DESC:METHOD AND SYSTEM FOR PREDICTIVE MAINTENANCE OF SMART VEHICLES
CROSS REFERENCE TO RELATED APPLICATIONS
The present application claims priority from Indian Provisional Patent Application No. 202421095006 filed on 03/12/2024, the entirety of which is incorporated herein by a reference.
TECHNICAL FIELD
Generally, the present disclosure relates to predictive maintenance. Particularly, the present disclosure relates to a system and method for predictive maintenance of smart vehicles.
BACKGROUND
The vehicle operates with high efficiency and reduced emissions compared to internal combustion engine vehicles, relying on components such as electric motors, battery management systems, regenerative braking systems, and power electronics for motion and energy control. The vehicles require maintenance, which involves monitoring the condition of mechanical and electrical components, evaluating wear and degradation patterns, and scheduling service operations to ensure operational efficiency, safety, and longevity. Predictive maintenance leverages sensor data, operational parameters, and analytical models to estimate component wear, enabling timely interventions that reduce unexpected failures and optimize overall vehicle performance.
Furthermore, existing technologies in vehicle maintenance include traditional scheduled maintenance based on fixed time or distance intervals, condition-based monitoring systems that utilize individual sensors such as vibration, speed, or torque sensors, and predictive maintenance approaches that employ statistical or machine learning models to forecast component degradation. The existing systems monitor operational parameters of vehicles, classify road surfaces using vibration signals, and estimate wear for maintenance scheduling. The existing technologies integrate accelerometers, gyroscopes, or speed sensors to assess driving conditions and suggest maintenance actions. The accelerometers capture vibration patterns that indicate road roughness, gyroscopes track angular velocity to assess steering dynamics and suspension stress, and speed sensors measure velocity to evaluate load and wear progression. Data from the sensors helps identify anomalies and suggest maintenance actions, such as inspecting suspension after high vibration exposure or checking steering after irregular angular velocity patterns.
However, there are certain problems associated with the existing or above-mentioned mechanism for predictive maintenance of a vehicle that arise from the inability of the existing technology to accurately quantify wear across diverse road surfaces and driving conditions, leading to maintenance scheduling that either underestimates or overestimates component degradation. The conventional approaches fail to assign surface-specific wear coefficients, aggregate distances over multiple road types, or dynamically compare effective wear against maintenance thresholds, resulting in inefficiencies, unexpected failures, or premature servicing.
Therefore, there exists a need for a secure, interoperable, and automated alternative for predictive maintenance of a vehicle.
SUMMARY
An object of the present disclosure is to provide a system for predictive maintenance of a vehicle.
Another object of the present disclosure is to provide a method for predictive maintenance of a vehicle.
Yet another object of the present disclosure is to provide a system and a method for improving maintenance accuracy.
In accordance with a first aspect of the present disclosure, there is provided a system for predictive maintenance of a vehicle, the system comprising:
at least one sensor module configured to sense at least one operational parameter of the vehicle;
at least one road classifier module communicably coupled to the at least one sensor module and configured to classify road surface characteristics based on the sensed data;
at least one wear computation module communicably coupled to the at least one road classifier module and configured to compute an effective wear distance; and
a maintenance module communicably coupled to the at least one wear computation module,
wherein the maintenance module is configured to schedule a predictive maintenance based on the classified road surface characteristics and the computed effective wear distance.
The system for predictive maintenance of a vehicle, as described in the present disclosure, is advantageous in terms of improving maintenance accuracy by aligning service schedules with actual road-induced wear patterns derived from real-time sensor data. Further, the road classifier module enables precise identification of the road surface characteristics through frequency-based attributes and machine learning models, ensuring adaptive classification under dynamic driving conditions. Furthermore, the wear computation module standardizes wear analysis by computing surface-specific coefficients using a baseline distance, allowing accurate aggregation of distance travelled on diverse surfaces. Moreover, the maintenance module enhances vehicle reliability by comparing the effective wear distance against predefined thresholds, generating timely alerts for maintenance interventions. Additionally, the integrated approach reduces operational costs, prevents unexpected failures, and extends vehicle lifespan by providing data-driven, condition-based maintenance recommendations across varied terrains.
In accordance with another aspect of the present disclosure, there is provided a method of predictive maintenance of a vehicle, the method comprises:
sensing at least one operational parameter of the vehicle, via at least one sensor module;
classifying road surface characteristics based on the sensed data, via a road classifier module;
computing an effective wear distance based on aggregated distance travelled on each classified surface, via a wear computation module;
performing a comparison between the effective wear distance and a maintenance threshold distance, via a maintenance module; and
generating an alert signal based on the comparison, via the maintenance module.
Additional aspects, advantages, features, and objects of the present disclosure would be made apparent from the drawings and the detailed description of the illustrative embodiments constructed in conjunction with the appended claims that follow.
It will be appreciated that features of the present disclosure are susceptible to being combined in various combinations without departing from the scope of the present disclosure as defined by the appended claims.
BRIEF DESCRIPTION OF DRAWINGS
The summary above, as well as the following detailed description of illustrative embodiments, is better understood when read in conjunction with the appended drawings. For the purpose of illustrating the present disclosure, exemplary constructions of the disclosure are shown in the drawings. However, the present disclosure is not limited to specific methods and instrumentalities disclosed herein. Moreover, those in the art will understand that the drawings are not to scale. Wherever possible, like elements have been indicated by identical numbers.
Embodiments of the present disclosure will now be described, by way of example only, with reference to the following diagrams wherein:
Figure 1 illustrates a block diagram of a system for predictive maintenance of a vehicle, in accordance with an embodiment of the present disclosure.
Figure 2 illustrates a flow chart of a method of predictive maintenance of a vehicle, in accordance with another embodiment of the present disclosure.
In the accompanying drawings, an underlined number is employed to represent an item over which the underlined number is positioned or an item to which the underlined number is adjacent. A non-underlined number relates to an item identified by a line linking the non-underlined number to the item. When a number is non-underlined and accompanied by an associated arrow, the non-underlined number is used to identify a general item at which the arrow is pointing.
DETAILED DESCRIPTION
The following detailed description illustrates embodiments of the present disclosure and ways in which they can be implemented. Although some modes of carrying out the present disclosure have been disclosed, those skilled in the art would recognize that other embodiments for carrying out or practicing the present disclosure are also possible.
As used herein, the term “predictive maintenance” refers to a proactive approach aimed at forecasting and addressing equipment degradation or failure before functional interruptions occur. Specifically, the predictive maintenance involves sensing key operational parameters such as, but not limited to, acceleration, braking force, steering angle, and vibrations through an array of sensors. The collected data undergoes processing through a road classifier module that extracts frequency-based attributes from vibration signals and assigns the frequency-based attributes to labeled datasets. Further, a machine learning model is trained on the datasets to classify road surface characteristics, identifying surfaces ranging from smooth highways to highly abrasive terrains. The classified surfaces are further used by a wear computation module to determine a baseline coefficient for the least abrasive surface and compute surface-specific coefficients by comparing baseline distances with actual distances travelled. Furthermore, aggregating distance travelled across classified surfaces yields an effective wear distance representing the cumulative wear experienced by vehicle components. The maintenance module compares the effective wear distance with predefined threshold values and generates alert signals prompting maintenance actions before critical wear occurs. Moreover, the predictive maintenance includes, but is not limited to, sensor-based monitoring, which relies on real-time data from multiple sensors to track operational health; model-based approaches, where algorithms and machine learning techniques analyze patterns and predict failures based on historical and current data; and condition-based maintenance, which schedules maintenance tasks in alignment with the condition and wear of components rather than fixed time intervals. The system integrates the aforementioned types into a unified predictive maintenance framework, enhancing reliability, reducing downtime, and optimizing maintenance schedules by accounting for road surface variations and driving conditions.
