Description:FIELD OF THE DISCLOSURE:
[001] The present application relates to the field of condition monitoring and predictive maintenance, and more particularly to a system and a method for estimating remaining useful life of a machinery component based on surface-defect progression.

BACKGROUND
[002] In industrial and defense applications, the service life of critical components is commonly assessed using point-based manual inspections, fixed replacement intervals, usage counters, or indirect condition-monitoring parameters using manual/digital gauges. However, these methods cannot provide the complete volumetric status of the components from all three dimensions. These approaches do not directly measure actual surface degradation, erosion, or localized deformation, particularly in components with complex geometry or non-uniform wear patterns, such as rotating machinery parts, load-bearing elements, and tubular components, including barrels. Consequently, components may be replaced prematurely, increasing cost and downtime, or retained beyond safe operating limits, thereby increasing the risk of failure.
[003] Existing remaining useful life (RUL) and predictive maintenance solutions fails to derive RUL directly from high-resolution three-dimensional surface geometry and defect progression, and instead rely primarily on indirect sensor signals, usage counters, or subjective inspection inputs, with the whole process taking a long time.
Furthermore, no known prior art employs a precise digital twin-based alignment workflow capable of performing deviation analysis between scanned in-service components and reference original equipment manufacturer (OEM) digital models to quantitatively map wear, material loss, and localized deformation.
[004] Current solutions do not provide a selectable multi-model RUL estimation architecture that allows users to choose among degradation trend modeling, artificial intelligence (AI)-based temporal prediction, and physics-based fatigue life analysis based on data availability, component geometry complexity, and required prediction accuracy.
Additionally, existing degradation-based RUL models generally lack Bayesian inference mechanisms to explicitly account for measurement noise, inspection variability, and uncertainty, and therefore do not generate probabilistic RUL outputs with confidence bounds.
[005] Physics-based fatigue life estimation tools available in the prior art are typically standalone analytical or simulation tools and are not integrated with machine-learning correction factors within an automated RUL prediction system.
[006] Moreover, no existing system is known to generate spatially resolved life-consumption maps from geometric degradation data to identify critical surface regions contributing disproportionately to component life consumption for localized, condition-based maintenance decisions.
[007] Therefore, there is a need for enhanced techniques to address the above-mentioned deficiencies.

SUMMARY
[008] This summary is provided to introduce a selection of concepts, in a simplified format, that are further described in the detailed description of the disclosure. This summary is neither intended to identify key nor essential inventive concepts of the disclosure, nor is it intended to determine the scope of the disclosure.
[009] According to an embodiment, a system for estimating remaining useful life of a machinery component based on surface-defect progression is provided. The system includes a scanning apparatus configured to capture surface geometry data of the machinery component. Further, the system includes a memory storing a reference digital model of the machinery component. Furthermore, the system includes a processing unit in communication with the memory. The processing unit is configured to generate a three-dimensional digital model from the captured surface geometry data. Additionally, the processing unit is configured to match the three-dimensional digital model with the reference digital model of the machinery component. Moreover, the processing unit is configured to extract a set of quantitative degradation metrics for identifying the surface-defect progression by performing deviation analysis between the three-dimensional digital model and the reference digital model. Furthermore, the processing unit is configured to estimate the remaining useful life of the machinery component based on the set of quantitative degradation metrics.
[010] According to an embodiment, a method for estimating remaining useful life of a machinery component based on surface-defect progression is provided. The method includes capturing surface geometry data of the machinery component. Further, the method includes generating a three-dimensional digital model from the captured surface geometry data. Furthermore, the method includes matching the three-dimensional digital model with a reference digital model of the machinery component. Additionally, the method includes extracting a set of quantitative degradation metrics for identifying the surface-defect progression by performing deviation analysis between the matched three-dimensional digital model and the reference digital model. Still further, estimating the remaining useful life of the machinery component based on the set of quantitative degradation metrics.
[011] To further clarify the advantages and features of the present disclosure, a more particular description of the disclosure will be rendered by reference to specific embodiments thereof, which are illustrated in the appended drawings. It is appreciated that these drawings depict only typical embodiments of the disclosure and are therefore not to be considered limiting of its scope. The disclosure will be described and explained with additional specificity and detail in the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS
[012] Features, aspects, and advantages of embodiments of the disclosure will be described below with reference to the accompanying drawings, in which reference numerals denote like elements, and wherein:
[013] Figure 1a illustrates an environment for the implementation of a system for estimating remaining useful life (RUL) of a machinery component based on surface-defect progression, in accordance with one or more embodiments of the disclosure;
[014] Figure 1b illustrates the system for estimating the RUL, in accordance with one or more embodiments of the disclosure;
[015] Figure 2 illustrates a block diagram of the AI model, in accordance with one or more embodiments of the disclosure; and
[016] Figure 3 illustrates a process flow of a method for estimating the RUL of the machinery component based on the surface-defect progression, in accordance with one or more embodiments of the disclosure.
