A control method and system for improving reliability of a new energy vehicle decelerator

Abstract

The application relates to the technical field of new energy vehicle control, and discloses a control method and system for improving the reliability of a new energy vehicle reducer, aiming to solve the defect that the existing reducer control strategy cannot effectively manage the historical load cumulative fatigue damage. The method and system build a cumulative damage digital model of key components of the reducer, prospectively evaluate future operation risks based on the model, and actively modulate the driving motor torque according to the reliability margin. Specifically, the method includes real-time multi-source data acquisition, load spectrum generation and cycle counting, multi-dimensional damage accumulation calculation, residual service life and reliability margin prediction, and active torque strategy generation and modulation modules. Through the above scheme, the service life of the reducer can be significantly prolonged, and the performance consistency and operation reliability of the reducer in the whole life cycle can be ensured.

Description

Technical Field

[0001] The present invention belongs to the field of new energy vehicle control technology, and specifically relates to a control method and system for improving the reliability of a new energy vehicle reducer. Background Art

[0002] With the global energy structure transformation and growing awareness of environmental protection, the new energy vehicle industry is experiencing unprecedented rapid development and has become the core driving force of technological change in the global automotive industry. In the overall architecture of new energy vehicles, the electric drive system is the key assembly for its power source and transmission. Its performance directly determines the power, economy, smoothness, and reliability of the entire vehicle. As a core component in the electric drive system, the reducer has the critical function of efficiently and smoothly converting the high-speed, low-torque output of the drive motor into the low-speed, high-torque required by the main drive axle. Therefore, the operational reliability and durability of the reducer assembly are not only related to the long-term stable operation of the vehicle, but also directly affect the user's driving experience and the safety performance of the entire vehicle, forming a vital link in the new energy vehicle technology system.

[0003] Under the existing technological framework, to ensure the operational stability and safety of reducer assemblies, the industry generally adopts passive protection control strategies based on real-time condition monitoring. Specifically, this control method typically relies on various sensors, such as temperature, speed, and torque sensors, deployed on the reducer or its associated components (such as the drive motor) to collect key reducer operating parameters in real time. The control system pre-sets a series of fixed safety thresholds, such as the maximum allowable operating temperature and maximum load torque. During vehicle operation, the onboard controller continuously monitors these real-time parameters. If any parameter exceeds the preset safety threshold, it immediately triggers the corresponding protection mechanism. These mechanisms typically include torque limiting, power reduction, and even, in extreme cases, power output cutoff of the drive motor, thereby preventing immediate and catastrophic damage to the reducer due to overheating, overload, and other obvious factors. This control strategy based on real-time monitoring and passive response effectively addresses the problem of acute failures caused by extreme operating conditions at a specific stage of technological development, providing the necessary guarantee for the basic safe operation of the reducer. Its design concept is straightforward and easy to implement.

[0004] However, with the continuous iteration and upgrading of new energy vehicle technology, particularly the increasingly stringent requirements for vehicle power performance, range, and driving experience (especially NVH (noise, vibration, and harshness)), the inherent limitations of the aforementioned passive protection strategy, centered around threshold triggering, have become increasingly apparent. This limitation lies not in its inability to respond to existing overload events, but rather in its inability to address the hidden performance degradation and reliability impairment caused by non-extreme, but long-term, cumulative operating loads. This is due to the fact that traditional control models are inherently "state memory-depleted" systems. They focus solely on whether instantaneous parameters exceed their limits, completely ignoring the significant time-accumulated effects of fatigue damage on the reducer's internal components (such as the gear train, bearings, and oil seals) as a mechanical system. In real-world driving scenarios, vehicles frequently experience a variety of complex operating conditions, such as acceleration, deceleration, coasting, and parking. These conditions generate numerous transient high-torque shocks, high-frequency torque fluctuations, and unsteady thermal cycles. Although the load or temperature rise generated by these operating conditions during a single occurrence may not necessarily trigger the preset protection threshold, their high frequency and long-term cumulative effects will cause continuous and irreversible damage to the tooth surface contact fatigue of key components inside the reducer, micro-damage to the bearing raceways, and shear stability of the lubricating oil.

[0005] Furthermore, in high-performance new energy vehicles, the drive motor can deliver peak torque within milliseconds. The cumulative destructive impact of these drastic dynamic load changes on the reducer gear train far exceeds the protective response provided by traditional control strategies based on slowly varying parameters such as temperature. Existing control methods are unable to effectively quantify, memorize, and evaluate the reducer's historical load sequence, making it impossible to identify the "sub-health" state of the reducer caused by cumulative damage. This can cause the reducer to operate for extended periods in a dangerous range where its reliability margin is continuously eroded. While no overt failures may occur in the short term, its NVH performance may be unnoticed, and even lead to premature failure beyond its expected lifespan under normal impact loads that do not reach the critical threshold. This creates a profound inherent contradiction between the "after-the-fact" nature of such control strategies and the "pre-emptive" nature of mechanical component damage accumulation.