As used herein, the term “sensor module” refers to a collection of electronic sensing devices designed to monitor, measure, and transmit physical or environmental parameters that affect the operation and performance of the system. Specifically, the sensor module plays a critical role by continuously capturing data related to the vehicle’s dynamic behavior and interaction with road conditions. The sensor module integrates multiple types of sensors including, but not limited to, inertial measurement unit sensors, which measure acceleration, angular velocity, and orientation; throttle position sensors, which monitor the extent of throttle opening and engine load; brake application sensors, which detect braking force and application patterns; torque demand sensors, which assess the torque output required by the drivetrain; steering angle sensors, which track steering inputs and wheel alignment; vehicle speed sensors, which measure real-time speed variations; and vibration sensors, which detect high-frequency oscillations caused by road irregularities and surface roughness. Further, the data from each sensor undergoes preprocessing to filter noise and enhance signal quality before being transmitted to the road classifier module for further analysis. The types of sensor modules include, but are not limited to, motion sensing modules, which gather data on acceleration and orientation; force sensing modules, which detect load and pressure variations; positional sensing modules, which measure angles and displacement; and acoustic or vibration sensing modules, which capture surface-induced oscillations and resonance patterns. The sensor module serves as the foundational component in the predictive maintenance system, providing the essential data required for classification, computation, and maintenance scheduling, thereby enhancing vehicle reliability and operational efficiency under varying road conditions.
As used herein, the term “operational parameter” refers to any measurable physical quantity or condition that reflects the performance, behavior, or state of a vehicle during operation. Specifically, the operational parameters include, but are not limited to, dynamic and static attributes that directly influence wear and degradation of components over time. Further, the types of operational parameters include, but are not limited to, motion-related parameters such as, but not limited to, speed, acceleration, and orientation; force-related parameters such as, but not limited to, braking pressure and torque output; position-related parameters such as, but not limited to, steering angle and throttle position; and surface interaction parameters such as, but not limited to, vibration frequency and amplitude. Aggregating data from the parameters enables comprehensive analysis of the driving behavior, road conditions, and component stress, which directly feed into the road classification algorithms, the wear computation models, and the maintenance scheduling. Ultimately, the operational parameters thus form the essential dataset that links the vehicle dynamics to predictive maintenance strategies, ensuring precise, data-driven insights for optimal maintenance planning and vehicle longevity.
As used herein, the term “road classifier module” refers to an analytical component within the vehicle maintenance system that processes the sensor data to identify and categorize the characteristics of the road surfaces based on measurable signals. Specifically, the primary function of the road classifier module involves interpreting the operational parameters, particularly the vibration data, to assess road roughness, texture, and abrasiveness, thereby enabling accurate determination of surface types that influence the vehicle wear and tear. The road classifier module receives high-frequency vibration signals from the sensor module and extracts the frequency-based attributes. Furthermore, once the model is trained, the road-classifier module classifies incoming data by mapping observed frequency patterns to predefined surface categories, continuously updating classification results based on current driving inputs and environmental changes. The types of road classifier modules include, but are not limited to, frequency-based classifiers that rely on vibration spectra, statistical classifiers that analyze trends in speed, acceleration, and braking patterns, and hybrid classifiers that combine multiple sensor inputs for enhanced accuracy. Moreover, the road classifier module integrates seamlessly with the wear computation module by providing surface-specific labels and coefficients, which feed into aggregated wear calculations and maintenance schedules. By accurately identifying road surface conditions, the road classifier module ensures precise assessment of component stress levels and effective wear distance computations, thereby supporting predictive maintenance strategies focused on extending vehicle life and improving safety under diverse driving conditions.
As used herein, the term “road surface characteristics” refers to the measurable and observable physical properties of a roadway that directly influence the vehicle dynamics, wear, and operational efficiency. Specifically, the road surface characteristics form the primary factors used to calculate effective wear distance and schedule maintenance activities. Detailed analysis of the road surface characteristics involves examining attributes such as, but not limited to, texture, roughness, hardness, and abrasiveness. The types of road surface characteristics include, but are not limited to, smooth surfaces such as, but not limited to, well-maintained asphalt highways; moderately rough surfaces such as, but not limited to, gravel paths or urban roads with minor cracks; highly abrasive surfaces like degraded concrete or heavily trafficked industrial areas; and irregular surfaces consisting of potholes, ruts, or sudden elevation changes. Furthermore, each classified surface associates with the baseline coefficient or wear factor that quantifies the road surface characteristics' contribution to component wear. The wear computation module uses the classified characteristics to determine a ratio between the baseline distances and the actual distances travelled, enabling accurate computation of wear for each surface type. Ultimately, the road surface characteristics thus form the critical link between driving conditions and component degradation, providing the data foundation for predictive maintenance algorithms aimed at ensuring optimal vehicle performance and durability across varied terrains.
As used herein, the term “wear computation module” refers to a computational system within the vehicle maintenance architecture that analyzes the classified road surface data and distance measurements to calculate the extent of the wear experienced by vehicle components over time. The wear computation module functions as an integrative processor that transforms raw driving data into actionable metrics, enabling informed maintenance scheduling based on real-world operating conditions. Specifically, the wear computation module selects the least abrasive surface from among classified road types and assigns the baseline coefficient to represent the minimal wear conditions. Further, the wear computation module then determines a ratio between the baseline distance and the actual distance travelled on each classified surface, computing surface-specific coefficients. The types of wear computation modules include, but are not limited to, baseline reference modules that establish minimal wear thresholds; ratio-based modules that calculate wear intensity by comparing actual and reference distances; coefficient aggregation modules that integrate wear effects across multiple surfaces; and threshold comparison modules that evaluate the computed wear distance against predefined maintenance thresholds to generate alert signals. Moreover, the effective wear distance computed by the module serves as a key parameter for the maintenance module to schedule timely interventions, ensuring that vehicle components receive necessary servicing before reaching critical levels of degradation. The wear computation module thus enables precise, data-driven wear estimation by accounting for road surface characteristics, travel patterns, and driving conditions, forming the analytical backbone of predictive maintenance systems aimed at enhancing vehicle longevity and safety.
As used herein, the term “effective wear distance” refers to a calculated metric that represents the cumulative impact of road surfaces on the wear and tear of the vehicle components over the distance travelled. Specifically, the effective wear distance arises from the aggregation of distance travelled across multiple classified road surfaces, each weighted by the specific wear coefficient. Further, for each classified surface encountered during driving, the wear computation module calculates a ratio between the baseline distance and the distance travelled on that surface, deriving surface-specific coefficients that quantify how much additional wear occurs relative to the baseline. Furthermore, the effective wear distance integrates real-world operational data by capturing how surface-induced forces, such as, but not limited to vibrations, friction, and shocks, contribute to component degradation across different terrains. The types of effective wear distance include, but are not limited to, baseline wear distance, which reflects wear under optimal surface conditions; weighted wear distance, which incorporates surface abrasiveness coefficients; aggregated wear distance, which sums wear contributions from multiple surfaces; and threshold comparison distance, which serves as a parameter for triggering maintenance alerts once the accumulated wear reaches predefined limits. Moreover, the effective wear distance serves as a precise and dynamic measure for maintenance planning, allowing maintenance scheduling based on actual usage patterns rather than fixed intervals, thereby enhancing vehicle reliability, safety, and lifespan across varied driving environments.
