1. An overall mass estimation system (100) for a vehicle, the system (100) comprising: a memory (102); at least one processor (104) operationally coupled to the memory (102); an input data acquisition module operationally coupled to the at least one processor (104), configured to collect one or more input parameters from pre-exiting vehicle data sources; a tractive force computation module (108) operationally coupled to the at least one processor (104), configured to determine a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data; a resistive force modelling module (110) operationally coupled to the at least one processor (104), configured to calculate longitudinal resistive forces acting on the vehicle, wherein the resistive forced comprises aerodynamic drag force, rolling resistance force, and gravitational grade resistance; an acceleration determination module (112) operationally coupled to the at least one processor (104), configured to compute vehicle acceleration from time-based vehicle speed data; a longitudinal dynamics module (114) operationally coupled to the at least one processor (104), configured to generate a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle; a mass estimation module (116) operationally coupled to the at least one processor (104), configured to rearrange the force balance equation and solve for total vehicle mass including payload wherein the mass estimation module (116) is configured to rearrange the force balance equation and solve for total vehicle mass including payload, and implement a supervised deep learning model (120) trained to learn nonlinear relationships between vehicle operating parameters and vehicle mass, wherein the supervised deep learning model (120) comprises a multi-layer fully connected artificial neural network including an input layer configured to receive vehicle operational signals; a first dense layer comprising 128 neurons with Rectified Linear Unit (ReLU) activation; a second dense layer comprising 64 neurons with ReLU activation; a third dense layer comprising 32 neurons with ReLU activation; and an output layer comprising a single neuron configured to generate a continuous mass estimation output using linear activation; wherein the model is trained using a Mean Squared Error (MSE) loss function and optimized using an Adaptive Moment Estimation (Adam) optimizer with a predefined learning rate; and an output module (118) operationally coupled to the at least one processor (104), configured to provide an estimated vehicle mass value and associated estimation metrics.

2. The system (100) as claimed in claim 1, wherein the one or more parameters comprises wheel radius, drivetrain efficiency, aerodynamic drag coefficient, rolling resistance coefficient, vehicle frontal area, gear ratios, and air density.

3. The system (100) as claimed in claim 1, wherein the input data acquisition module is configured to collect engine torque, engine speed, vehicle speed, and time data from a Controller Area Network (CAN) bus.

4. The system (100) as claimed in claim 1, wherein the acceleration determination module (112) computes acceleration as a derivative of vehicle speed with respect to time.

5. The system (100) as claimed in claim 1, wherein the output module (118) is configured to generate a mass estimation report including error margins and safety warnings related to overloading.

6. The system (100) as claimed in claim 1, wherein the at least one processor (104) is configured to validate the estimated mass against known curb weight or gross vehicle weight value when available.

7. The system (100) as claimed in claim 1, wherein no additional hardware sensors are installed on the vehicle for the mass measurement.

8. A method (200) for operating overall mass estimation system (100) for a vehicle, the method (200) comprising: collecting, via an input data acquisition module (106), one or more input parameters from pre-exiting vehicle data sources; determining, via a tractive force computation module (108), a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data; calculating, via a resistive force modelling module (110), longitudinal resistive forces acting on the vehicle; computing, via an acceleration determination module (112), vehicle acceleration from time-based vehicle speed data; generating, via a longitudinal dynamics module (114), a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle; rearranging, via a mass estimation module (116), force balance equation and solve for total vehicle mass including payload; and providing, via an output module (118), an estimated vehicle mass value and associated estimation metrics.

Description:OVERALL MASS ESTIMATION SYSTEM FOR A VEHICLE
FIELD OF THE DISCLOSURE
This invention generally relates to vehicle dynamics, automotive analytics, and computational estimation systems, and more particularly to an overall mass estimation system for a vehicle.
BACKGROUND OF THE INVENTION