[0006] In summary, existing technologies only address the "acute disease" protection problem of reducers, namely, how to prevent them from immediate damage under extreme operating conditions. However, there is a lack of effective technical means to manage their "chronic disease"—that is, how to use intelligent control to delay the cumulative fatigue damage caused by normal use and thereby maximize their reliability and performance consistency throughout their life cycle. Therefore, how to establish a control model that can account for the cumulative effects of historical loads and conduct a forward-looking assessment of future operating risks, thereby achieving a paradigm shift from "passive fault response" to "active reliability management," has become a key challenge and a technical problem that needs to be solved urgently for those skilled in the art. Summary of the Invention

[0007] The purpose of the present invention is to overcome the drawback of existing control strategies for new energy vehicle reducers, which only provide passive threshold protection but are unable to effectively manage the cumulative fatigue damage caused by historical loads. This method and system aims to improve the reliability of reducers for new energy vehicles. By constructing a digital model of cumulative damage to key reducer components and conducting a forward-looking assessment of future operational risks based on this model, the method and system achieve a control paradigm shift from passive fault response to active reliability management, aiming to significantly extend the service life of the reducer and ensure its performance consistency and operational reliability throughout its entire life cycle.

[0008] To achieve the aforementioned objectives, the present invention provides a control system for improving the reliability of new energy vehicle reducers. The system is installed within the vehicle control unit (VCU) or dedicated power domain controller (PDU) of a new energy vehicle and communicates with the drive motor controller (MCU) and the onboard sensor network. The control system includes a real-time multi-source data acquisition module, a load spectrum generation and cycle counting module, a multi-dimensional damage accumulation calculation module, a remaining service life and reliability margin prediction module, and an active torque strategy generation and modulation module.

[0009] The real-time multi-source data acquisition module synchronously collects multi-channel data streams related to the reducer's load and operating environment at a preset high sampling frequency. Specifically, the module communicates with the drive motor controller via a high-speed controller area network (FlexRay) or Ethernet bus to obtain real-time output torque and speed data from the drive motor. The sampling frequency of torque and speed data is no less than 1kHz, ensuring accurate capture of transient torque shocks and high-frequency fluctuations caused by driving maneuvers or road surface changes. Furthermore, the module connects to the vehicle's sensor network via a standard controller area network (CAN) bus to obtain the reducer's housing temperature measured by an external temperature sensor, the lubricating oil temperature measured by a temperature sensor within the reducer's lubricating oil sump, and the vehicle's real-time speed signal provided by the anti-lock braking system (ABS) or vehicle stability control (ESP). All collected data is accurately timestamped and transferred to a dedicated circular data buffer for subsequent access.

[0010] The load spectrum generation and cycle counting module, connected to the data output of the real-time multi-source data acquisition module, converts non-steady-state, time-domain torque-speed time series data into a structured load spectrum matrix capable of characterizing fatigue loading. To achieve this, the load spectrum generation and cycle counting module incorporates a stress cycle analysis core based on the Rainflow Counting algorithm. This stress cycle analysis core first preprocesses the received high-frequency torque time series, identifying all local maxima and minima using a peak-valley detection algorithm to form a torque peak-valley sequence. The analysis core then iteratively processes the torque peak-valley sequence using a standard four-point method, decomposing the complex, irregular load history into a series of closed stress hysteresis loops, each corresponding to a complete stress cycle. Each identified stress cycle is uniquely defined by two parameters: its stress amplitude (half the cycle amplitude) and mean stress (the mean value of the cycle). Finally, the load spectrum generation and cycle counting module divides all counted stress cycles into intervals based on preset stress amplitudes and mean stresses, and generates a two-dimensional load spectrum histogram matrix. The row indices of the two-dimensional load spectrum histogram matrix correspond to different stress amplitude intervals, and the column indices correspond to different mean stress intervals. The value of each cell in the matrix represents the cumulative number of stress cycles that fell within that specific amplitude and mean interval during vehicle operation.

[0011] The multi-dimensional damage accumulation calculation module, which is connected to the output end of the load spectrum generation and cycle counting module, is the core of the technical solution of the present invention. Its function is to build and update in real time a digital twin model that characterizes the health status of the reducer. The multi-dimensional damage accumulation calculation module does not use a single damage assessment indicator, but calculates the cumulative damage degree of multiple dimensions in parallel for different key components and potential failure modes inside the reducer. The multi-dimensional damage accumulation calculation module specifically includes a gear contact fatigue damage calculation unit, a bearing rolling contact fatigue damage calculation unit, and a lubricating oil performance degradation calculation unit.

[0012] Furthermore, the gear contact fatigue damage calculation unit incorporates a mathematical model of the Wöhler curve (SN) for the reducer's driving and driven gears. This SN curve describes the number of fatigue cycles a gear material can withstand under a specific stress amplitude. Its mathematical expression is defined as N*S^m=C, where N is the number of fatigue life cycles, S is the stress amplitude, and m and C are material constants determined by the gear material (e.g., 20CrMnTi alloy steel that has undergone carburizing and quenching) and the heat treatment process. The gear contact fatigue damage calculation unit receives the load spectrum matrix output by the load spectrum generation and cycle counting module. For each cell (i, j) in the matrix, the corresponding stress amplitude is S_i, the mean stress is S_m,j, and the cumulative number of cycles is n_ij. The gear contact fatigue damage calculation unit first corrects the fatigue limit using the mean stress S_m,j according to the Goodman or Gerber criterion, converting each cycle into an equivalent symmetrical cyclic stress S_eq,i. The SN curve formula is then used to calculate the allowable number of cycles at this equivalent stress level: N_ij = C / (S_eq,i)^m. Finally, the gear contact fatigue damage calculation unit calculates the gear system's current total cumulative damage, D_gear, by linearly superimposing the damage at all stress levels according to the Palmgren-Miner linear cumulative damage criterion. The calculation formula is: D_gear = Σ(n_ij / N_ij). The gear system's current total cumulative damage, D_gear, is a dimensionless value ranging from 0 (brand new) to 1 (theoretical failure).