[017] Further, skilled artisans will appreciate that elements in the drawings are illustrated for simplicity and may not have necessarily been drawn to scale. For example, the flow charts illustrate the method in terms of the most prominent steps involved to help improve understanding of aspects of the present invention. Furthermore, in terms of the construction of the device, one or more components of the device may have been represented in the drawings by conventional symbols, and the drawings may show only those specific details that are pertinent to understanding the embodiments of the present invention so as not to obscure the drawings with details that will be readily apparent to those of ordinary skill in the art having the benefit of the description herein.

DETAILED DESCRIPTION
[018] For the purpose of promoting an understanding of the principles of the invention, reference will now be made to the various embodiments, and specific language will be used to describe the same. It will nevertheless be understood that no limitation of the scope of the invention is thereby intended, such alterations and further modifications in the illustrated system, and such further applications of the principles of the invention as illustrated therein, being contemplated as would normally occur to one skilled in the art to which the invention relates.
[019] It will be understood by those skilled in the art that the foregoing general description and the following detailed description are explanatory of the invention and are not intended to be restrictive thereof.
[020] Reference throughout this specification to “an aspect,” “another aspect,” or similar language means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrase “in an embodiment”, “in another embodiment”, and similar language throughout this specification may, but do not necessarily, all refer to the same embodiment.
[021] The terms “comprises”, “comprising”, or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of steps does not include only those steps but may include other steps not expressly listed or inherent to such process or method. Similarly, one or more devices or sub-systems or elements or structures or components preceded by “comprises... a” does not, without more constraints, preclude the existence of other devices or other sub-systems or other elements or other structures or other components or additional devices or additional sub-systems or additional elements or additional structures or additional components.
[022] The present disclosure relates to techniques for estimating remaining useful life (RUL) based on direct geometric measurement of surface defect progression and advanced degradation modelling. In particular, the present disclosure provides a method and a system for estimating the RUL of a machinery component based on surface-defect progression. The disclosed method and the system integrate a three-dimensional surface inspection, digital model comparison, and data-driven and physics-informed life estimation techniques, estimating the RUL. The disclosed method and the system enable objective, repeatable, and condition-based life assessment of machinery components across both civilian industrial and defense applications, thereby improving safety, reliability, and lifecycle management.
[023] Figure 1a illustrates an environment 100 for the implementation of a system 130 for estimating the remaining useful life (RUL) of a machinery component 120 based on surface-defect progression, in accordance with one or more embodiments of the disclosure.
[024] In an exemplary embodiment, the environment 100 may include an industrial system 110 where an assessment and estimation of RUL of the machinery component 120 subjected to wear, fatigue, erosion, and surface degradation is required. In a non-limiting example, the industrial system 110 may include industrial machinery maintenance industry, manufacturing industry, heavy engineering industry, transportation industry, energy systems, defense manufacturing industry, ordnance production industry, aerospace industry, and the like.
[025] In an exemplary embodiment, the industrial system 110 may be, but is not limited to, a power generation system. Further, the machinery component 120 may be a rotating component, such as a turbine blade used in the power generation system. The turbine blade may be subjected to continuous high-speed rotation, cyclic mechanical loading, and elevated thermal stresses during operation, resulting in progressive surface wear, erosion, and localized deformation. The system 130 may enable periodic high-resolution surface scanning of such rotating components to generate an accurate three-dimensional digital model representing their in-service condition. The system 130 may derive quantitative degradation metrics by comparing the three-dimensional digital model with a corresponding original equipment manufacturer (OEM) reference model. The quantitative degradation metrics may reflect the surface-defect progression, which may be subsequently utilized for estimating the RUL of the turbine blade.
[026] The quantitative degradation metrics, including material loss, average wear depth, and localized deformation patterns, may be extracted and stored as part of the historical degradation profile of the machinery component 120. The localized deformation may represent geometric changes in the machinery component 120 that may not involve material removal but rather displacement or distortion of the component geometry. The localized deformation may include bending, warping, bulging, or other geometric distortions. Using the multi-model RUL estimation framework, a degradation trend-based model with Bayesian inference may be applied to analyze the progression of wear and predict the future degradation path of the turbine blades. The analysis may result in an RUL estimate with associated confidence bounds, enabling proactive, condition-based maintenance planning. In an advantageous aspect, the system 130 may minimize risks of catastrophic failure by identifying wear trends early and scheduling timely maintenance or replacement. Further, the system 130 may enhance operational reliability and significantly extend the turbine blade’s overall service life. The detailed functioning of the system 130 is explained in the forthcoming paragraphs.