The subject matter discussed in the background section should not be assumed to be prior art merely as a result of its mention in the background section. Similarly, a problem mentioned in the background section or associated with the subject matter of the background section should not be assumed to have been previously recognized in the prior art. The subject matter in the background section merely represents different approaches, which in and of themselves may also correspond to implementations of the claimed technology.
Accurate determination of a vehicle’s overall mass is essential for performance evaluation, design validation, regulatory compliance, and operational safety. Conventional approaches primarily rely on physical weighing using weighbridges, dynamometer setups, or dedicated test rigs. The methods are often expensive, time-consuming, and dependent on specialized infrastructure, making them impractical during early vehicle development stages when multiple prototypes or design variants must be assessed quickly. As a result, engineers frequently rely on estimated or assumed mass values, which may lead to inaccurate predictions of fuel consumption, emissions, braking performance, durability, and other key performance indicators, thereby slowing development cycles and increasing the risk of design iterations.
In addition to development challenges, lack of reliable knowledge of actual vehicle mass during real-world operation may lead to safety and compliance issues. Vehicles operating with unknown or underestimated loads may experience excess loading, causing increased mechanical wear, reduced stability, longer stopping distances, and potential violations of road safety regulations. Existing indirect mass estimation techniques attempt to address these issues but often require additional sensors, instrumentation, or complex hardware installations, which add cost, system complexity, and maintenance burden. Accordingly, there remains a need for a reliable, cost-effective, and infrastructure-independent technique capable of estimating vehicle mass using already available data without requiring dedicated weighing systems or added sensors.
According to the patent application number “WO2013075280A1” titled “Vehicle mass estimation method and system”, discloses a vehicle mass estimation method and system, wheel speed, driving torque and longitudinal acceleration of the vehicle are obtained, and an estimated vehicle mass is calculated using an estimation equation group which comprises wheel speed, driving torque and longitudinal acceleration as input parameters, and vehicle mass and driving resistance of the vehicle as variables. Lower and upper thresholds are considered in the calculating process of the estimated vehicle mass. The reference discloses about a method and system for estimating the mass of a vehicle, particularly an electric vehicle, using a longitudinal dynamic model that relates wheel speed, driving torque and acceleration to determine vehicle mass and driving resistance, which combines both rolling and air resistances into a single variable. An estimation equation group is used to calculate the estimated vehicle mass, taking into account driving resistance. The estimation is preferably performed during the vehicle’s starting phase, where the vehicle mass may be treated as constant during operation. However, the process may also be repeated during driving if needed. However, the reference does not disclose aerodynamic drag, rolling resistance and grade resistance models as parameters for mass calculation.
According to another patent application number “IN202311068636” titled “A method for estimating mass of a moving vehicle”, discloses a method for estimating mass of a moving vehicle. The method comprises determining an overall ratio, a Coefficient of Rolling Resistance (CoRR) (224), and an α angle based on pre-processed vehicle parameters (102) and data (104) received from one or more sensors installed in a vehicle. The one or more parameters determined by pre-processing are used to determine a first and a second set of parameters. The first set of parameters is used to determine an acceleration factor (denominator) (112). The second set of parameters is used to determine a force factor (118). The force factor (numerator) (118) and the acceleration factor (denominator) (112) are used to determine a raw mass (122), and the raw mass (122) is used to compute an actual mass of the vehicle. The reference discloses about vehicle mass estimation is performed while the vehicle is in motion using data collected from onboard sensors via the Controller Area Network (CAN) bus. Predefined vehicle parameters such as tire radius, coefficient of rolling resistance, and transmission efficiency are stored in the vehicle memory. Sensor data including torque, speed, acceleration, slope angle and environmental conditions are collected from devices like the IMU and speed sensors. These inputs are preprocessed to calculate force variables such as net engine torque, aerodynamic resistance, effective gearing and inertial force. Using these force and acceleration factors, a raw vehicle mass is continuously estimated over time and multiple estimations are aggregated to determine the final vehicle mass. However, the reference does not disclose about integrating aerodynamic drag, rolling resistance and grade resistance models with driveline parameters to estimate vehicle mass.
The cited prior arts for vehicle mass estimation suffer from several limitations. The system disclosed in WO2013075280A1 relies on simplified resistance modelling and does not separately account for aerodynamic drag, rolling resistance, and grade resistance as distinct physical components within a comprehensive longitudinal force balance, which may limit estimation accuracy under varying real-world driving conditions. Its approach is also oriented toward specific operating phases, such as vehicle start, and may not robustly address continuously changing dynamics. Meanwhile, the method described in IN202311068636 depends heavily on multiple onboard sensors and extensive signal pre-processing, increasing hardware dependency, system complexity, calibration requirements, and cost. Neither prior art provides a structured, sensor-free analytical framework that integrates driveline parameters with distinct resistive force models using only existing design and departmental data, thereby leaving a gap for a simpler, infrastructure-independent, and development-friendly solution for reliable vehicle mass estimation.
OBJECTIVES OF THE INVENTION
An objective of the invention is to provide an overall mass estimation system for a vehicle.
Furthermore, the objective of the invention is to provide a method for operating the overall mass estimation system for a vehicle.
Furthermore, the objective of the invention is to provide the system that enable fast, repeatable, and cost-effective mass estimation during early vehicle development stages, where access to weighbridges or test infrastructure may be limited or impractical.
Furthermore, the objective of invention is to provide the system that improve accuracy of mass estimation by distinctly modelling aerodynamic drag, rolling resistance, and gravitational grade resistance as separate resistive force components within a structured force balance framework.
Furthermore, the objective of the present invention is to provide the system to detect and prevent vehicle overloading, enhancing road safety, reducing mechanical wear, and supporting compliance with regulatory load limits.
SUMMARY
The invention relates to an overall mass estimation system of a vehicle.
According to an aspect, an overall mass estimation system of a vehicle. The system comprising a memory, at least one processor operationally coupled to the memory. Further, an input data acquisition module operationally coupled to the at least one processor, configured to collect one or more input parameters from pre-exiting vehicle data sources. Further, a tractive force computation module operationally coupled to the at least one processor, configured to determine a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data. Further, a resistive force modelling module operationally coupled to the at least one processor, configured to calculate longitudinal resistive forces acting on the vehicle, wherein the resistive forced comprises aerodynamic drag force, rolling resistance force, and gravitational grade resistance. Further, an acceleration determination module operationally coupled to the at least one processor, configured to compute vehicle acceleration from time-based vehicle speed data. Further, a longitudinal dynamics module operationally coupled to the at least one processor, configured to generate a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle. Further, a mass estimation module operationally coupled to the at least one processor, configured to rearrange the force balance equation and solve for total vehicle mass including payload; and an output module operationally coupled to the at least one processor, configured to provide an estimated vehicle mass value and associated estimation metrics.
According to an another aspect, a method for operating overall mass estimation system for a vehicle. Further, the method comprising collecting, via an input data acquisition module, one or more input parameters from pre-exiting vehicle data sources. Further, determining, via a tractive force computation module, a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data. Further, calculating, via a resistive force modelling module, longitudinal resistive forces acting on the vehicle. Further, computing, via an acceleration determination module, vehicle acceleration from time-based vehicle speed data. Further, generating, via a longitudinal dynamics module, a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle. Further, rearranging, via a mass estimation module, force balance equation and solve for total vehicle mass including payload. Further, providing, via an output module, an estimated vehicle mass value and associated estimation metrics.
 