As used herein, the term “maintenance module” refers to a component for monitoring vehicle health and scheduling servicing actions based on analyzed data from the operational parameters, road surface classifications, and computed wear metrics. Specifically, the maintenance module functions as the decision-making element within the predictive maintenance framework, ensuring that necessary interventions occur before excessive wear compromises the vehicle's performance or safety. The maintenance module evaluates that the aggregated wear accumulated across various classified surfaces exceeds the threshold using a predefined comparison, and upon identifying excess wear, generates an alert signal to prompt maintenance activities. Further, the maintenance module integrates communication protocols to relay maintenance notifications to vehicle operators or fleet management systems, ensuring timely response and intervention. The types of maintenance modules include, but are not limited to, threshold comparison modules that assess computed wear distances against preset limits; alert generation modules that create warning signals or notifications; scheduling modules that plan maintenance tasks based on wear forecasts; and reporting modules that log wear data, driving patterns, and maintenance actions for tracking and optimization purposes. Furthermore, by leveraging effective wear distance and road surface data, the maintenance module ensures maintenance activities align with actual usage conditions rather than generic schedules, enabling precise, data-driven decision-making that enhances vehicle reliability, safety, and operational efficiency across diverse driving scenarios.
As used herein, the term “inertial measurement unit sensor” refers to an electronic device that measures and reports a vehicle’s specific force, angular rate, and sometimes magnetic field in multiple axes, providing critical data related to motion, orientation, and dynamic behavior. Specifically, the inertial measurement unit sensor measures acceleration along longitudinal, lateral, and vertical axes, angular velocity around pitch, roll, and yaw axes, providing a comprehensive profile of the vehicle movements and responds to road irregularities. Further, the data from the inertial measurement unit sensor undergoes preprocessing to filter noise and isolate frequency-based attributes relevant to the road classification algorithms. Furthermore, the types of inertial measurement unit sensors include, but are not limited to, three-axis accelerometers, which detect linear motion in multiple directions; gyroscopes, which measure rotational movement; and integrated units combining both sensors to provide fused data for motion analysis. Some variants incorporate magnetometers to account for orientation relative to Earth’s magnetic field, enhancing directional awareness. Moreover, by delivering accurate, high-resolution motion data, the inertial measurement unit sensor ensures maintenance decisions align with real-world forces acting on the vehicle structure and components.
As used herein, the term “throttle position sensor” refers to an electronic device that measures the angular position or movement of the throttle valve within a vehicle’s intake system, providing real-time information about engine load and power demand. Specifically, the throttle position sensor measures the degree of throttle opening, expressed as a percentage relative to the fully closed and fully open positions, and transmits the data continuously to the sensor module. Further, data from the throttle position sensor integrates with other operational parameters, such as, but not limited to, vehicle speed, torque demand, and brake application, to provide a holistic view of driving conditions that influence wear on the drivetrain, tires, and suspension systems. The throttle position sensor output assists the road classifier module by correlating acceleration patterns and torque variations with specific road surface characteristics, enabling classification based on driving intensity and force distribution. Furthermore, the types of throttle position sensors include, but are not limited to, potentiometer-based sensors, which use a variable resistor to measure angular displacement; Hall-effect sensors, which rely on magnetic fields to detect movement without physical contact; and non-contact optical sensors, which use light-based measurements to enhance accuracy and reduce wear over time. Moreover, by accurately measuring throttle engagement, the sensor supports classification of road surfaces, computation of effective wear distance, and generation of maintenance schedules tailored to actual driving patterns, thereby enhancing vehicle reliability, performance, and safety across varied terrains and operating environments.
As used herein, the term “brake application sensor” refers to an electronic device that measures the force, pressure, or extent of braking applied by the vehicle operator, providing critical data regarding deceleration patterns, driving behavior, and stress imposed on braking components. Specifically, the brake application sensor detects parameters such as, but not limited to, hydraulic pressure within the brake lines, mechanical displacement of brake pads, or electronic signals from brake actuators, depending on the braking system architecture. Furthermore, the road classifier module uses data from the brake application sensor with braking patterns to differentiate between smooth deceleration on well-maintained roads and abrupt braking caused by rough or hazardous terrains, further refining road surface classification models. The types of brake application sensors include, but are not limited to, pressure-based sensors, which measure hydraulic force within braking circuits; displacement sensors, which track mechanical movement of braking elements; and electronic sensors, which record actuator signals or pedal position data. Moreover, some sensors incorporate temperature monitoring to assess heat buildup, contributing additional insights into brake system health. Additionally, by enabling precise correlation between braking stress and road conditions, the brake application sensor supports wear computation algorithms, classification of surface abrasiveness, and maintenance scheduling that aligns with actual driving forces, ensuring optimized vehicle performance, safety, and longevity across diverse driving scenarios.
As used herein, the term “torque demand sensor” refers to a device that measures the mechanical power or rotational force required by the vehicle’s drivetrain to achieve desired acceleration, speed, or load conditions, providing essential information regarding engine output, drivetrain stress, and driving effort. Specifically, the torque demand sensor measures parameters such as, but not limited to, torque output from the electric motor or internal combustion engine, drive shaft rotational force, or wheel torque using strain gauges, current sensors, or rotational speed differentials. Further, the parameters assist in identifying patterns of heavy acceleration, frequent load changes, or sustained high torque conditions, which directly impact wear on drivetrain components, bearings, and joints. The road classifier module uses torque demand patterns in conjunction with vibration and speed data to distinguish between driving scenarios, such as, but not limited to, smooth highway cruising, uphill driving, or traversing rough, abrasive surfaces, refining the classification of the road surface characteristics. Furthermore, the types of torque demand sensors include, but are not limited to, strain gauge sensors that measure deformation due to applied torque; magnetic sensors that infer torque from electromagnetic changes; current sensors that relate electric motor power draw to torque output; and speed differential sensors that compare rotational speeds between drivetrain segments. The torque sensors integrate temperature compensation and signal filtering algorithms to enhance measurement accuracy under variable environmental and load conditions. Additionally, the torque demand sensor’s contribution ensures maintenance decisions reflect actual mechanical loads, optimizing vehicle reliability, safety, and performance across diverse driving environments and load conditions.
As used herein, the term “steering angle sensor” refers to an electronic device that measures the angular position and rate of change of the steering wheel or steering column, providing precise information about vehicle handling, directional control, and driving behavior. Specifically, the steering angle sensor continuously monitors the degree of wheel rotation relative to the vehicle’s forward axis, capturing data on steering adjustments, cornering intensity, lane changes, and abrupt directional shifts. Further, by integrating steering data with operational parameters such as, but not limited to, vehicle speed, torque demand, and brake application, the maintenance system assesses how directional inputs correlate with stress on mechanical components and surface interactions. Furthermore, the types of steering angle sensors include, but are not limited to, resistive sensors that measure changes in electrical resistance corresponding to angular displacement; optical sensors that detect movement through light-based measurements; magnetic sensors that use Hall-effect or magneto-resistive principles to sense rotation; and inductive sensors that track changes in electromagnetic fields caused by steering motion. Moreover, by correlating steering patterns with surface abrasiveness, vibration signals, and load conditions, the sensor supports the computation of wear on steering linkages, tires, and suspension systems, contributing to the aggregation of effective wear distance and informing maintenance scheduling. Additionally, the steering angle sensor’s role ensures that maintenance activities align with actual steering demands, promoting vehicle safety, handling precision, and longevity across diverse driving conditions and terrains.
As used herein, the term “vehicle speed sensor” refers to an electronic device that measures the rate at which a vehicle travels over a surface, providing accurate, real-time information regarding velocity, acceleration, and deceleration patterns. Specifically, the vehicle speed sensor continuously captures instantaneous speed data and transmits the information to the sensor module for further processing. The vehicle speed sensor monitors changes in speed during various driving conditions, including, but not limited to, steady cruising, rapid acceleration, sudden braking, and maneuvering over uneven or abrasive surfaces. Further, the data integrates with information from the sensor module to create a comprehensive profile of operational parameters influencing component wear. Furthermore, the types of vehicle speed sensors include, but are not limited to, magneto-resistive sensors that detect rotational speed of the wheel hub or axle through changes in magnetic fields; Hall-effect sensors that generate signals from rotational components passing magnetic poles; optical sensors that measure wheel rotation using light-based detection methods; and GPS-based sensors that calculate speed by tracking geographic displacement over time. Additionally, the vehicle speed sensor’s contribution ensures maintenance schedules reflect real-world usage, optimizing vehicle safety, performance, and component longevity across diverse terrains and operational environments.