[027] Figure 1b illustrates the system 130 for estimating the RUL, in accordance with one or more embodiments of the disclosure.
[028] In an embodiment, the system 130 may be implemented within supervisory control and industrial automation environments such as supervisory control and data acquisition (SCADA) systems, distributed control systems (DCS), industrial internet of things (IIoT)-based systems, condition monitoring systems (CMS), manufacturing execution systems (MES), computerized maintenance management systems (CMMS), and digital‑twin or product lifecycle management (PLM) systems. The system 130 may acquire operational data using sensors, programmable logic controllers (PLCs), or edge devices. The system 130 may be integrated with these supervisory control and industrial automation environments by ingesting surface geometry data captured through periodic or event-driven scanning operations. Further, the system 130 may extract quantitative degradation metrics using deviation analysis and transmit the RUL estimates to the supervisory or maintenance-planning layer. The integration may enable closed-loop, condition-based maintenance workflows by combining geometric degradation intelligence with traditional sensor-based monitoring. In an embodiment, the system 130 may include processing unit 132, a memory 134, a data unit 136, a scanning apparatus 138, a plurality of modules 140, and an artificial intelligence (AI) model 142.
[029] In an embodiment, the processing unit 132 may be in communication with the memory 134. The processing unit 132 may be configured to execute instructions stored in the memory 134 and to perform various processes for estimating the RUL, as discussed throughout the disclosure. The processing unit 132 may be configured to initiate or stop one or more routines or a process based on the estimation of the RUL. The processing unit 132 may be a single processing unit or several units, all of which could include multiple computing units. The processing unit 132 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors (DSPs), central processing units (CPUs), an application processor (AP), or like, a graphics-only processing unit such as a graphics processing unit (GPU), a visual processing unit (VPU), and/or an Artificial intelligence (AI) dedicated processor such as a neural processing unit (NPU), state machines, logic circuitries, and/or any devices that manipulate signals based on operational instructions. Among other capabilities, the processing unit 132 may be configured to fetch and execute computer-readable instructions and data stored in the memory 134.
[030] In an embodiment, the memory 134 stores instructions to be executed by the processing unit 132 for voice-based navigation, as discussed throughout the disclosure. The memory 134 may include non-volatile storage elements. Examples of such non-volatile storage elements may include magnetic hard discs, optical discs, floppy discs, flash memories, or forms of electrically programmable memories (EPROM) or electrically erasable and programmable (EEPROM) memories. In addition, the memory 134 may, in some examples, be considered a non-transitory storage medium. The term “non-transitory” may indicate that the storage medium is not embodied in a carrier wave or a propagated signal. However, the term “non-transitory” should not be interpreted as the memory 134 is non-movable. In some examples, the memory 134 can be configured to store larger amounts of information. In certain examples, a non-transitory storage medium may store data that can, over time, change (e.g., in Random Access Memory (RAM) or cache). The memory 134 may be an internal storage unit, or an external storage unit of the system 130, a cloud storage, or any other type of external storage. In an embodiment, the memory 134 may be configured to store a reference digital model of the machinery component 120. The reference digital model may correspond to a reference OEM digital model of a new, unused machinery component 120. The reference digital model serves as a baseline against which the current condition of the machinery component 120 may be compared to determine the extent of degradation. The memory 134 may further store structured historical data comprising quantitative degradation metrics collected over multiple inspection cycles. In a non-limiting example, the structured historical data may be organized by component identifier and the inspection cycle, enabling retrieval and analysis of degradation progression over time. The structured historical data may include records of the quantitative degradation metrics extracted during each inspection cycle, along with associated metadata such as inspection date, operating conditions, and maintenance actions performed. The structured historical data may form the basis for degradation trend analysis and the RUL.
[031] The metrics may be spatially mapped and stored as structured historical data for each component and inspection cycle. The structured historical data may enable tracking of degradation progression over time, which may be essential for accurate the RUL estimation.
[032] In an embodiment, the data unit 136, amongst other things, may include routines, programs, objects, components, data structures, and the like, which may help in performing particular tasks or implementing data types. The data unit 136 may be implemented as signal processor(s), state machine(s), logic circuitries, or other devices/components that manipulate signals based on operational instructions. Further, the data unit 136 may be implemented in hardware, instructions executed by the processing unit 132, or by a combination thereof.