BRIEF DESCRIPTION OF THE DRAWINGS
The accompanying drawings illustrate the embodiment of the system. Any person with ordinary skills in the art will appreciate that the illustrated element boundaries (e.g., boxes, groups of boxes, or other shapes) in the figures represent one example of the boundaries. It may be that in some examples one element may be designed as multiple elements or that multiple elements may be designed as one element. In some examples, an element shown as an internal component of one element may be implemented as an external component in another, and vice versa. Furthermore, elements may not be drawn to scale. Non-limiting and non-exhaustive descriptions are described with reference to the following drawings. The components in the figures are not necessarily to scale, emphasis instead being placed upon illustrating principles.
FIG. 1 illustrates a block diagram of an overall mass estimation system of a vehicle, according to an embodiment of the present invention.
FIG. 2 illustrates an architectural representation of a deep learning module, according to an embodiment of the present invention.
FIG. 3 illustrates a flow chart of a method for operating the overall mass estimation system of a vehicle, according to an embodiment of the present invention.
FIG. 4 illustrates a tabular and graphical representation of a vehicle operational data collected across all available gear conditions without applying additional filtering or gear-specific segregation, according to an embodiment of the present invention.
FIG. 5 illustrates a tabular and graphical representation of operational data restricted to higher gears (5, 6, and 7) using deep learning model (120), according to an embodiment of the present invention.
 