[0013] Furthermore, the bearing rolling contact fatigue damage calculation unit performs life assessments on key support bearings (e.g., tapered roller bearings or deep groove ball bearings) on the reducer's input and output shafts. This unit incorporates a bearing life calculation model based on the ISO 281 standard. The bearing rolling contact fatigue damage calculation first calculates the radial and axial force cycles acting on the bearing based on the input torque cycle data, combined with the bearing's geometric parameters and the gear train's transmission parameters. The dynamic equivalent load P_ij for each cycle is then calculated based on the bearing type. The bearing rolling contact fatigue damage calculation unit also calculates the cumulative bearing damage, D_bearing, based on the Palmgren-Meiner criterion, combining the bearing's basic dynamic load rating, C, and the life exponent, p (p = 10 / 3 for roller bearings and p = 3 for ball bearings). The calculation formula is: D_bearing = Σn_ij * (P_ij / C)^p. The cumulative damage degree of the bearing, D_bearing, is also a dimensionless value ranging from 0 to 1, which represents the degree of fatigue damage of the bearing raceway or rolling element.

[0014] Furthermore, the lubricant degradation calculation unit is used to assess the performance degradation of the reducer's internal lubricant due to thermal oxidation and mechanical shear. This unit establishes a comprehensive lubricant health model. For thermal oxidation damage, the unit utilizes a model based on the Arrhenius equation, using real-time historical lubricant temperature data to calculate the cumulative impact of temperature on lubricant additive consumption and base oil oxidation. For mechanical shear damage, the unit uses shear stress and shear rate calculated from torque-speed data, combined with the lubricant's shear stability parameters, to assess the permanent viscosity loss caused by long-chain molecular breakage in the oil's viscosity index improver. The unit weights thermal oxidation damage and mechanical shear damage to output a lubricant health index (H_oil), ranging from 100% (new oil) to a preset replacement threshold (e.g., 30%).

[0015] The RLS and RMA prediction module's inputs are connected to the three outputs (D_gear, D_bearing, and H_oil) of the multi-dimensional damage accumulation calculation module. Its function goes beyond simply reporting the current damage status to provide a forward-looking forecast of the reducer's future reliability trends. The RLS and RMA prediction module first determines the systemic critical damage indicator (D_crit) = max(D_gear, D_bearing) and uses it as a proxy for the fatigue state of the entire vehicle. It then uses a Kalman filter algorithm to process the time series of the critical damage indicator (D_crit). The Kalman filter's state vector contains the current damage value (D_crit(t)) and the damage accumulation rate (dD_crit / dt). Through the filter's prediction and update steps, the model smooths noise and dynamically estimates the future damage accumulation rate based on recent driving behavior patterns. Based on this, the Remaining Service Life and Reliability Margin Prediction module calculates the predicted remaining service life (RUL) of the reducer using the formula: RUL = (1-D_crit(t)) / (dD_crit / dt), where the unit can be remaining kilometers or remaining operating hours. The module also defines and calculates a dynamic reliability margin, M_relia. This dynamic reliability margin is a nonlinear function of the current cumulative damage level, D_crit. For example, M_relia = (1-D_crit)^k, where the exponent k > 1, causes the margin to decrease slowly in the early stages of damage and accelerate in the later stages, thus providing more sensitive risk warnings.

[0016] The active torque strategy generation and modulation module is the final actuator of the control loop of the present invention. Its input is connected to the output of the remaining service life and reliability margin prediction module, and it also receives the driver's original torque request signal (converted by the accelerator pedal position sensor). Its function is to intelligently and smoothly intervene in the torque output of the drive motor based on the predicted reliability margin, thereby slowing the rate of damage accumulation without significantly affecting the driving experience. The active torque strategy generation and modulation module adopts a hierarchical torque modulation logic, which includes at least three operating ranges, divided by two preset reliability margin thresholds M_th1 and M_th2 (M_th1>M_th2).

[0017] In the first operating range, when the calculated reliability margin M_relia is greater than the first threshold M_th1, the retarder is in a healthy state. Within this range, the active torque strategy generation and modulation module does not modify the driver's torque request signal, and the motor controller delivers torque in full compliance with the driver's intent, ensuring optimal vehicle performance.

[0018] In the second operating range, when the reliability margin M_relia drops below the first threshold M_th1 but remains above the second threshold M_th2, it indicates that the gearbox has entered a period of potential acceleration wear. During this range, the active torque strategy generation and modulation module initiates the first level of active intervention, namely, torque rate limiting. Specifically, the active torque strategy generation and modulation module processes the driver's raw torque request signal through a digital low-pass filter. The filter's cutoff frequency (or time constant) is positively correlated with the reliability margin M_relia. When the margin is high, the filter's cutoff frequency is high, minimizing the smoothing effect on the torque request. As the margin decreases, the cutoff frequency decreases, smoothing the high-frequency, drastic changes in the torque request and effectively reducing transient shock loads on the gear train. This processing significantly reduces the rate of fatigue damage accumulation while barely affecting the vehicle's steady-state acceleration performance.