[033] In an embodiment, the scanning apparatus 138 may be configured to capture surface geometry data of the machinery component 120. The scanning apparatus 138 may comprise a laser scanner configured to capture surface geometry data of the machinery component 120 from multiple angles. The laser scanner emits laser beams that reflect off the surface of the machinery component 120, and the reflected signals are captured to determine precise three-dimensional coordinates of points on the component surface. The laser scanner is configured to capture surface geometry data from multiple angles to ensure complete coverage along the surface of the machinery component 120, including regions that may not be visible from a single scanning position. The acquired scan data may be processed to generate the three-dimensional digital model of the used component, accurately representing the current physical condition of the machinery component 120. In an implementation, the scanning apparatus 138 may be implemented in fixed installations within inspection facilities, or may be integrated into mobile inspection or maintenance units for field deployment. The scanning apparatus 138 may be operatively coupled to the processing unit 132 to transmit the captured surface geometry data for subsequent processing.
[034] In an embodiment, the processing unit 132 may be operatively coupled to the scanning apparatus 138 and the memory 134. The processing unit 132 may be configured to generate the three-dimensional digital model from the captured surface geometry data.
[035] In an embodiment, the processing unit 132 may be configured to match the three-dimensional digital model with the reference digital model of the machinery component 120.
[036] In an embodiment, the processing unit 132 may be configured to extract a set of quantitative degradation metrics to identify the surface-defect progression by performing deviation analysis between the three-dimensional digital model and the reference digital model. The set of quantitative degradation metrics may include a first wear, a second wear, a material loss distribution, and a localized deformation. The first wear may correspond to a maximum wear, and the second wear may correspond to an average wear. For example, the first wear may be represented by a scenario in which the turbine blade, originally manufactured with a surface thickness of 6.0 mm at a critical section, is found, during a periodic inspection, to have reduced to 5.4 mm due to long-term operational exposure. The total cumulative wear may correspond to a material loss of 0.6 mm, which may reflect the aggregated degradation that may have occurred over the entire service life of the turbine blade. In another example, the second wear may be sequential measurements captured over multiple inspection intervals. For instance, if the same turbine blade exhibits wear of 0.15 mm after 500 operating hours, 0.32 mm after 1000 hours, and 0.50 mm after 1500 hours, the progression of these values may represent the wear sequence over time. The progression may reflect that the wear may be accumulated across successive operating cycles, contributing incrementally to the overall cumulative wear of the turbine blade. In another example, the material loss distribution may be illustrated by a cylindrical machinery component 120 whose 3D deviation analysis reveals non‑uniform erosion across the surface. The scanned results may show that the upper region has experienced 0.10 mm of loss, the middle region 0.48 mm, and the lower region only 0.05 mm. Such a non‑uniform pattern may form a spatial distribution map indicating that the middle region is subjected to significantly higher wear intensity relative to the rest of the turbine blade. In yet another example, localized deformation may be observed in a gearbox housing where repeated cyclic loading results in a small but measurable geometric distortion. Deviation analysis may reveal an outward bulge of approximately 0.25 mm concentrated within a 10 mm × 15 mm area on one side of the housing. The deformation may reflect a localized structural displacement rather than uniform material wear and may be indicative of stress concentration or plastic deformation in that specific region.
[037] In an embodiment, the deviation analysis may involve computing the distance between corresponding points on the three-dimensional digital model and the reference digital model. Positive deviations may indicate material loss, while negative deviations may indicate material buildup or deformation. The processing unit 132 may generate a deviation map that may spatially represent the magnitude and direction of deviations across the entire component surface.
[038] In an embodiment, the processing unit 132 may be configured to estimate the RUL of the machinery component 120 based on the set of quantitative degradation metrics. In an implementation, the processing unit may be configured to estimate the RUL of the machinery component 120 using the AI model 142. The AI model 142 may be configured to estimate a degradation trend using the set of quantitative degradation metrics stored as structured historical data. Further, the AI model 142 may be configured to generate a probabilistic degradation distribution by applying Bayesian inference on the estimated degradation trend to quantify uncertainty and variability. Additionally, the AI model 142 may be configured to determine the RUL with an upper confidence bound and a lower confidence bound based on a correlation between the probabilistic degradation distribution and a predefined safety threshold. In an implementation, the processing unit 132 may be configured to estimate the RUL using the AI model 142, which may be trained on learned temporal degradation patterns. The AI model 142 may be selected based on historical failure data, maintenance history, and operational context associated with the machinery component 120.