DETAILED DESCRIPTION OF THE INVENTION
Some embodiments of this disclosure, illustrating all its features, will now be discussed in detail. The words “comprising,” “having,” “containing,” and “including,” and other forms thereof, are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items or meant to be limited to only the listed item or items. It must also be noted that as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.
Although any systems and methods similar or equivalent to those described herein can be used in the practice or testing of embodiments of the present disclosure, the preferred, systems and methods are now described. Embodiments of the present disclosure will be described more fully hereinafter with reference to the accompanying drawings in which like numerals represent like elements throughout the several figures, and in which example embodiments are shown. Embodiments of the claims may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. The examples set forth herein are non-limiting examples and are merely examples among other possible examples.
The present invention discloses an overall mass estimation system of a vehicle.
FIG. 1 illustrates a block diagram of an overall mass estimation system (100) of a vehicle, according to an embodiment of the present invention. FIG. 2 illustrates an architectural representation (200) of a deep learning module, according to an embodiment of the present invention.
In some embodiments, the overall mass estimation system (100) of a vehicle comprises a memory (102), at least one processor (104), input data acquisition module (106), tractive force computation module (108), resistive force modelling module (110), acceleration determination module (112), longitudinal dynamics module (114), mass estimation module (116), and an output module (118).
In some embodiments, the system (100) comprises the memory (102) configured to store program instructions, vehicle parameters, analytical models, and intermediate computational data, and the at least one processor (104) operationally coupled to the memory (102). Further, the at least one processor (104) configured to execute the stored instructions to perform data acquisition, force modelling, longitudinal dynamics calculations, and vehicle mass estimation. The at least one processor (104) retrieves input datasets and mathematical models from the memory (102), processes real-time and stored data, and generates estimated mass outputs and related metrics to enable automated and repeatable execution of the mass estimation methodology.
In some embodiments, the input data acquisition module operationally coupled to the at least one processor (104). The input data acquisition module configured to collect one or more input parameters from the pre-exiting vehicle data sources. Further, the one or more parameters comprises wheel radius, drivetrain efficiency, aerodynamic drag coefficient, rolling resistance coefficient, vehicle frontal area, gear ratios, and air density. The input data acquisition module configured to gather design and engineering parameters such as wheel radius, gear ratios, drivetrain efficiency, aerodynamic drag coefficient, rolling resistance coefficient, and frontal area from stored databases or departmental records, and may also obtain operational data such as engine torque, engine speed, vehicle speed, and time information from existing vehicle communication networks. The collected data is formatted and supplied to the at least one processor (104) for use in subsequent force modelling and vehicle mass estimation computations.
In some embodiments, the tractive force computation module (108) operationally coupled to the at least one processor (104). Further, the tractive force computation module (108) is configured to determine a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data. The tractive force computation module (108) is configured to process engine torque and engine speed data along with gear ratio information, drivetrain or transmission efficiency, and wheel radius to determine the effective tractive force generated at the tire–road interface. The computed driving force represents the propulsion input used in the longitudinal dynamics model for subsequent estimation of vehicle mass.
In some embodiments, the resistive force modelling module (110) operationally coupled to the at least one processor (104). The resistive fore modelling module is configured to calculate longitudinal resistive forces acting on the vehicle. Further, the resistive forced comprises aerodynamic drag force, rolling resistance force, and gravitational grade resistance. The resistive force modelling module (110) is configured to determine aerodynamic drag force based on air density, drag coefficient, frontal area, and relative vehicle speed using the following equation:
Air Resistance (F_air): F_air=1/2.ρ.C_d.A.v_relative^2
Further, rolling resistance force is determined based on the rolling resistance coefficient and gravitational acceleration using the following equation:
Rolling Resistance (F_rolling): F_rolling=C_r.m.g
Further, the gravitational grade resistance is determined based on vehicle mass, gravitational acceleration, and road slope angle using the following equation:
Gravitational Resistance (F_gravity): F_gravity=m.g.sin⁡θ
These resistive force components are provided to the at least one processor (104) for use in a longitudinal force balance for vehicle mass estimation.
In some embodiments, the acceleration determination module (112) operationally coupled to the at least one processor (104). Further, the acceleration determination module (112) is configured to compute vehicle acceleration from time-based vehicle speed data. Further, the acceleration determination module (112) computes acceleration as a derivative of vehicle speed with respect to time. The acceleration determination module (112) is configured to process sequential speed samples associated with time stamps and computes acceleration as a temporal derivative, expressed as follow:
a=(dv_vehicle)/dt=(Change in vehicle speed)/(change in time)
The equation shows the rate of change of vehicle speed over time. To improve numerical stability and accuracy, the acceleration determination module (112) may apply signal conditioning techniques including noise filtering, smoothing, or window-based differentiation to reduce the effect of sensor noise and transient disturbances. Further, the resulting acceleration value is synchronized with corresponding drivetrain data and supplied to the at least one processor (104) for use in a longitudinal force balance model.
In some embodiments, the longitudinal dynamics module (114) operationally coupled to the at least one processor (104). The longitudinal dynamics module (114) configured to generate a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle. The longitudinal dynamics module (114) is configured to receive the computed tractive driving force and the modelled resistive forces, including aerodynamic drag, rolling resistance, and gravitational grade resistance, and combines these with the inertial force associated with vehicle acceleration. Using Newton’s second law of motion, the longitudinal dynamics module (114) formulates a longitudinal force balance equation in which the net propulsion force acting on the vehicle is equal to the product of vehicle mass and acceleration. The equation serves as the core analytical framework that links propulsion inputs and opposing resistances to vehicle inertial response, enabling subsequent determination of unknown vehicle mass by the at least one processor (104).
The net force acting on the vehicle is:
F_drive-F_air-F_rolling-F_gravity=ma
In some embodiments, the mass estimation module (116) operationally coupled to the at least one processor (104). Further, the mass estimation module (116) is configured to rearrange the force balance equation and solve for total vehicle mass including payload. The mass estimation module (116) using the at least one processor (104) is configured to algebraically rearrange the longitudinal force balance equation generated by the longitudinal dynamics module (114). Using the relationship in which the net tractive force minus aerodynamic drag, rolling resistance, and gravitational grade resistance equals the inertial force of the vehicle, the mass estimation module (116) is configured to isolate the mass variable and computes it based on known propulsion inputs and measured acceleration. In one implementation, the mass estimation module (116) evaluates mass using a power-based driving force expression and solves an equation:

m=((P_engine.η_DT)/v_vehicle -(1/2.ρ.C_d.A.v_relative^2+m.g.sin⁡θ+C_r.m.g))/a

m=((P_engine. η_DT)/v_vehicle - 1/2.ρ.C_d.A.v_relative^2)/(a+m.g.sin⁡θ+C_r.m.g)

The mass estimation module (116) may employ iterative, recursive, or optimization-based numerical techniques to converge on a stable mass estimate, especially under dynamic driving conditions. The computed mass value is then provided to subsequent modules for reporting, validation, or overload assessment.
In some embodiments, the at least one processor (104) is configured to validate the estimated mass against known curb weight or gross vehicle weight value when available. The at least one processor (104) is further configured to perform the validation routine in which the estimated vehicle mass is compared against reference mass values such as a known curb weight, rated gross vehicle weight (GVW), or previously recorded baseline data when such information is available. During the process, the at least one processor (104) is configured to determine a deviation or error margin between the estimated mass and the reference value, and may apply statistical checks, threshold comparisons, or confidence interval analysis to assess estimation reliability. If discrepancies exceed predefined limits, the at least one processor (104) may trigger recalibration of model parameters, flag the estimation as uncertain, or generate a diagnostic or warning output, thereby improving robustness, traceability, and confidence in the mass estimation results.
In some embodiments, the mass estimation module (116) is configured to estimate the total vehicle mass including payload through a hybrid computational approach combining physics-based modelling and supervised deep learning. In a first stage, the mass estimation module (116) rearranges the force balance equation generated by the longitudinal dynamics module to analytically solve for vehicle mass based on the relationship between tractive force, resistive forces, and inertial force. In a second stage, the mass estimation module (116) implements a deep learning model (120) trained to capture nonlinear dependencies between vehicle operational parameters and actual vehicle mass, thereby improving estimation accuracy under dynamic and real-world operating conditions where modelling uncertainties, noise, and parameter variations may exist.
The deep learning model (120) comprises a multi-layer fully connected artificial neural network architecture. The network includes an input layer configured to receive vehicle operational signals such as engine torque, engine speed, vehicle speed, acceleration, and other derived parameters. The input layer is followed by a first dense layer comprising 128 neurons with Rectified Linear Unit (ReLU) activation, a second dense layer comprising 64 neurons with ReLU activation, and a third dense layer comprising 32 neurons with ReLU activation. These hidden layers progressively learn hierarchical nonlinear feature representations associated with vehicle load behaviour. The network further comprises an output layer including a single neuron with linear activation configured to generate a continuous numerical output corresponding to the estimated vehicle mass.
The deep learning model (120) is trained using historical and labeled vehicle operational datasets, wherein known vehicle mass values are used as ground truth. During training, a Mean Squared Error (MSE) loss function is employed to minimize the deviation between predicted mass and actual mass values. Optimization of the neural network parameters is performed using an Adaptive Moment Estimation (Adam) optimizer with a predefined learning rate, enabling efficient convergence and stable training performance. The trained deep learning model (120) is deployed within the mass estimation module (116) to provide real-time mass prediction and enhanced robustness against signal noise and nonlinear system behaviour.
In some embodiments, the output module (118) operationally coupled to the at least one processor (104). Further, the output module (118) is configured to provide an estimated vehicle mass value and associated estimation metrics. An output module (118) is operationally coupled to the at least one processor (104) and is configured to present the results of the vehicle mass estimation process to users, on-board system, or remote platforms. The output module (118) provides the calculated vehicle mass along with associated estimation metrics such as confidence intervals, uncertainty values, deviation from reference weights, and data quality indicators. Further, the output module (118) is configured to generate a comprehensive mass estimation report that may include historical trends, validation results, and computed error margins. When the estimated mass exceeds predefined safety or regulatory thresholds, the output module (118) further issues alerts or safety warnings related to potential vehicle overloading, thereby supporting operational safety, compliance monitoring, and maintenance decision-making.
FIG. 3 illustrates a flow chart of a method (300) for operating the overall mass estimation system (100) of a vehicle, according to an embodiment of the present invention.
At step 302, the input data acquisition module (106) is configured to collect one or more input parameters from pre-exiting vehicle data sources. The input data acquisition module (106) configured to collect one or more input parameters from the pre-exiting vehicle data sources. Further, the one or more parameters comprises wheel radius, drivetrain efficiency, aerodynamic drag coefficient, rolling resistance coefficient, vehicle frontal area, gear ratios, and air density. The input data acquisition module (106) configured to gather design and engineering parameters such as wheel radius, gear ratios, drivetrain efficiency, aerodynamic drag coefficient, rolling resistance coefficient, and frontal area from stored databases or departmental records, and may also obtain operational data such as engine torque, engine speed, vehicle speed, and time information from existing vehicle communication networks.