[0019] In the third operating range, when the reliability margin M_relia further drops below the second threshold M_th2, the reducer has entered a high-risk operating state and its remaining service life is approaching the warning limit. Within this range, the active torque strategy generation and modulation module, while continuing to implement torque rate limiting, initiates a second-level active intervention strategy: dynamic torque ceiling. The active torque strategy generation and modulation module calculates a dynamic, allowable maximum output torque, T_max_dyn, which is a monotonically increasing function of the reliability margin M_relia and is always less than or equal to the physical peak torque of the drive motor. The active torque strategy generation and modulation module compares the rate-limited torque request signal with the dynamic torque ceiling and takes the smaller value as the final torque command to the drive motor controller. This approach achieves flexible and gradual derating of vehicle dynamics, sacrificing some extreme performance to maximize driving safety and extend the service life of remaining components. At the same time, the active torque strategy generation and modulation module will send instructions to the vehicle instrument system or central information display system through the CAN bus, light up the maintenance reminder light and display information such as "transmission system requires maintenance", prompting the user to perform maintenance.

[0020] The present invention also provides a control method for improving the reliability of a new energy vehicle reducer. The method is implemented based on the aforementioned control system hardware and software architecture and includes the following steps:

[0021] Step S100: System initialization. After the vehicle is powered on, the system loads the accumulated damage data (D_gear, D_bearing) and the lubricant health index H_oil saved at the last time the engine was turned off from the non-volatile memory.

[0022] Step S200: Real-time data acquisition: During vehicle driving, the real-time multi-source data acquisition module continuously collects drive motor torque and speed data at a frequency of no less than 1kHz, and simultaneously obtains reducer oil temperature, housing temperature, and vehicle speed information.

[0023] Step S300: Real-time load spectrum construction. The load spectrum generation and cycle counting module continuously receives torque time series data, applies the rainflow counting method to perform real-time analysis, and dynamically updates the two-dimensional load spectrum histogram matrix.

[0024] Step S400: Multi-dimensional damage calculation and update. The multi-dimensional damage accumulation calculation module periodically (e.g., every 1 second) reads the latest load spectrum matrix and, based on the newly added stress cycle count, calls the gear contact fatigue damage calculation unit, bearing rolling contact fatigue damage calculation unit, and lubricant performance degradation calculation unit to update the cumulative damage values ​​D_gear, D_bearing, and the lubricant health index H_oil.

[0025] Step S500: Reliability status prediction. The remaining service life and reliability margin prediction module receives the updated damage data, estimates the current damage accumulation rate through the Kalman filter algorithm, and calculates the predicted remaining service life RUL and dynamic reliability margin M_relia based on the estimated damage accumulation rate.

[0026] Step S600: Active Torque Strategy Decision and Execution. The active torque strategy generation and modulation module compares the calculated reliability margin M_relia with preset thresholds M_th1 and M_th2 to determine the current operating range. Based on this determination, it selects a control strategy: no intervention, torque rate limit, or torque rate limit plus dynamic torque ceiling. The module then processes the driver's original torque request and generates a final torque command.

[0027] Step S700: Command issuance and closed-loop control. The system sends the generated final torque command via a high-speed bus to the drive motor controller, which accurately executes the command. The system continuously loops through steps S200 to S700, forming a complete closed-loop active control process based on cumulative damage and future risk prediction.

[0028] Step S800: Data Storage and Communication. When the vehicle is turned off, the system writes the final cumulative damage data (D_gear, D_bearing) and the lubricant health index H_oil to non-volatile memory. Simultaneously, the system uploads key health indicators to a cloud server via the vehicle's telematics unit (T-BOX) for fleet health management and early fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a functional module block diagram of the control system provided by the present invention;

[0030] Figure 2 It is a flow chart of the control method provided by the present invention;

[0031] Figure 3 It is a schematic diagram of the working principle of the load spectrum generation and cycle counting module in the present invention;

[0032] Figure 4 It is a structural diagram of the multi-dimensional damage accumulation calculation module in the present invention;

[0033] Figure 5 This is a schematic diagram of the working principle of the remaining service life and reliability margin prediction module in the present invention;

[0034] Figure 6 It is a hierarchical logic diagram of the active torque modulation strategy in the present invention. DETAILED DESCRIPTION

[0035] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the following detailed description of a control method and system for improving the reliability of a new energy vehicle reducer provided by the present invention is provided in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments herein are intended only to explain the present invention and do not constitute any form of limitation to the present invention.

[0036] Reference Figure 1, which illustrates the overall functional architecture of the control system of the present invention. The system can be physically integrated into the vehicle's vehicle control unit (VCU) or a dedicated power domain controller (PDU). For example, it can be deployed on a controller hardware platform based on a 32-bit multi-core microprocessor from the Infineon Aurix TC3xx series. This processor offers ample floating-point computing power and multiple high-speed communication interfaces, meeting the intensive data processing and real-time control requirements of the present invention. The system logically consists of a series of collaborative software modules, including a real-time multi-source data acquisition module, a load spectrum generation and cycle counting module, a multi-dimensional damage accumulation calculation module, a remaining service life and reliability margin prediction module, and an active torque strategy generation and modulation module. These modules exchange information via well-defined data interfaces, forming a complete closed-loop control system from state perception to decision execution.