As used herein, the term “vibration sensor” refers to an electronic device that measures oscillatory motion, shock, or mechanical disturbances in the vehicle’s structure, providing critical data regarding surface-induced forces, component stress, and dynamic interactions between the vehicle and road conditions. Specifically, the vibration sensor continuously records oscillations across multiple axes, including, but not limited to, vertical, lateral, and longitudinal directions, using accelerometers or piezoelectric elements that convert mechanical motion into electrical signals. Further, the road classifier module extracts frequency-based attributes from the vibration data, identifying patterns. Furthermore, the types of vibration sensors include, but are not limited to, accelerometer-based sensors that measure linear motion across multiple axes; piezoelectric sensors that generate electrical signals proportional to mechanical stress; velocity sensors that derive oscillatory movement from relative displacement; and strain sensors that monitor deformation within structural elements. Moreover, by enabling classification of road surfaces based on the vibration signatures and supporting computation of surface-specific wear coefficients, the vibration sensor directly contributes to calculating effective wear distance and scheduling maintenance actions aligned with real driving conditions. Additionally, the vibration sensor’s high-resolution data acquisition ensures that wear assessment reflects actual stress factors encountered during operation, enhancing vehicle safety, reducing downtime, and extending component lifespan across varied terrains and usage patterns.
As used herein, the term “frequency-based attributes” refers to measurable characteristics derived from analyzing the spectral content of vibration signals, providing insight into the types of forces, surface irregularities, and dynamic behaviors that a vehicle experiences during motion. Specifically, frequency-based attributes are extracted from vibration data recorded by the vibration sensor, and signal processing techniques are applied to the time-domain vibration signals, isolating dominant frequencies, harmonic components, energy distribution, and noise levels. Further, the extracted attributes include, but are not limited to, peak frequency, which identifies the most prominent oscillation rate; frequency bandwidth, which reflects the range of significant vibrations present; signal amplitude at specific frequency bands, which correlates with the severity of surface irregularities; root mean square (RMS) values, representing the overall energy content of the vibration; and spectral kurtosis, indicating transient events and impulsive forces caused by sharp bumps or abrupt changes in road texture. Furthermore, the types of frequency-based attributes include, but are not limited to, narrowband attributes focused on specific frequency ranges associated with particular disturbances; broadband attributes capturing energy spread across multiple frequencies; transient attributes that highlight sudden changes or impacts; and statistical attributes that quantify overall vibration behavior over defined intervals. Moreover, through accurate extraction and analysis of frequency components, the attributes allow the predictive maintenance system to compute surface-specific wear coefficients and aggregate effective wear distance, ensuring maintenance decisions reflect actual driving forces and road-induced stresses. The comprehensive analysis of frequency-based attributes provides a reliable foundation for understanding how vibrations influence vehicle dynamics, thereby enhancing maintenance scheduling, safety, and component longevity across diverse operational environments.
As used herein, the term “labelled dataset” refers to a structured collection of data samples in which each sample is paired with a corresponding identifier or classification that describes the corresponding characteristics, enabling supervised learning algorithms to learn relationships between input signals and expected outputs. Specifically, the labelled datasets consist of vibration signals, frequency attributes, and associated road surface types, each tagged with identifiers that reflect specific surface profiles. Further, the dataset includes, but is not limited to, various driving conditions, speeds, and load scenarios, ensuring comprehensive representation of real-world usage patterns. Furthermore, the types of labelled datasets include, but are not limited to, static datasets, which contain pre-recorded signals collected under controlled conditions and manually annotated for accuracy; dynamic datasets, which continuously update with new driving data and real-time labelling based on automated classification algorithms; balanced datasets, designed to ensure equal representation of all surface types to prevent bias; and augmented datasets, which synthetically expand sample diversity through noise addition, scaling, or signal transformations to enhance model robustness. Moreover, through structured labelling and comprehensive representation, the datasets ensure that the classification models achieve high accuracy and generalization, thereby improving the calculation of the effective wear distance and facilitating precise maintenance scheduling tailored to real-world driving conditions and road environments. The use of well-curated labelled datasets ensures that maintenance interventions align with actual wear factors, enhancing vehicle reliability, safety, and operational efficiency.
As used herein, the term “least abrasive surface” refers to the road surface classification that exhibits minimal friction, texture irregularities, and structural imperfections, resulting in the lowest rate of wear and tear on vehicle components during operation. Specifically, the least abrasive surface represents the category of roads that induce the smallest mechanical stress on the vehicle’s tires, suspension, and drivetrain, allowing the wear computation module to assign the baseline coefficient and the reference distance that quantify minimal wear impact. The types of least abrasive surfaces include, but are not limited to, well-maintained asphalt highways, characterized by uniform texture, minimal cracking, and low debris accumulation; concrete surfaces with smooth joints and properly sealed seams; urban roads recently resurfaced with high-quality materials that reduce shock loads; and controlled test tracks designed to provide optimal traction and minimal vibration interference. Furthermore, each surface type undergoes analysis to confirm minimal impact on vehicle motion, allowing the wear computation module to assign the baseline coefficient that serves as a standard for comparison across other, more abrasive surfaces. Moreover, by using the least abrasive surface as the reference point, the predictive maintenance system achieves reliable wear estimation, reduces false-positive maintenance alerts, and enhances vehicle durability through informed decision-making based on actual road-induced stresses. The classification of least abrasive surfaces ensures that maintenance interventions reflect the true extent of wear caused by harsher driving environments, thereby optimizing vehicle performance and longevity across diverse terrains.
As used herein, the term “baseline distance” refers to a reference measurement that represents the standard distance travelled by the vehicle on the least abrasive surface, serving as the comparative metric for assessing wear caused by more abrasive or irregular road conditions. Specifically, the baseline distance corresponds to the distance the vehicle travels on the surface classified as least abrasive, characterized by smooth texture, low friction, and minimal vibration. Further, the frequency-based attributes and vibration data from sensors confirm that the road condition aligns with minimal surface irregularities, validating the selection of the baseline distance for the wear calculations. Furthermore the types of baseline distance include, but are not limited to, fixed baseline distances, which are predefined for specific road types such as, but not limited to, asphalt highways or smooth concrete surfaces; dynamic baseline distances, which adjust based on environmental factors such as, but not limited to, temperature, load, or driving speed to ensure continued relevance across varying operational scenarios; segment-based baseline distances, where the distance is calculated over specific sections of a journey to account for transient changes in road condition; and adaptive baseline distances, which are continuously updated through machine learning models that refine reference measurements based on historical driving data and sensor feedback. Accurate determination of the baseline distance ensures predictive maintenance systems deliver precise, data-driven assessments that enhance vehicle reliability, safety, and longevity across diverse driving environments and conditions.
As used herein, the term “baseline coefficient” refers to a numerical factor assigned to the road surface classified as least abrasive, representing the minimal rate of wear experienced by vehicle components over a defined distance. Specifically, the baseline coefficient is assigned to the least abrasive surface identified through frequency-based attributes extracted from vibration data and validated against sensor measurements. Further, the wear computation module uses the baseline coefficient in conjunction with the baseline distance to determine ratios comparing actual travelled distances on classified surfaces to the least abrasive standard, thereby calculating effective wear distance for each surface type. The types of baseline coefficients include, but are not limited to, static baseline coefficients, which remain constant for specific surface categories such as, but not limited to, well-maintained asphalt or smooth concrete; dynamic baseline coefficients, which adjust based on vehicle load, speed, or environmental conditions to reflect changing stress factors; segment-specific baseline coefficients, calculated for discrete portions of a journey to account for local variations in road smoothness; and adaptive baseline coefficients, updated over time through machine learning algorithms that refine wear estimates based on historical sensor data and surface classifications. Furthermore, the baseline coefficient ensures that effective wear distance calculations precisely reflect real-world operational stresses, thereby optimizing maintenance interventions, improving component lifespan, and enhancing vehicle reliability under diverse driving conditions.