[039] The processing unit 132 may control the processing of the surface geometry data of the machinery component 120 in accordance with a predefined operating rule or the AI model 142 stored in the non-volatile memory and the volatile memory. The predefined operating rule or AI model may be provided through training or learning. Here, being provided through learning means that, by applying learning techniques to learning data, a predefined operating rule or the AI model 142 of the desired characteristic may be prepared. The learning may be performed in a device itself, in which the AI model 142 may be implemented through a separate server/system. The learning techniques may be a method for training a predetermined target device (for example, a robot) using a plurality of learning data to cause, allow, or control the target device to decide or predict. Examples of learning techniques include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning. The AI model 142 may consist of a plurality of neural network layers. Each layer has a plurality of weight values and performs a layer operation through a calculation of a previous layer and an operation of a plurality of weights.
[040] In an embodiment, the AI model 142 may continuously improve as new inspection and operational data become available. The AI model 142 may be effective in capturing nonlinear, irregular, or accelerated degradation behaviors that cannot be accurately represented by traditional analytical or rule-based models. The AI model 142 may be trained on historical data from multiple machinery components and may be updated as new data becomes available. The AI model 142 may be adapted to changing operating conditions and to improve prediction accuracy over time.
[041] In an embodiment, the plurality of modules 140 or the like may be physically implemented by the processing unit 132, analog or digital circuits such as logic gates, integrated circuits, microprocessors, microcontrollers, memory circuits, passive electronic components, active electronic components, optical components, hardwired circuits, and may optionally be driven by firmware and software. The plurality of modules 140 may include a generation module 140a, a matching module 140b, an extraction module 140c, and an estimation module 140d. The generation module 140a may be configured to generate the three-dimensional digital model from the captured surface geometry data. The three-dimensional digital model may accurately represent the current physical condition of the machinery component 120. The matching module 140b may be configured to precisely align the three-dimensional digital model and compare the aligned three-dimensional digital model with the reference digital model corresponding to the new and unused machinery component 120. The alignment may enable accurate deviation analysis between the current state and the original state of the machinery component 120. The extraction module 140c may be configured to identify surface-to-surface deviation and perform profile deformation analysis. The system 130 may extract the set of quantitative degradation metrics, including the average wear, maximum wear, the material loss distribution, and the localized deformation across the component surface. The set of quantitative degradation metrics may be spatially mapped and stored as structured historical data for the machinery component 120 and the inspection cycle.
[042] The estimation module 140d may be configured to estimate the RUL of the machinery component 120 based on the set of quantitative degradation metrics using a selectable RUL estimation model. The stored degradation data may be processed using a configurable multi-model RUL estimation framework. The configurable multi-model RUL estimation framework may include various complementary analytical approaches. The detailed functioning of the complementary analytical approaches of the AI model 142 is explained in the forthcoming paragraphs.
[043] Figure 2 illustrates a block diagram of the AI model 142, in accordance with one or more embodiments of the disclosure.
[044] In an embodiment, the AI model 142 may include, but are not limited to, a degradation trend-based RUL model 202, an AI-based degradation prediction model 204, and a hybrid physics-based model 206.
[045] In an embodiment, the degradation trend-based RUL model 202 may incorporate the Bayesian inference to quantify uncertainty and variability. The AI model-based degradation prediction model may use deep learning models to capture nonlinear and time-dependent wear behavior. The hybrid life estimation model may combine physics-based fatigue analysis with machine-learning-derived correction factors. The degradation trend-based RUL model 202, an AI-based degradation prediction model 204, and a hybrid physics-based model 206 may be applied individually or selectively based on component geometry, data availability, and operational complexity. In an advantageous aspect, the use of complementary analytical approaches may provide a flexible, adaptive, and risk-aware mechanism for estimating the RUL of machinery component 120 under diverse operating conditions.
[046] In an embodiment, the degradation trend-based RUL model 202 may estimate the RUL of the machinery component 120 by analyzing the evolution of the wear over time relative to a defined safety threshold. The degradation trend-based RUL model 202 may rely on historical wear data collected across multiple inspection cycles of the same machinery component 120.
[047] In an embodiment, the processing unit 132 may continuously track historical degradation indicators such as the average wear and the maximum wear, and the like, and fit a mathematical degradation trend using models such as, but not limited to, linear functions, polynomial functions, or exponential functions, depending on the observed wear behavior. Such models used to identify the trend may represent the expected progression of degradation under normal operating conditions. Further, the degradation trend-based RUL model 202 may incorporate the Bayesian inference to account for measurement noise, inspection variability, and uncertainty in wear progression. Instead of producing a single deterministic degradation curve, the Bayesian inference may generate a probabilistic degradation distribution, reflecting confidence levels in the prediction. In a non-limiting example, the RUL may be estimated by determining the intersection point between the probabilistic degradation curve and the predefined safety threshold. The output may include the expected RUL value as well as upper and lower confidence bounds, enabling risk-aware maintenance planning. The degradation trend-based RUL model 202 may be particularly well suited for the machinery component 120 that may exhibit consistent and gradually evolving wear patterns.