[0037] Specifically, the real-time multi-source data acquisition module is the perception foundation of the entire system. Its core responsibility is to capture physical quantities directly related to the reducer's operating status with high fidelity. In one specific embodiment, the real-time multi-source data acquisition module exchanges deterministic data with the drive motor controller (MCU) via the vehicle's powertrain high-speed bus, such as the 10Mbit / s FlexRay bus or the 100Mbit / s in-vehicle Ethernet. Through this high-speed link, the real-time multi-source data acquisition module acquires the real-time output torque and speed of the drive motor at a fixed period of 1 millisecond (i.e., a sampling frequency of 1kHz). This high sampling rate is crucial because it ensures accurate capture of transient torque spikes lasting only tens of milliseconds, caused by sudden acceleration or deceleration by the driver or sudden road changes (such as driving over potholes or speed bumps). These spikes are key factors in the accumulation of fatigue damage. Simultaneously, the real-time multi-source data acquisition module connects to the vehicle's sensor network via a standard 500kbit / s Controller Area Network (CAN) bus. It subscribes to and parses specific CAN messages. For example, message ID 0x1F1 acquires the reducer housing temperature measured by an NTC thermistor mounted at specific points on the reducer housing (e.g., near the master and slave bearing seats); and message ID 0x1F2 acquires the real-time lubricant temperature measured by a temperature sensor immersed in the reducer's lubricant reservoir. Furthermore, the real-time multi-source data acquisition module extracts high-precision, sensor-fusion-processed real-time vehicle speed signals from messages from the anti-lock braking system (ABS) or vehicle stability control (ESP) (e.g., message ID 0x0B4). Before entering the processing pipeline, all collected heterogeneous data is assigned a microsecond-level timestamp generated by the microprocessor's internal high-precision clock to ensure absolute synchronization for subsequent time series analysis. The collected data stream is sequentially stored in a dedicated circular data buffer of 8192 frames. This buffer can cache approximately the last eight seconds of high-frequency data, providing a stable data source for the subsequent load spectrum analysis module.

[0038] Next, the load spectrum generation and cycle counting module performs in-depth processing on the data output by the real-time multi-source data acquisition module. Its goal is to transform the seemingly chaotic torque time series into structured data that can quantitatively describe the fatigue load distribution. Figure 3The core of the load spectrum generation and cycle counting module is an implementation of the Rainflow Counting Algorithm defined in the ASTM E1049-85 standard. First, the 1kHz torque time series signal extracted from the ring buffer undergoes a digital preprocessing stage. This stage includes a fourth-order Butterworth low-pass filter with a cutoff frequency set to 100Hz to remove high-frequency electrical noise introduced by the motor controller PWM chopping or sensor circuitry, while retaining all effective load fluctuations affecting the mechanical structure. The preprocessed signal then enters a peak-valley detector. This detector compares the slope changes between adjacent sampling points to identify all local maxima and minima in the time series, thereby simplifying the continuous waveform into a discrete alternating sequence of torque peaks and valleys. The load spectrum generation and cycle counting module then iteratively processes this peak-valley sequence using the classic four-point method. This algorithm conceptually simulates the process of rainwater flowing over the eaves of a pagoda, accurately decomposing the complex load history into a series of closed stress hysteresis loops. Each successfully identified loop represents a complete load cycle that causes fatigue damage to the material. Each load cycle is uniquely characterized by two key parameters: stress amplitude (S_a), defined as half the difference between the maximum and minimum values ​​in the cycle; and mean stress (S_m), defined as the arithmetic mean of the maximum and minimum values ​​in the cycle. Finally, the load spectrum generation and cycle counting module classifies and counts all counted load cycles according to preset classification criteria. In a preferred embodiment, the load spectrum generation and cycle counting module maintains a two-dimensional load spectrum histogram matrix. The two-dimensional load spectrum histogram matrix is ​​set to a dimension of 64 rows and 32 columns. The row index corresponds to 64 different stress amplitude intervals. For example, the maximum contact stress of the reducer gear from 0 to 1500 MPa is divided into 64 equal-width intervals, each with a width of approximately 23.4 MPa. The column index corresponds to 32 different mean stress intervals, for example, covering the range from -400 MPa to +400 MPa. Each time a new load cycle is counted, its stress amplitude S_a and mean stress S_m are calculated. Based on the interval in which these values ​​fall, the count value of the corresponding matrix cell (i, j) is incremented by one. This load spectrum matrix is ​​dynamically and continuously updated, accurately recording the distribution of all fatigue loads experienced by the reducer since the vehicle was started.

[0039] The multi-dimensional damage accumulation calculation module is the calculation core of the technical solution of the present invention. It receives the load spectrum matrix as input and, based on a precise physical model, performs a parallel evaluation of the health status of different key components inside the reducer. Figure 4,The multi-dimensional damage accumulation calculation module does not use a single macro damage ,index, but is subdivided into three independent calculation units, targeting different ,failure mechanisms and components.