As used herein, the term “surface-specific coefficient” refers to a numerical factor assigned to the classified road surface that quantifies the relative wear imposed on vehicle components compared to the least abrasive reference surface. Specifically, the wear computation module determines surface-specific coefficients by first establishing the baseline coefficient and baseline distance for the least abrasive surface. Further, the surface-specific coefficients incorporate data from the frequency-based attributes extracted from the vibration sensors, as well as the operational parameters, ensuring that the computed coefficient accurately represents the mechanical stress experienced by tires, suspension, and drivetrain components under real-world driving conditions. The types of surface-specific coefficients include, but are not limited to, static coefficients assigned to fixed road types, such as, but not limited to, gravel, smooth asphalt, or degraded concrete; dynamic coefficients adjusted based on varying vehicle load, speed, or environmental conditions; segment-based coefficients calculated for discrete portions of a journey to account for localized variations in road condition; and adaptive coefficients continuously refined through machine learning models that integrate historical driving data and sensor feedback to improve predictive accuracy. Furthermore, by quantifying the relative wear imposed by diverse road conditions, surface-specific coefficients ensure that predictive maintenance strategies align with actual operational stresses, thereby optimizing maintenance scheduling, extending vehicle component lifespan, and enhancing overall reliability and safety across varied driving environments.
As used herein, the term “maintenance threshold distance” refers to a predefined value of the effective wear distance at which the vehicle component requires maintenance intervention to prevent excessive degradation or failure. Specifically, the maintenance module receives the effective wear distance computed by the wear computation module, which aggregates distances travelled across classified road surfaces weighted by surface-specific coefficients. Further, the maintenance threshold distance integrates multiple factors, including, but not limited to, component durability specifications, expected lifespan under normal usage, road surface variability, and the operational stresses such as, but not limited to, acceleration, braking, torque demand, and vibration intensity. The types of maintenance threshold distances include, but are not limited to, static threshold distances, which are fixed based on manufacturer specifications or standard wear limits for particular components; dynamic threshold distances, which adjust according to real-time operational conditions, driving patterns, and environmental factors; component-specific thresholds, which vary for different parts such as, but not limited to, brake pads, tires, suspension elements, or drivetrain components; and adaptive thresholds, which are refined using historical wear data and machine learning algorithms to optimize predictive accuracy and minimize unscheduled downtime. Furthermore, by utilizing the maintenance threshold distance, the system prevents premature component failure, reduces unnecessary servicing, and enhances vehicle reliability, safety, and operational efficiency across diverse driving environments.
As used herein, the term “alert signal” refers to an electronic notification generated by the predictive maintenance system to indicate that the vehicle component has reached or exceeded the predefined wear limit, prompting immediate or scheduled maintenance action. Specifically, the alert signal originates from the maintenance module after comparing the effective wear distance, computed by the wear computation module, against the maintenance threshold distance assigned for specific vehicle components. Further, the alert signal integrates data from surface-specific coefficients, operational parameters, ensuring that notifications reflect actual wear conditions influenced by road surface characteristics and driving behavior. The types of alert signals include, but is not limited to, visual alerts displayed on the vehicle dashboard or operator interface, providing immediate awareness of maintenance requirements; auditory alerts such as beeps or alarms that draw attention to critical wear conditions; electronic communication signals transmitted to fleet management systems or maintenance scheduling platforms for centralized monitoring; and haptic alerts, including, but is not limited to, vibrations or pedal/seat feedback, that directly notify the driver of impending maintenance needs. Furthermore, the alert signals also vary by priority level, distinguishing between routine maintenance advisories and urgent warnings that require immediate action. Moreover, by delivering precise, context-sensitive notifications based on actual wear data and road-induced stresses, the alert signal enables proactive maintenance planning, reduces unexpected failures, and aligns service actions with operational realities.
In accordance with a first aspect of the present disclosure, there is provided a system for predictive maintenance of a vehicle, the system comprising:
at least one sensor module configured to sense at least one operational parameter of the vehicle;
at least one road classifier module communicably coupled to the at least one sensor module and configured to classify road surface characteristics based on the sensed data;
at least one wear computation module communicably coupled to the at least one road classifier module and configured to compute an effective wear distance; and
a maintenance module communicably coupled to the at least one wear computation module,
wherein the maintenance module is configured to schedule a predictive maintenance based on the classified road surface characteristics and the computed effective wear distance.
Referring to figure 1, in accordance with an embodiment, there is described a system 100 for predictive maintenance of a vehicle. The system 100 comprises at least one sensor module 102, configured to sense at least one operational parameter of the vehicle. Further, the system 100 comprises at least one road classifier module 104 is communicably coupled to the at least one sensor module 102 and is configured to classify road surface characteristics based on the sensed data. Furthermore, the system 100 comprises at least one wear computation module 106 communicably coupled to the at least one road classifier module 104 and is configured to compute an effective wear distance. Moreover, the system 100 comprises a maintenance module 108 communicably coupled to the at least one wear computation module 106, wherein the maintenance module 108 is configured to schedule a predictive maintenance based on the classified road surface characteristics and the computed effective wear distance.
The system 100 for predictive maintenance of the vehicle operates through seamless interaction between the at least one sensor module 102, the road classifier module 104, the wear computation module 106, and the maintenance module 108. The at least one sensor module 102 senses at least one operational parameter of the vehicle, such as, but not limited to, acceleration, vibration, throttle position, brake application, torque demand, steering angle, and vehicle speed. Further, the sensed data is transmitted to the road classifier module 104, which extracts frequency-based attributes from the vibration data and classifies the road surface characteristics by applying machine learning models trained on labelled datasets. The classification output identifies specific road types travelled by the vehicle, associating each segment of travel with corresponding surface characteristics. Furthermore, the wear computation module 106 computes the effective wear distance based on the classified road surfaces and the distance travelled on each surface. The module 106 selects a least abrasive surface and assigns a baseline coefficient to the baseline distance of the surface. Moreover, ratios between the baseline distance and the distance travelled on each classified surface are calculated to derive surface-specific coefficients. The wear computation module 106 aggregates the distance travelled on each surface multiplied by the corresponding coefficient to compute the total effective wear distance. Additionally, the maintenance module 108 receives the computed effective wear distance and compares the effective wear distance against a maintenance threshold distance. Upon determining that the effective wear distance exceeds the maintenance threshold distance, the maintenance module 108 generates an alert signal, triggering a scheduled maintenance action to preserve vehicle integrity. Consequently, the system 100 delivers accurate, usage-based maintenance scheduling that reflects real-world operating conditions. By integrating real-time sensor data with the road classification algorithms and wear computation logic, the system 100 ensures that maintenance intervals correspond to actual wear patterns rather than fixed schedules. Additionally, the system 100 enhances vehicle reliability, reduces downtime, and optimizes maintenance resources by avoiding unnecessary servicing and preventing failure due to unforeseen road-induced wear. The predictive maintenance approach improves fleet management efficiency and extends vehicle lifespan by accounting for diverse operational scenarios encountered during driving.