At step 304, the tractive force computation module (108) is configured to determine a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data. The tractive force computation module (108) is configured to determine a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data. The tractive force computation module (108) is configured to process engine torque and engine speed data along with gear ratio information, drivetrain or transmission efficiency, and wheel radius to determine the effective tractive force generated at the tire–road interface. The computed driving force represents the propulsion input used in the longitudinal dynamics model for subsequent estimation of vehicle mass.
At step 306, the resistive force modelling module (110) is configured to calculate longitudinal resistive forces acting on the vehicle. The resistive fore modelling module is configured to calculate longitudinal resistive forces acting on the vehicle. Further, the resistive forced comprises aerodynamic drag force, rolling resistance force, and gravitational grade resistance.
At step 308, the acceleration determination module (112) is configured to compute vehicle acceleration from time-based vehicle speed data. Further, the acceleration determination module (112) is configured to compute vehicle acceleration from time-based vehicle speed data. Further, the acceleration determination module (112) computes acceleration as a derivative of vehicle speed with respect to time.
At step 310, the longitudinal dynamics module (114) is configured to generate the force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle. The longitudinal dynamics module (114) configured to generate a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle. The longitudinal dynamics module (114) is configured to receive the computed tractive driving force and the modelled resistive forces, including aerodynamic drag, rolling resistance, and gravitational grade resistance, and combines these with the inertial force associated with vehicle acceleration.
At step 312, the mass estimation module (116) is configured to rearrange force balance equation and solve for total vehicle mass including payload. Further, the mass estimation module (116) is configured to rearrange the force balance equation and solve for total vehicle mass including payload. The mass estimation module (116) using the at least one processor (104) is configured to algebraically rearrange the longitudinal force balance equation generated by the longitudinal dynamics module (114).
At step 314, the output module (118) is configured to provide an estimated vehicle mass value and associated estimation metrics. Further, the output module (118) is configured to provide an estimated vehicle mass value and associated estimation metrics. An output module (118) is operationally coupled to the at least one processor (104) and is configured to present the results of the vehicle mass estimation process to users, on-board systems, or remote platforms.
FIG. 4 illustrates a tabular and graphical representation (400) of a vehicle operational data collected across all available gear conditions without applying additional filtering or gear-specific segregation, according to an embodiment of the present invention.
In some embodiments, the supervised deep learning model (120) is trained using vehicle operational data collected across all available gear conditions without applying additional filtering or gear-specific segregation. This ensures that the neural network learns generalized nonlinear relationships between engine parameters, vehicle dynamics, and overall vehicle mass under diverse real-world driving scenarios. The training and validation loss curves demonstrate rapid convergence during the initial epochs, followed by stable minimization, indicating effective learning behavior without significant overfitting. Both training and validation Mean Absolute Error (MAE) curves closely follow each other, confirming good generalization capability of the model.
The predicted versus actual vehicle mass plot shows a strong linear correlation across all load categories, with predictions closely aligned to the ideal prediction line. This confirms that the network successfully captures the underlying mass-dependent dynamics of the vehicle system. Even at higher payload categories (cat4 and cat5), the deviation remains relatively small, demonstrating robustness under heavy-load conditions.
As illustrated in table, the MAE across load categories ranges from 95.45 kg (cat1) to 832.66 kg (cat3). Although cat3 exhibits comparatively higher MAE, the percentage error analysis in table shows that the relative error remains below 1% for all categories, with values ranging between 0.22% and 0.70%. This indicates that absolute error magnitude increases with higher vehicle mass, but proportional prediction accuracy remains consistently high. The average predicted mass per category closely matches the actual mass values, with percentage errors well within acceptable automotive engineering tolerances.
Table: Mean Absolute Error (MAE) of Vehicle Mass Prediction Across Load Categories
Category MAE (kg)
cat1 95.45
cat2 232.81
cat3 832.66
cat4 443.96
cat5 207.67