[0040] Specifically, the gear contact fatigue damage calculation unit focuses on evaluating the cumulative damage of tooth surface contact fatigue (pitting) and tooth root bending fatigue on the reducer's driving and driven gears. The gear contact fatigue damage calculation unit's knowledge base contains detailed performance parameters of gear materials. For example, 20CrMnTi alloy steel, a common high-performance gear material, undergoes carburizing, quenching, and grinding processes. Its stress-life curve (SN) can be accurately described by a mathematical model. For contact fatigue, the SN curve follows the form N*S_H^m=C, where S_H is the Hertzian contact stress and N is the number of cycles allowed to reach the Hertzian contact stress level. In this embodiment, the material constant m, obtained by fitting material fatigue test data, is set to 9.8, and the constant C is set to 1.2e30 MPa^9.8. The gear contact fatigue damage calculation unit periodically (e.g., once per second) processes the load spectrum matrix updated by the load spectrum generation and cycle counting module. Each nonzero cell (i,j) in the matrix represents n_ij cycles with a stress amplitude S_a,i and a mean stress S_m,j. Because fatigue life is highly sensitive to mean stress, the gear contact fatigue damage calculation unit first applies the Goodman criterion to correct for the effects of mean stress. The Goodman criterion is mathematically expressed as S_a / S_e + S_m / S_u = 1 / n, where S_e is the fatigue limit of the material under symmetrical cycles (e.g., 600 MPa for 20CrMnTi steel), S_u is the ultimate tensile strength of the material (e.g., 1100 MPa), and n is the safety factor. This formula can be used to convert any stress cycle (S_a,i, S_m,j) into an equivalent symmetrical stress cycle S_eq,i with the same life. Subsequently, using the aforementioned SN curve formula, the total number of allowable cycles at this equivalent stress level S_eq,i is calculated as N_ij = C / (S_eq,i)^m. Finally, the gear contact fatigue damage calculation unit uses the widely accepted Palmgren-Miner linear cumulative damage criterion to linearly superimpose the damage incurred at all different stress levels to determine the current total cumulative damage of the gear system, D_gear. This is calculated using the formula: D_gear = Σ(n_ij / N_ij), summed across all cells in the load spectrum matrix. D_gear is a dimensionless value ranging from 0 (representing a brand new, undamaged state) to 1 (representing the theoretical fatigue failure point), and it constitutes a key dimension of the gear reducer's health.

[0041] In parallel, the Bearing Rolling Contact Fatigue Damage Calculation Unit assesses the lifespan of key support bearings on the reducer's input and output shafts. Taking the 32206 tapered roller bearing commonly used on the input shaft as an example, the Bearing Rolling Contact Fatigue Damage Calculation Unit's calculations adhere to the ISO 281:2007 standard. Its database stores key parameters for this bearing type, including its basic dynamic load rating C (e.g., 55.5 kN, according to the manufacturer's manual) and geometric parameters such as the contact angle. Upon receiving torque cycle data (derived from the load spectrum matrix, but back-calculated to the torque on the bearing shaft using the gear ratio), the Bearing Rolling Contact Fatigue Damage Calculation Unit first decomposes the tangential and radial forces acting on the gears into the radial load Fr and axial load Fa acting on the bearings based on the reducer's gear geometry (e.g., the helix angle and pressure angle of the helical gears). It then calculates the dynamic equivalent load P_ij corresponding to each load cycle (i, j) using the ISO standard formula for tapered roller bearings. This calculation takes into account the coupled effects of radial and axial loads. Similar to gear damage calculation, the bearing rolling contact fatigue damage calculation unit also applies the Palmgren-Miner criterion to accumulate damage. For rolling bearings, the life formula is L10 = (C / P)^p, where p is the life exponent. For roller bearings, p is 10 / 3. Therefore, the cumulative bearing damage D_bearing is calculated as: D_bearing = Σn_ij * (P_ij / C)^p. This value is also a dimensionless number ranging from 0 to 1, which accurately quantifies the material fatigue damage caused by cyclic contact stress in the bearing raceways and rolling elements.

[0042] Furthermore, the lubricant degradation calculation unit assesses the health of the reducer's internal lubricant from both a chemical and physical perspective. It integrates a two-channel degradation model. The first is a model for thermal oxidative damage. This model is based on the Arrhenius equation and has the form: degradation rate = A*exp(-E_a / (R*T)), where T is the real-time absolute temperature of the lubricant, R is the ideal gas constant, and the activation energy E_a and pre-exponential factor A are parameters (e.g., E_a ≈ 90 kJ / mol) derived from aging tests at different temperatures on a specific lubricant (e.g., SAE 75W-90 fully synthetic gear oil with a certain PAO base oil). The thermal oxidative damage model continuously integrates over time to quantify the cumulative effects of antioxidant additive depletion and base oil oxidative deterioration due to high temperatures. The second model is a model for mechanical shear damage. High speeds and high torques create extreme shear stresses in the gear mesh and bearings, breaking the long-chain polymer molecules that contribute to the viscosity index, resulting in permanent viscosity loss. A mechanical shear damage model estimates shear stress and shear rate based on real-time torque-speed data and, combined with the lubricant's shear stability index (SSI), calculates the cumulative viscosity loss. Finally, the lubricant degradation calculation unit combines the results of thermal-oxidative damage and mechanical shear damage using a weighting function (e.g., a weighting factor determined based on historical data analysis or expert experience) to produce a comprehensive lubricant health index (H_oil). The oil health index is designed to decrease unidirectionally from 100% (representing new oil). When it falls below a preset replacement threshold (e.g., 30%), the system provides a clear maintenance recommendation.