[048] In an embodiment, the AI-based degradation prediction model 204 may employ deep learning techniques to forecast future degradation behavior by learning from historical operational and inspection data. In one embodiment, the AI-based degradation prediction model 204 may utilize an attention-based long short-term memory (LSTM) neural network, which may be specifically designed to model temporal dependencies in sequential data. In a non-limiting example, the LSTM neural network may be designed to capture long-term dependencies in sequential data, making the LSTM neural network well-suited for modeling degradation progression over extended time periods. The AI-based degradation prediction model 204 may take a multivariate time series as input that may include parameters such as the number of operations. The number of operations may include rounds fired, the average wear, the maximum wear, the maintenance history, operations since last maintenance, relevant environmental factors, and the like. The AI-based degradation prediction model 204 may learn complex, nonlinear relationships between usage patterns and degradation progression by processing the time-series data. The AI-based degradation prediction model 204 may implement an attention mechanism. The attention mechanism may enable the AI-based degradation prediction model 204 to focus on critical time intervals or degradation events that may have a greater influence on future wear behavior. In a non-limiting example, the attention mechanism may assign weights to different time steps in the input sequence, allowing the AI-based degradation prediction model 204 to emphasize time intervals that may be most relevant for predicting future degradation. The AI-based degradation prediction model 204 may identify critical degradation events, such as sudden increases in wear rate or changes in operating conditions, that have a disproportionate influence on the RUL. Further, the AI-based degradation prediction model 204 may predict the future degradation trajectory of the component and estimate the RUL based on the learned temporal patterns. Furthermore, the AI-based degradation prediction model 204 may continuously improve as new inspection and operational data may be available. The AI-based degradation prediction model 204 may be particularly effective in capturing nonlinear, irregular, or accelerated degradation behaviors that may not be accurately represented by traditional analytical or rule-based models. In an advantageous aspect, the AI-based degradation prediction model 204 may be selected based on the historical failure data, the maintenance history, and the operational context associated with the machinery component 120.
[049] In an embodiment, the AI model 142 may include the hybrid physics-based model 206. The hybrid physics-based model 206 may be configured to convert geometric deformation data from the deviation analysis into strain values. Further, the hybrid physics-based model 206 may be configured to fatigue life using the strain values and predefined material strain/stress life curve associated with the material of the machinery component 120. Additionally, the hybrid physics-based model 206 may be configured to apply a machine learning-based correction factor to adjust life estimates using historical failure data, maintenance history, and operational context associated with the machinery component 120.
[050] In an embodiment, the hybrid physics-based model 206 may combine physics-based fatigue life estimation with machine learning-based correction factors to achieve higher prediction accuracy under real-world operating conditions. The hybrid physics-based model 206 may leverage both physical understanding of material behavior and data-driven adaptation.
[051] In an embodiment, in the hybrid physics-based model 206, geometric deformation data extracted from 3D profile comparison may be converted into strain values, which are then used in conjunction with material stress-life (S-N) curves to estimate fatigue life in terms of allowable cycles. The S-N curves may represent the relationship between applied stress or strain amplitude and the number of cycles to failure for a given material. The processing unit 132 may use the computed strain values to look up the corresponding fatigue life from the material S-N curves. Calculation of the fatigue life may account for material properties such as yield strength, ultimate tensile strength, and fatigue limit. The calculation of the fatigue life may further account for loading conditions such as mean stress, stress ratio, and loading frequency. Environmental factors such as temperature, humidity, and corrosive environments may also be considered for the calculation of the fatigue life. The physics-based calculation may account for material properties, loading conditions, environmental factors, and known failure mechanisms such as crack initiation and growth. While the hybrid physics-based model 206 may provide strong theoretical grounding, the hybrid physics-based model 206 may not fully capture real-world variability. To address such a limitation, the physics-based model 206 may apply AI-based correction factors that may adjust the physics-derived life estimates using historical failure data, maintenance history, and operational context. The hybrid physics-based model 206 may identify critical regions on the component surface that contribute disproportionately to life consumption. Additionally, the hybrid physics-based model 206 may construct a behavioral life map representing spatial variation in fatigue damage. Further, a spatially resolved life-consumption map may represent a distribution of fatigue damage across the surface of the machinery component 120. Regions with higher life consumption may be identified as critical regions that require particular attention during maintenance. The final output may be a refined RUL estimate that may combine the rigor of physical laws with the adaptability of machine learning, delivering greater reliability than standalone physics-based or data-driven models.