Table: Average Predicted Vehicle Mass per Category and Error with Respect to Actual Load
Category Actual Mass (kg) Avg of Predicted Mass (kg) Error (kg) Percentage Error (%)
cat1 5802 5782.41 19.59 0.34
cat2 8977 9040.07 -63.07 0.70
cat3 12151 12123.85 27.15 0.22
cat4 15326 15371.16 -45.16 0.29
cat5 18500 18384.88 115.12 0.62

Overall, the deep learning approach trained with all gear data demonstrates high prediction accuracy, stable convergence characteristics, and strong generalization across varying load conditions. The model effectively integrates multi-gear operational data to provide reliable real-time vehicle mass estimation without requiring additional hardware sensors.
FIG. 5 illustrates a tabular and graphical representation (500) of operational data restricted to higher gears (5, 6, and 7) using deep learning model (120), according to an embodiment of the present invention.
In some embodiments, the deep learning model (120) is trained using operational data restricted to higher gears (5, 6, and 7). This selective dataset focuses on steady-state and highway driving conditions, where gear shifts are minimal and torque transmission characteristics are relatively stable. The training and validation MAE curves exhibit rapid convergence within the initial epochs, followed by smooth stabilization, indicating efficient learning and good generalization. The close alignment between training and validation loss curves further confirms that the model does not suffer from significant overfitting despite the reduced dataset scope.
The predicted versus actual mass plot demonstrates strong linear correlation across all load categories, with predicted values closely following the ideal reference line. From Table, the MAE values show improvement in lower load categories (cat1 and cat2), with errors of 61.50 kg and 120.16 kg respectively, compared to the all-gear model. Although moderate MAE values are observed for higher load categories (cat3 to cat5), the performance remains consistent and stable.
Table indicates that the percentage error across all categories remains below 0.5%, ranging from 0.07% to 0.50%. The predicted average vehicle mass for each category is very close to the actual mass, with minimal deviation even under heavy load conditions. These results suggest that training the model using higher gear data enhances prediction stability under steady driving conditions, while maintaining high overall mass estimation accuracy suitable for real-time implementation.
Table: Mean Absolute Error (MAE) of Vehicle Mass Prediction Across Load Categories
Category MAE (kg)
cat1 61.50
cat2 120.16
cat3 562.05
cat4 652.01
cat5 254.35
Table: Average Predicted Vehicle Mass per Category and Error with Respect to Actual Load
Category Actual Mass (kg) Predicted Mass (kg) Error (kg) Percentage Error (%)
cat1 5802 5807.55 -5.55 0.10
cat2 8977 8983.20 -6.20 0.07
cat3 12151 12206.91 -55.91 0.46
cat4 15326 15248.80 77.20 0.50
cat5 18500 18412.75 87.25 0.47

Filtering the dataset to gears 5, 6, and 7 results in improved model stability and enhanced prediction accuracy, particularly for lower and moderate load categories, as these higher gears typically represent steady-state driving conditions with reduced transient effects and smoother torque delivery. Consequently, the subset gear model exhibits comparatively lower errors and more consistent convergence behavior. In the overall comparative analysis, the deep learning approach significantly outperforms the purely physics-based model in terms of prediction accuracy, adaptability, and robustness across varying load categories. While the physics-based model offers interpretability through explicit force balance relationships, it remains sensitive to parameter estimation, environmental variations, and simplifying assumptions. In contrast, the deep learning model (120) effectively captures complex nonlinear interactions among vehicle parameters, thereby delivering superior mass estimation performance under diverse real-world operating conditions.
The present invention offers a cost-effective, accurate, and scalable solution for estimating vehicle mass without requiring additional sensors, weighing infrastructure, or hardware modifications. By relying solely on existing vehicle design data and operational parameters, the system (100) enables fast and repeatable analytical mass estimation that is particularly valuable during early development and prototype stages. The approach is simple and safe to implement since no physical vehicle handling or test setups are needed. Accuracy is maintained through physics-based modelling and quality input data, while the system (100) remains adaptable across multiple vehicle variants through parameter updates. Additionally, the present invention supports periodic or continuous estimation for indirect load monitoring, thereby improving safety, compliance, and operational efficiency.
It has thus been seen the overall mass estimation system (100) of a vehicle, as described. The overall mass estimation system (100) of a vehicle any case could undergo numerous modifications and variants, all of which are covered by the same innovative concept; moreover, all of the details can be replaced by technically equivalent elements. In practice, the components used, as well as the numbers, shapes, and sizes of the components can be whatever according to the technical requirements. The scope of protection of the invention is therefore defined by the attached claims.