[0043] Next, the remaining service life and reliability margin prediction module conducts comprehensive analysis and forward-looking prediction of the above multi-dimensional health status information. Its working principle is as follows: Figure 5As shown in Figure 2, the remaining service life and reliability margin prediction module first selects a key systemic damage indicator from three inputs (D_gear, D_bearing, H_oil). In this embodiment, D_crit = max(D_gear, D_bearing), representing the fatigue state of the entire reducer, is defined because failure of any gear or bearing will result in failure of the entire reducer. To robustly predict the future trend of D_crit, the remaining service life and reliability margin prediction module utilizes a Kalman filter algorithm. The Kalman filter's state vector x is defined as [D_crit(t), dD_crit / dt]^T, which includes the current damage value and its first-order derivative (damage accumulation rate). The Kalman filter's state transition model is x_k = A*x_{k-1} + w_{k-1}, where the state transition matrix A = [[1, Δt], [0, 1]] (Δt is the calculation period), and the process noise w represents the uncertainty in the damage rate caused by changes in driving behavior patterns. The observation model is z_k=H*x_k+v_k, where the observation matrix H=[1,0]. The observation value z_k is D_crit calculated by the multi-dimensional damage accumulation calculation module, and the observation noise v represents errors in the model itself and the calculation process. Through the standard Kalman filter iteration of prediction and update steps, the Kalman filter algorithm effectively smooths out short-term fluctuations in the calculated D_crit value. Based on the recent actual load history, it dynamically and adaptively estimates the most reliable damage accumulation rate dD_crit / dt. Based on this filtered and optimized rate, the Remaining Service Life and Reliability Margin Prediction module proactively calculates the predicted remaining useful life (RUL) of the reducer. The calculation formula is very intuitive: RUL=(1-D_crit(t)) / (dD_crit / dt). This RUL can be converted into remaining mileage based on vehicle speed history, providing users with valuable prognostic information. Furthermore, to enable more refined control, the Remaining Service Life and Reliability Margin Prediction module also defines and calculates a dynamic reliability margin M_relia. It is designed as a nonlinear function of the current cumulative damage level D_crit, for example, using the form M_relia = (1-D_crit)^k. In this embodiment, the exponent k is preferably 2. This quadratic relationship ensures that the reliability margin decreases very slowly in the early stages of damage (when D_crit is low), barely impacting vehicle performance. However, in the later stages of damage (when D_crit approaches 1), the reliability margin decreases more rapidly, providing a more sensitive and timely early warning signal for impending risks.

[0044] Finally, the active torque strategy generation and modulation module is the final executive body of the control logic of the present invention. Its core task is to convert the predictive analysis results of the upstream remaining service life and reliability margin prediction module into actual intervention on vehicle power output. Figure 6 As shown, the active torque strategy generation and modulation module receives the driver's raw torque request signal via the accelerator pedal and intelligently and smoothly modulates the torque request signal based on the real-time reliability margin M_relia calculated by the remaining service life and reliability margin prediction module. It internally implements a hierarchical torque modulation logic, divided by two pre-calibrated reliability margin thresholds, M_th1 and M_th2 (for example, M_th1 = 0.6, M_th2 = 0.25), defining three distinct operating ranges.

[0045] In the first operating range, when M_relia > M_th1, this indicates that the cumulative damage to the reduction gear is minimal and the reliability margin is sufficient. In this healthy state, the active torque strategy generation and modulation module does not take any intervention measures. The driver's original torque request signal is directly transmitted to the drive motor controller as the final torque command without any modification. This ensures that the vehicle can fully realize its designed extreme dynamic performance and meet customer expectations early in its life cycle or under mild driving conditions.

[0046] When the vehicle continuously operates under high load, causing the reliability margin M_relia to drop to the range of M_th2 < M_relia ≤ M_th1, the system enters the second operating interval, namely the potential accelerated wear period. At this time, the active torque strategy generation and modulation module activates the first-level active intervention strategy, namely the slew rate limiting. Specifically, the original torque request signal is processed through a first-order digital low-pass filter. The mathematical expression of this filter is y[n]=α*x[n]+(1-α)*y[n-1], where x[n] is the original torque request at the current moment, and y[n] is the filtered output torque command. The key is that the filter coefficient α is a function of the reliability margin M_relia, for example, α=α_min+(M_relia-M_th2) / (M_th1-M_th2)*(α_max-α_min). When M_relia approaches M_th1, α approaches its maximum value α_max (such as 0.95), the filtering effect is weak, and the response is rapid; when M_relia approaches M_th2, the value of α decreases, and the smoothing effect of the filter increases. This processing method can effectively "flatten" the high-frequency and violently changing components in the torque request signal, especially those transient impacts caused by the driver's subconscious "ankle jitter" or rapid pedal pressing, thus significantly reducing the impact load on the gear train and the cumulative rate of fatigue damage. Since it mainly affects the torque establishment rate rather than the steady-state value, in most driving scenarios, the driver can hardly perceive this intervention, achieving active protection without sacrificing the subjective driving experience.