[052] In an embodiment, the processing unit 132 may perform the model selection. In a non-limiting example, the model selection may be performed by a user based on data availability, component geometry complexity, and required accuracy. Degradation trend modeling may be suited for linear wear behavior and simple geometries with limited data, while AI-based temporal prediction may provide higher accuracy for nonlinear degradation when sufficient historical data is available. The hybrid physics-based model 206 may utilize finite element analysis and material fatigue properties to deliver the highest accuracy for complex and critical components.
[053] Figure 3 illustrates a process flow of a method 300 for estimating the RUL of the machinery component 120 based on the surface-defect progression, in accordance with one or more embodiments of the disclosure.
[054] At step 302, the method 300 may include capturing the surface geometry data of the machinery component 120.
[055] At step 304, the method 300 may include generating the three-dimensional digital model from the captured surface geometry data.
[056] At step 306, the method 300 may include matching the three-dimensional digital model with a reference digital model of the machinery component 120.
[057] At step 308, the method 300 may include extracting a set of quantitative degradation metrics for identifying the surface-defect progression by performing deviation analysis between the matched three-dimensional digital model and the reference digital model. The set of quantitative degradation metrics may include the first wear, the second wear, the material loss distribution, and the localized deformation. The first wear corresponds to the total cumulative wear, and the second wear corresponds to a wear sequence over time, adding to the cumulative wear of the machinery component 120.
[058] At step 310, the method 300 may include estimating the RUL of the machinery component 120 based on the set of quantitative degradation metrics. In an implementation, the estimation of the RUL may include estimating the degradation trend using the set of quantitative degradation metrics stored as structured historical data. Further, the estimation may include generating the probabilistic degradation distribution by applying the Bayesian inference on the estimated degradation trend to quantify uncertainty and variability. Additionally, the estimation may include determining the RUL with the upper confidence bound and the lower confidence bound based on the correlation between the probabilistic degradation distribution and the predefined safety threshold.
[059] The RUL may be estimated using the AI model 142, which may be based on the learned temporal degradation patterns. The AI model 142 may be selected based on the historical failure data, the maintenance history, and the operational context associated with the machinery component 120.
[060] In an embodiment, the AI model 142 may further include the hybrid physics-based model 206. The hybrid physics-based model 206 may be configured for converting geometric deformation data from the deviation analysis into strain values. Further, the hybrid physics-based model 206 may be configured for determining fatigue life using the strain values and predefined material strain/stress life curve associated with the material of the machinery component 120. Additionally, the hybrid physics-based model 206 may be configured for applying a machine learning-based correction factor to adjust life estimates using historical failure data, maintenance history, and operational context associated with the machinery component 120.
[061] The present disclosure provides the following advantages:
[062] The present disclosure provides an objective, geometry-driven RUL estimation system 130 and method 300 based on direct measurement of surface degradation, eliminating subjectivity and limitations of sensor-only or schedule-based maintenance approaches. The disclosed system 130 and method 300 deliver higher prediction accuracy and reliability through digital twin-based surface comparison and spatial wear mapping, enabling precise identification of material loss and localized deformation. Further, the disclosed system 130 and method 300 introduce a flexible, user-selectable multi-model RUL framework that adapts the estimation approach based on data availability, component geometry complexity, and required accuracy. Additionally, the disclosed system 130 and method 300 enable risk-aware and uncertainty-quantified life prediction using the Bayesian inference and probabilistic modelling, providing confidence bounds to support safety-critical maintenance decisions. Furthermore, the disclosed system 130 and method 300 achieve optimized maintenance planning and asset utilization by integrating physics-based fatigue life estimation with AI-based correction and spatial life-consumption mapping. As a result, the disclosed system 130 and method 300 allow targeted maintenance, reduced downtime, and extended component life. Further, the disclosed system 130 and method 300 integrate techniques such as high-resolution geometric measurement, probabilistic degradation modeling, AI-based temporal learning, and physics-informed fatigue estimation. Such techniques enable the method 300 and the system 130 to perform accurate, repeatable, and safety-oriented prediction of component remaining life, supporting condition-based maintenance strategies, enhanced reliability, and optimized utilization of mechanical assets.