In an embodiment, the at least one sensor module 102 comprises one or more sensors selected from the group of sensors comprising inertial measurement unit sensors, throttle position sensors, brake application sensors, torque demand sensors, steering angle sensors, vehicle speed sensors, and vibration sensors. Specifically, the inertial measurement unit sensors measure linear acceleration and angular velocity across multiple axes with high precision, such as, but not limited to, ±16 g for acceleration and ±250 °/s for angular velocity. The throttle position sensors record the percentage of throttle opening from 0% to 100% with a resolution of 0.1%. Further, the brake application sensors detect hydraulic pressure variations up to 200 bar with a sampling rate of 500 Hz. The torque demand sensors monitor torque requests from the drivetrain in real-time, offering data with a resolution of ±1 Nm. Furthermore, the steering angle sensors detect angular displacement of the steering system within a range of ±450° with accuracy up to 0.5°. The vehicle speed sensors capture wheel rotation data with precision up to ±0.1 km/h, and vibration sensors record oscillations from the chassis or suspension system at frequencies between 1 Hz and 2 kHz with sampling rates exceeding 1,000 Hz. The throttle position sensors, brake application sensors, and torque demand sensors contribute to assessing driving behavior under specific road conditions. Subsequently, the steering angle sensors assist in identifying turns, sharp curves, and vehicle handling characteristics, and the vehicle speed sensors offer information on dynamic load conditions and distance travelled on various surfaces. The road classifier module 104 utilizes the parameters to associate sensor data patterns with known road surfaces through machine learning algorithms. Furthermore, the predictive maintenance process improves reliability and safety by aligning maintenance schedules with actual usage patterns rather than predefined intervals. The predictive maintenance approach minimizes unexpected failures, extends component life, and optimizes operational costs by targeting maintenance interventions based on scientifically derived wear estimations.
In an embodiment, the road classifier module 104 is configured to extract at least one frequency-based attribute from vibration data received from the sensor module 102. Specifically, the vibration sensors within the sensor module 102 record the acceleration signals along multiple axes at a sampling rate of 1,000 Hz, capturing oscillations ranging from 1 Hz to 2 kHz. The road classifier module 104 applies signal processing techniques such as, but not limited to, Fast Fourier Transform (FFT) over time windows of 10 seconds, converting the time-domain vibration signals into frequency-domain representations. Further, the road classifier module 104 identifies the dominant frequency peaks, amplitude envelopes, root mean square (RMS) values, and spectral energy distributions across frequency bands such as, but not limited to, 5–50 Hz, 50–200 Hz, and 200–1,000 Hz. The frequency-based attributes reflect the interaction between the vehicle’s suspension system and the underlying road surface, enabling differentiation between surfaces such as asphalt, gravel, and dirt roads. Furthermore, the extracted frequency-based attributes from the vibration data form the foundation for the road surface classification. The road classifier module 104 normalizes the frequency data, removes noise through filtering techniques such as, but not limited to, band-pass and low-pass filters, and applies feature selection algorithms to retain the most discriminative attributes. Additionally, the road classifier module 104 continuously updates the classification process as new vibration data streams in from the sensor module 102, ensuring adaptive recognition of road types across varying driving conditions. The classified output, including, but not limited to, surface type labels and associated confidence scores, is transmitted to the wear computation module 106 for further analysis and aggregation into effective wear distance metrics. Consequently, extracting the frequency-based attributes from vibration data ensures accurate identification of the road surface characteristics based on objective, quantifiable vehicle dynamics. The approach eliminates reliance on external mapping databases or manual input, providing an autonomous, real-time solution that reflects actual road interactions. Ultimately, the accurate classification of road surfaces enhances maintenance scheduling through precise wear computation, improves vehicle safety by monitoring road-induced stress factors, and supports fleet management by delivering actionable insights into driving patterns and maintenance needs. The integration of the frequency-based attributes into predictive maintenance systems strengthens reliability, reduces downtime, and extends vehicle operational efficiency.
In an embodiment, the road classifier module 104 is configured to assign the extracted at least one frequency-based attribute to a plurality of labelled datasets and train a machine learning model based on the assigned labelled datasets. Specifically, the frequency-based attributes extracted from the vibration data by the road classifier module 104 undergo preprocessing steps, including, but not limited to, normalization, outlier removal, and noise filtering using algorithms such as, but not limited to, band-pass filters and moving average smoothing. The road classifier module 104 associates the processed attributes with the pre-compiled labeled datasets containing known road surface types, such as, but not limited to, asphalt, concrete, gravel, and dirt. Further, each labeled dataset includes, but is not limited to, multiple samples, each defined by frequency peaks, amplitude variations, RMS values, and energy distributions. The road classifier module 104 aligns the real-time vibration attributes with the labeled datasets through distance metrics, such as, but not limited to, Euclidean distance and correlation coefficients, to determine similarity. Furthermore, the aligned datasets serve as inputs to supervised learning algorithms, including, but not limited to, Random Forest, Support Vector Machine (SVM), or Gradient Boosting classifiers for model training. The training process involves partitioning the labeled datasets into training and validation sets with ratios such as, but not limited to, 80:20 or 70:30, depending on dataset size and variability. Moreover, the road classifier module 104 applies cross-validation techniques such as, but not limited to, k-fold validation to ensure model generalization and avoid overfitting. The trained model receives incoming vibration data in real-time, transforming incoming attributes into structured feature sets and predicting the most likely road surface classification with confidence scores. Subsequently, the road classifier module 104 stores the model parameters and continuously updates the classifier through retraining based on newly accumulated labeled datasets and feedback from wear computation results. The robust model training approach ensures high classification accuracy, reduces false positives and negatives, and provides confidence metrics that guide subsequent computations. Furthermore, the adaptive retraining process enhances resilience against environmental changes and sensor drift. The integration of machine learning into the road classifier module 104 advances predictive maintenance by offering precise road surface identification, which directly influences the computation of the effective wear distance and maintenance scheduling. Moreover, the approach improves vehicle reliability, lowers maintenance costs, and extends the operational lifespan of mechanical components by enabling data-driven decision-making based on actual road conditions.
In an embodiment, the road classifier module 104 is configured to classify the road surface characteristics based on the trained machine learning model. Specifically, after training the model using the labeled datasets associated with extracted frequency-based attributes, the road classifier module 104 receives real-time vibration data from the sensor module 102. The road classifier module 104 preprocesses the incoming data by applying filtering algorithms such as, but not limited to, band-pass and low-pass filters to eliminate noise and enhance signal quality. Further, the frequency-based attributes, including but not limited to dominant peaks, amplitude variations, RMS values, and spectral energy distributions, are calculated over defined time windows, such as, but not limited to, 10-second intervals. The frequency-based attributes are structured into feature vectors and input into the trained machine learning model. Furthermore, the model computes probability distributions for each known road surface type and assigns the classification label corresponding to the road surface that exhibits the highest probability score. Subsequently, the road classifier module 104 maintains logs of classification sequences, facilitating long-term analysis and adaptive retraining of the model to improve performance over extended operational periods. Consequently, classifying the road surface characteristics based on the trained machine learning model ensures accurate, real-time identification of road types influencing vehicle wear patterns. Furthermore, the accurate identification of road surfaces enhances predictive maintenance by providing reliable inputs to the wear computation module 106, allowing precise calculation of the effective wear distance. The process increases maintenance scheduling efficiency, minimizes unscheduled downtime, and improves vehicle longevity through targeted service interventions aligned with actual road-induced wear factors. Moreover, the integration of machine learning into the classification process advances the system 100, delivering data-driven insights and operational readiness across diverse terrains.