, Claims:I/We claim,
1. An overall mass estimation system (100) for a vehicle, the system (100) comprising:
a memory (102);
at least one processor (104) operationally coupled to the memory (102);
an input data acquisition module operationally coupled to the at least one processor (104), configured to collect one or more input parameters from pre-exiting vehicle data sources;
a tractive force computation module (108) operationally coupled to the at least one processor (104), configured to determine a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data;
a resistive force modelling module (110) operationally coupled to the at least one processor (104), configured to calculate longitudinal resistive forces acting on the vehicle, wherein the resistive forced comprises aerodynamic drag force, rolling resistance force, and gravitational grade resistance;
an acceleration determination module (112) operationally coupled to the at least one processor (104), configured to compute vehicle acceleration from time-based vehicle speed data;
a longitudinal dynamics module (114) operationally coupled to the at least one processor (104), configured to generate a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle;
a mass estimation module (116) operationally coupled to the at least one processor (104), configured to rearrange the force balance equation and solve for total vehicle mass including payload
wherein the mass estimation module (116) is configured to rearrange the force balance equation and solve for total vehicle mass including payload, and implement a supervised deep learning model (120) trained to learn nonlinear relationships between vehicle operating parameters and vehicle mass,
wherein the supervised deep learning model (120) comprises a multi-layer fully connected artificial neural network including an input layer configured to receive vehicle operational signals; a first dense layer comprising 128 neurons with Rectified Linear Unit (ReLU) activation; a second dense layer comprising 64 neurons with ReLU activation; a third dense layer comprising 32 neurons with ReLU activation; and an output layer comprising a single neuron configured to generate a continuous mass estimation output using linear activation;
wherein the model is trained using a Mean Squared Error (MSE) loss function and optimized using an Adaptive Moment Estimation (Adam) optimizer with a predefined learning rate; and
an output module (118) operationally coupled to the at least one processor (104), configured to provide an estimated vehicle mass value and associated estimation metrics.
2. The system (100) as claimed in claim 1, wherein the one or more parameters comprises wheel radius, drivetrain efficiency, aerodynamic drag coefficient, rolling resistance coefficient, vehicle frontal area, gear ratios, and air density.
3. The system (100) as claimed in claim 1, wherein the input data acquisition module is configured to collect engine torque, engine speed, vehicle speed, and time data from a Controller Area Network (CAN) bus.
4. The system (100) as claimed in claim 1, wherein the acceleration determination module (112) computes acceleration as a derivative of vehicle speed with respect to time.
5. The system (100) as claimed in claim 1, wherein the output module (118) is configured to generate a mass estimation report including error margins and safety warnings related to overloading.
6. The system (100) as claimed in claim 1, wherein the at least one processor (104) is configured to validate the estimated mass against known curb weight or gross vehicle weight value when available.
7. The system (100) as claimed in claim 1, wherein no additional hardware sensors are installed on the vehicle for the mass measurement.
8. A method (200) for operating overall mass estimation system (100) for a vehicle, the method (200) comprising:
collecting, via an input data acquisition module (106), one or more input parameters from pre-exiting vehicle data sources;
determining, via a tractive force computation module (108), a driving force acting on the vehicle using engine torque, engine speed, gear ratio, drivetrain efficiency, and wheel radius data;
calculating, via a resistive force modelling module (110), longitudinal resistive forces acting on the vehicle;
computing, via an acceleration determination module (112), vehicle acceleration from time-based vehicle speed data;
generating, via a longitudinal dynamics module (114), a force balance equation representing a relationship between the driving force, the resistive forces, and inertial force of the vehicle;
rearranging, via a mass estimation module (116), force balance equation and solve for total vehicle mass including payload; and
providing, via an output module (118), an estimated vehicle mass value and associated estimation metrics.