[063] While specific language has been used to describe the disclosure, any limitations arising on account of the same are not intended. As would be apparent to a person in the art, various working modifications may be made to the method to implement the inventive concept as taught herein. The drawings and the foregoing description give examples of embodiments. Those skilled in the art will appreciate that one or more of the described elements may well be combined into a single functional element. Alternatively, certain elements may be split into multiple functional elements. Elements from one embodiment may be added to another embodiment. For example, orders of processes described herein may be changed and are not limited to the manner described herein. , Claims:1. A system (130) for estimating remaining useful life of a machinery component (120 based on surface-defect progression, the system (130) comprising:
a scanning apparatus (138) configured to capture surface geometry data of the machinery component (120);
a memory (134) storing a reference digital model of the machinery component (120); and
a processing unit (132) in communication with the memory (134), the processing unit (132) being configured to:
generate a three-dimensional digital model from the captured surface geometry data;
match the three-dimensional digital model with the reference digital model of the machinery component (120);
extract a set of quantitative degradation metrics for identifying the surface-defect progression by performing deviation analysis between the three-dimensional digital model and the reference digital model; and
estimate the remaining useful life of the machinery component (120) based on the set of quantitative degradation metrics.

2. The system (130) as claimed in claim 1, wherein the set of quantitative degradation metrics comprise a first wear, a second wear, a material loss distribution, and a localized deformation, wherein the first wear corresponds to a total cumulative wear, and the second wear corresponds to a wear sequence over time adding to the cumulative wear of the component.

3. The system (130) as claimed in claim 1, wherein the processing unit is configured to estimate the remaining useful life of the machinery component (120) using an artificial intelligence (AI) model by:
estimating a degradation trend using the set of quantitative degradation metrics stored as structured historical data;
generating a probabilistic degradation distribution by applying Bayesian inference on the estimated degradation trend, and to quantify uncertainty and variability; and
determining the remaining useful life with an upper confidence bound and a lower confidence bound based on a correlation between the probabilistic degradation distribution and a predefined safety threshold.

4. The system (130) as claimed in claim 3, wherein the processing unit is configured to estimate the remaining useful life using an artificial intelligence (AI) model based on learned temporal degradation patterns, the AI model being selected based on historical failure data, maintenance history, and operational context associated with the machinery component (120).

5. The system (130) as claimed in claim 3, wherein the AI model comprises a hybrid physics-based model (206) configured to:
convert geometric deformation data from the deviation analysis into strain values;
determine fatigue life using the strain values and predefined material strain/stress life curve associated with material of the machinery component (120); and
apply a machine learning-based correction factor to adjust life estimates using historical failure data, maintenance history, and operational context associated with the machinery component (120).

6. A method (300) for estimating remaining useful life of a machinery component (120) based on surface-defect progression, the method (300) comprising:
capturing (302) surface geometry data of the machinery component (120);
generating (304) a three-dimensional digital model from the captured surface geometry data;
matching (306) the three-dimensional digital model with a reference digital model of the machinery component (120);
extracting (308) a set of quantitative degradation metrics for identifying the surface-defect progression by performing deviation analysis between the matched three-dimensional digital model and the reference digital model; and
estimating (310) remaining useful life of the machinery component (120) based on the set of quantitative degradation metrics.

7. The method (300) as claimed in claim 6, wherein the set of quantitative degradation metrics comprise a first wear, a second wear, a material loss distribution, and a localized deformation, wherein the first wear corresponds to a total cumulative wear, and the second wear corresponds to a wear sequence over time adding to the cumulative wear of the machinery component (120).

8. The method (300) as claimed in claim 6, wherein estimating (310) the remaining useful life, the estimating comprises:
estimating a degradation trend using the set of quantitative degradation metrics stored as structured historical data;
generating a probabilistic degradation distribution by applying Bayesian inference on the estimated degradation trend to quantify uncertainty and variability; and
determining the remaining useful life with an upper confidence bound and a lower confidence bound based on a correlation between the probabilistic degradation distribution and a predefined safety threshold.

9. The method (300) as claimed in claim 8, wherein estimating (310) the remaining useful life using an artificial intelligence (AI) model (142) based on learned temporal degradation patterns, the AI model (142) being selected based on historical failure data, maintenance history, and operational context associated with the machinery component (120).

10. The method (300) as claimed in claim 8, wherein the AI model (142) comprises a hybrid physics-based model (206) configured for:
converting geometric deformation data from the deviation analysis into strain values;
determining fatigue life using the strain values and predefined material strain/stress life curve associated with material of the machinery component (120); and
applying a machine learning-based correction factor to adjust life estimates using historical failure data, maintenance history, and operational context associated with the machinery component (120).