In an embodiment, the wear computation module 106 is configured to select a least abrasive surface and a corresponding baseline distance, wherein the wear computation module is configured to assign a baseline coefficient to the least abrasive surface. Specifically, among the classified road surface types provided by the road classifier module 104, the wear computation module 106 identifies the surface with the lowest wear potential based on predefined engineering data or historical wear profiles. For instance, asphalt is designated as the least abrasive surface due to the uniform texture and lower impact on tire and suspension wear compared to gravel or dirt surfaces. Furthermore, the process employed by the wear computation module 106 involves correlating the classified road surfaces with the respective wear profiles. After selecting the least abrasive surface and establishing the baseline distance and coefficient, the wear computation module 106 accesses accumulated distance travelled data segmented by classified road surface type. Moreover, the wear computation module 106 uses the baseline coefficient to determine relative wear impacts by calculating ratios between the baseline distance and the distance travelled on each classified surface. The baseline coefficient provides a standard for interpreting wear across heterogeneous road conditions, ensuring consistency in the effective wear distance computations and enabling informed maintenance decisions. Additionally, the approach enables accurate aggregation of distance metrics, improves predictive maintenance by quantifying the relative wear induced by different surfaces, and ensures fair comparisons across diverse operational environments. The assignment of the baseline coefficient enhances computational efficiency by providing a fixed reference for ratio-based calculations and simplifying model interpretation. Subsequently, the integration of the baseline selection into the wear computation module 106 advances the vehicle maintenance strategies by aligning service schedules with real-world wear patterns, reducing downtime, and optimizing component lifespan while providing robust and scalable deployment across fleets operating in varied terrains.
In an embodiment, the wear computation module 106 is configured to determine a ratio of the baseline distance and a distance travelled associated with each classified surface and compute a plurality of surface-specific coefficients for each classified surface. Specifically, after selecting the least abrasive surface and assigning the baseline coefficient as defined above, the wear computation module 106 retrieves the distance travelled on each classified surface from stored or real-time data. The baseline Distance refers to the reference distance assigned to the least abrasive surface, and Distance Travelled on surface is the cumulative distance covered on that particular surface type. The wear computation module 106 applies the calculations over predefined time intervals, such as, but not limited to, every 100 km travelled or after each trip segment. Further, the computed coefficients are stored in structured data formats, associating each surface-specific coefficient with the corresponding road classification and timestamp, ensuring accurate tracking and reproducibility of wear patterns. The wear computation module 106 integrates the coefficients across all surfaces encountered during the vehicle operation, enabling aggregation and further processing in effective wear distance computations. Moreover, the coefficients are updated dynamically as new data streams from the sensor module 102 and classification output from the road classifier module 104, ensuring real-time responsiveness to changing driving environments. Additionally, the approach enhances the predictive maintenance framework by offering detailed, surface-specific insights into how different terrains contribute to vehicle degradation. The normalization process enables the comparative analysis across heterogeneous road types, supporting informed maintenance planning and resource allocation. Subsequently, the computed coefficients drive accurate, effective wear distance calculations, improving the reliability of maintenance schedules, extending component lifespan, and reducing unscheduled repairs.
In an embodiment, the wear computation module 106 is configured to compute an effective wear distance based on aggregated distance travelled on each classified surface. Specifically, after calculating surface-specific coefficients for each classified surface using the ratio of baseline distance to distance travelled as defined above, the wear computation module 106 multiplies the distance travelled on each surface by the respective coefficient. The wear computation module 106 continuously aggregates the weighted distances in real-time or over defined intervals, such as, but not limited to, trip segments, daily operation, or service cycles. Further, the computed effective wear distance reflects the combined wear impact of all road surfaces encountered by the vehicle during operation and is stored for further processing by the maintenance module 108. The wear computation module 106 ensures that data integrity is preserved by validating distance inputs, filtering out erroneous readings using statistical techniques such as, but not limited to, moving averages and outlier detection algorithms. Moreover, the aggregation process applies weighted summation using matrix operations for efficiency and scalability across multiple surfaces. The module 106 timestamps each data entry and maintains a historical record of the wear distance computations, enabling trend analysis, model recalibration, and validation checks across operational periods. Additionally, by accounting for differences in road abrasiveness through weighted calculations, the system 100 ensures that maintenance schedules correspond to actual wear patterns rather than estimated or fixed intervals. The approach improves maintenance accuracy, reduces operational costs by preventing premature servicing, and extends vehicle component longevity through targeted interventions. Subsequently, the aggregation framework enhances data-driven decision-making, supports fleet optimization, and enables scalability across vehicles operating in varied terrains, all while delivering real-time responsiveness and long-term wear tracking for predictive maintenance planning.
In an embodiment, the maintenance module 108 is configured to perform a comparison between the effective wear distance and a maintenance threshold distance and generate an alert signal based on the comparison. Specifically, the maintenance module 108 receives the computed effective wear distance from the wear computation module 106, which aggregates the distance travelled on each classified surface weighted by the surface-specific coefficients. The maintenance threshold distance is predefined based on manufacturer specifications, historical wear data, and safety margins, and is stored within the maintenance module 108. Further, if the effective wear distance equals or exceeds the maintenance threshold distance, the maintenance module 108 triggers the alert signal to notify operators or the vehicle subsystems that maintenance scheduling must be initiated. The method employed by the maintenance module 108 involves continuous evaluation of the vehicle’s operational state through periodic or real-time comparisons between the effective wear distance and the maintenance threshold distance. Moreover, the maintenance module 108 applies data validation steps to ensure that transient anomalies or sensor noise produce true alerts by using smoothing algorithms and filtering techniques such as, but not limited to, exponential weighted moving averages. The maintenance module 108 stores historical comparison data to support trending analyses and recalibration of maintenance schedules based on usage patterns over extended periods. Consequently, performing the comparison between the effective wear distance and the maintenance threshold distance ensures proactive maintenance interventions aligned with actual vehicle usage and wear profiles. The approach improves resource utilization by targeting maintenance actions precisely when required, thus extending component life, reducing downtime, and lowering operational costs. Subsequently, the integration of continuous comparison logic into the maintenance module 108 establishes a robust and adaptive system for fleet management, ensuring maintenance planning aligns with empirical data and operational realities across diverse driving conditions.
In an exemplary embodiment, the system 100 operates through the interaction of the at least one sensor module 102, the road classifier module 104, the wear computation module 106, and the maintenance module 108. The at least one sensor module 102 includes, but is not limited to, an inertial measurement unit sensor, vibration sensor, vehicle speed sensor, throttle position sensor, brake application sensor, and steering angle sensor. Moreover, the sensors transmit data to the road classifier module 104, which processes vibration signals using Fast Fourier Transform over 10-second windows to extract dominant frequency peaks, RMS values, and spectral energy in frequency bands of 5–50 Hz, 50–200 Hz, and 200–1,000 Hz. The road classifier module 104 assigns the extracted frequency-based attributes to the labeled datasets representing road surfaces such as, but not limited to, asphalt, gravel, and dirt roads. For instance, asphalt may exhibit dominant frequencies at 20 Hz and 75 Hz with RMS values below 0.5 g, whereas gravel shows peaks at 30 Hz and 120 Hz with RMS values around 1.2 g. The road classifier module 104 trains a Random Forest model using 5,000 labeled samples with 80% for training and 20% for validation. After training, the road classifier module 104 classifies incoming data in real-time and outputs road types with associated confidence scores. Additionally, the classification results are transmitted to the wear computation module 106, which selects asphalt as the least abrasive surface, assigns a baseline distance of 10,000 km, and defines the baseline coefficient as 1. For gravel and dirt roads, the wear computation module 106 calculates surface-specific coefficients using the equation:
Coefficient= (Baseline Distance)/(Distance Traveled on Surface)
For instance, if the vehicle travels 3,000 km on asphalt, 1,000 km on gravel, and 500 km on dirt, the coefficients for gravel and dirt are computed as Coefficient (gravel) as 10 and Coefficient (dirt) as 20. The wear computation module 106 then computes the effective wear distance as 23,000 km. The maintenance module 108 compares the effective wear distance with a predefined maintenance threshold distance of 20,000 km using the condition: 23,000 = 20,000. Subsequently, as the condition evaluates as true, the maintenance module 108 generates the alert signal, notifying the operator that maintenance scheduling must be performed. The alert includes, but is not limited to, the total effective wear distance, breakdown by road surface, timestamp of threshold crossing, and recommended maintenance action, such as, but not limited to, tire inspection or suspension check. Ultimately, the maintenance module 108 also stores historical data for trending analysis, enabling recalibration of threshold values based on cumulative wear patterns.
In accordance with a second aspect, there is described a method for predictive maintenance of a vehicle, the method comprising:
sensing at least one operational parameter of the vehicle, via at least one sensor module;
classifying road surface characteristics based on the sensed data, via a road classifier module;
computing an effective wear distance based on aggregated distance travelled on each classified surface, via a wear computation module;
performing a comparison between the effective wear distance and a maintenance threshold distance, via a maintenance module; and
generating an alert signal based on the comparison, via the maintenance module.
Referring to figure 2, in accordance with an embodiment, there is described a method 200 for predictive maintenance of a vehicle. At step 202, the method 200 comprises sensing at least one operational parameter of the vehicle, via at least one sensor module 102. At step 204, the method 200 comprises classifying road surface characteristics based on the sensed data, via a road classifier module 104. At step 206, the method 200 comprises computing an effective wear distance based on aggregated distance travelled on each classified surface, via a wear computation module 106. At step 208, the method 200 comprises performing a comparison between the effective wear distance and a maintenance threshold distance, via a maintenance module 108. At step 210, the method 200 comprises generating an alert signal based on the comparison, via the maintenance module 108.
In an embodiment, the method 200 comprises extracting at least one frequency-based attribute from vibration data received from the sensor module 102, via the road classifier module 104.
In an embodiment, the method 200 comprises assigning the extracted at least one frequency-based attribute to a plurality of labelled datasets and training a machine learning model based on the assigned labelled datasets, via the road classifier module 104.
In an embodiment, the method 200 comprises selecting a least abrasive surface and a corresponding baseline distance and assigning a baseline coefficient to the least abrasive surface, via the wear computation module 106.
In an embodiment, the method 200 comprises determining a ratio of the baseline distance and a distance travelled associated with each classified surface and computing a plurality of surface-specific coefficients for each classified surface, via the wear computation module 106.
In an embodiment, the method 200 comprises sensing at least one operational parameter of the vehicle, via at least one sensor module 102. Further, the method 200 comprises extracting at least one frequency-based attribute from vibration data received from the sensor module 102, via the road classifier module 104. Furthermore, the method 200 comprises assigning the extracted at least one frequency-based attribute to a plurality of labelled datasets and training a machine learning model based on the assigned labelled datasets, via the road classifier module 104. Moreover, the method 200 comprises classifying road surface characteristics based on the sensed data, via the road classifier module 104. Additionally, the method 200 comprises selecting a least abrasive surface and a corresponding baseline distance and assigning a baseline coefficient to the least abrasive surface, via the wear computation module 106. Subsequently, the method 200 comprises determining a ratio of the baseline distance and a distance travelled associated with each classified surface and computing a plurality of surface-specific coefficients for each classified surface, via the wear computation module 106. Further, the method 200 comprises computing an effective wear distance based on aggregated distance travelled on each classified surface, via a wear computation module 106. Furthermore, the method 200 comprises performing a comparison between the effective wear distance and a maintenance threshold distance, via a maintenance module 108. Moreover, the method 200 comprises generating an alert signal based on the comparison, via the maintenance module 108.
The system for predictive maintenance of a vehicle, as described in the present disclosure, is advantageous in terms of improving maintenance accuracy by aligning service schedules with actual road-induced wear patterns derived from real-time sensor data. Further, the road classifier module 104 enables precise identification of the road surface characteristics through frequency-based attributes and machine learning models, ensuring adaptive classification under dynamic driving conditions.
It would be appreciated that all the explanations and embodiments of the system 100 also apply mutatis-mutandis to the method 200.
In the description of the present disclosure, it is also to be noted that, unless otherwise explicitly specified or limited, the terms “disposed,” “mounted,” and “connected” are to be construed broadly, and may for example be fixedly connected, detachably connected, or integrally connected, either mechanically or electrically. They may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meaning of the above terms in the present disclosure can be understood in specific cases to those skilled in the art.
Modifications to embodiments and combinations of different embodiments of the present disclosure described in the foregoing are possible without departing from the scope of the present disclosure as defined by the accompanying claims. Expressions such as “including”, “comprising”, “incorporating”, “have”, and “is” used to describe and claim the present disclosure are intended to be construed in a non-exclusive manner, namely allowing for items, components or elements not explicitly described also to be present. Reference to the singular is also to be construed to relate to the plural where appropriate.
Although embodiments have been described with reference to a number of illustrative embodiments thereof, it should be understood that numerous other modifications and embodiments can be devised by those skilled in the art that will fall within the scope of the principles of this disclosure. More particularly, various variations and modifications are possible in the component parts and/or arrangements of the subject combination arrangement within the scope of the present disclosure, the drawings, and the appended claims. In addition to variations and modifications in the component parts and/or arrangements, alternative uses will also be apparent to those skilled in the art. ,CLAIMS:WE CLAIM:
1. A system (100) for predictive maintenance of a vehicle, the system (100) comprises:
- at least one sensor module (102) configured to sense at least one operational parameter of the vehicle;
- at least one road classifier module (104) communicably coupled to the at least one sensor module (102) and configured to classify road surface characteristics based on the sensed data;
- at least one wear computation module (106) communicably coupled to the at least one road classifier module (104) and configured to compute an effective wear distance; and
- a maintenance module (108) communicably coupled to the at least one wear computation module (106),
wherein the maintenance module (108) is configured to schedule a predictive maintenance based on the classified road surface characteristics and the computed effective wear distance.

2. The system (100) as claimed in claim 1, wherein the at least one sensor module (102) comprises one or more sensors selected from the group of sensors comprising inertial measurement unit sensors, throttle position sensors, brake application sensors, torque demand sensors, steering angle sensors, vehicle speed sensors, and vibration sensors.

3. The system (100) as claimed in claim 1, wherein the road classifier module (104) is configured to extract at least one frequency-based attribute from vibration data received from the sensor module (102).

4. The system (100) as claimed in claim 1, wherein the road classifier module (104) is configured to assign the extracted at least one frequency-based attribute to a plurality of labelled datasets and train a machine learning model based on the assigned labelled datasets.

5. The system (100) as claimed in claim 1, wherein the road classifier module (104) is configured to classify the road surface characteristics based on the trained machine learning model.

6. The system (100) as claimed in claim 1, wherein the wear computation module (106) is configured to select a least abrasive surface and a corresponding baseline distance, wherein the wear computation module is configured to assign a baseline coefficient to the least abrasive surface.

7. The system (100) as claimed in claim 1, wherein the wear computation module (106) is configured to determine a ratio of the baseline distance and a distance travelled associated with each classified surface and compute a plurality of surface-specific coefficients for each classified surface.

8. The system (108) as claimed in claim 1, wherein the wear computation module (106) is configured to compute an effective wear distance based on aggregated distance travelled on each classified surface.

9. The system (100) as claimed in claim 1, wherein the maintenance module (108) is configured to perform a comparison between the effective wear distance and a maintenance threshold distance and generate an alert signal based on the comparison.

10. The method (200) of predictive maintenance of a vehicle, the method (200) comprising:
- sensing at least one operational parameter of the vehicle, via at least one sensor module (102);
- classifying road surface characteristics based on the sensed data, via a road classifier module (104);
- computing an effective wear distance based on aggregated distance travelled on each classified surface, via a wear computation module (106);
- performing a comparison between the effective wear distance and a maintenance threshold distance, via a maintenance module (108); and
- generating an alert signal based on the comparison, via the maintenance module (108).