[0047] When further damage accumulates to the reducer, causing the reliability margin M_relia to drop further to the dangerous range of M_relia ≤ M_th2, the system enters the third operating range. This indicates that the reducer has entered a high-risk state and its remaining service life may not be sufficient to last until the next regular maintenance period. Within this range, the active torque strategy generation and modulation module, while continuing to implement the aforementioned torque rate limit, initiates a more aggressive second-level active intervention strategy, namely, dynamic torque ceiling. The active torque strategy generation and modulation module calculates a dynamic, maximum allowable output torque, T_max_dyn. The maximum output torque is a monotonically increasing function of the reliability margin M_relia, for example, T_max_dyn = T_min_safe + (T_peak - T_min_safe) * (M_relia / M_th2)^γ, where T_peak is the motor's physical peak torque and T_min_safe is the minimum torque required to ensure basic driving safety. The shaping exponent γ (e.g., 0.5) is used to adjust the derating curve. The active torque strategy generation and modulation module compares the rate-limited torque request signal with the dynamic torque ceiling T_max_dyn and takes the smaller value as the final torque command. This achieves flexible and gradual derating of vehicle power performance. As M_relia decreases, the torque ceiling also decreases smoothly, sacrificing some extreme acceleration performance to maximize driving safety and significantly extend the service life of remaining components, providing customers with a sufficient maintenance window. Simultaneously, the active torque strategy generation and modulation module sends a command to the instrument cluster controller and central infotainment system via the CAN bus, illuminating the transmission maintenance reminder light (e.g., a yellow gear icon) and displaying a clear text message such as "Transmission system performance is limited. Please contact a service center as soon as possible" to ensure the user is fully informed.

[0048] Reference Figure 2 The present invention also provides a control method for improving the reliability of a new energy vehicle reducer. The method is run on the above-mentioned system hardware and software architecture, and its specific execution process includes the following steps:

[0049] Step S100: System Initialization. When the vehicle is powered on (IG-ON), the control system awakens. It first accesses the controller's internal non-volatile memory (e.g., EEPROM or Flash) to read and load the final cumulative damage data saved before the vehicle was last powered off (IG-OFF). This includes the gear cumulative damage degree D_gear, the bearing cumulative damage degree D_bearing, and the lubricant health index H_oil. This ensures continuous damage assessment and a complete record of the load history throughout the vehicle's lifecycle.

[0050] Step S200: Real-time data acquisition. After the vehicle enters the drivable state, the real-time multi-source data acquisition module begins to continuously acquire torque and speed data from the drive motor controller at a high frequency of 1kHz. It also simultaneously parses information such as reducer oil temperature, case temperature, and vehicle speed from the CAN bus, providing real-time, high-fidelity input for subsequent calculations.

[0051] Step S300: Real-time load spectrum construction. The load spectrum generation and cycle counting module continuously processes the torque time series data from step S200. It applies the rainflow counting method for online analysis in real time, decomposing the irregular load flow into independent fatigue cycles and dynamically updating the internally stored two-dimensional load spectrum histogram matrix.

[0052] Step S400: Multi-dimensional damage calculation and update. The multi-dimensional damage accumulation calculation module executes periodically at a low frequency, for example, every one second. It reads the incremental portion of the load spectrum matrix since the last calculation, representing the number of newly generated fatigue cycles. It then calls the gear contact fatigue damage calculation unit, the bearing rolling contact fatigue damage calculation unit, and the lubricant performance degradation calculation unit, respectively, to update the current values ​​of D_gear, D_bearing, and H_oil based on their respective physical models and the newly added load data.

[0053] Step S500: Reliability Status Prediction. The Remaining Useful Life and Reliability Margin Prediction module receives the damage data updated in step S400. It processes D_crit = max(D_gear, D_bearing) as input and uses its internal Kalman filter algorithm to calculate the smoothed current damage accumulation rate. Based on this rate, the Remaining Useful Life and Reliability Margin Prediction module calculates the predicted Remaining Useful Life (RUL) and the dynamic reliability margin (M_relia).

[0054] Step S600: Active Torque Strategy Decision and Execution. The active torque strategy generation and modulation module obtains the reliability margin M_relia calculated in step S500 and compares it with preset thresholds M_th1 and M_th2. Based on the comparison results, the active torque strategy generation and modulation module determines the current operating range (healthy, potential wear, or high risk) and selects the appropriate torque modulation strategy: no intervention, torque rate limiting only, or a combination of torque rate limiting and dynamic torque ceiling. It then applies this strategy to the driver's original torque request to generate a final, intelligently modulated torque command.

[0055] Step S700: Command issuance and closed-loop control. The system sends the final torque command generated in step S600 to the drive motor controller via a high-speed bus such as FlexRay or in-vehicle Ethernet. As a faithful executor, the drive motor controller precisely controls the inverter to generate the corresponding current, driving the motor to output the torque required by the current command. The system then seamlessly returns to step S200 and continuously loops through steps S200 to S700, forming a complete, dynamic, closed-loop, proactive reliability management control process based on historical cumulative damage and future risk prediction.

[0056] Step S800: Data Storage and Communication. When the vehicle detects a stall signal, the system executes the final storage procedure before powering off. It securely writes the current final cumulative damage data (D_gear, D_bearing) and lubricant health index H_oil to the controller's non-volatile memory for recall upon the next startup. As a preferred embodiment, the system also uploads these key health indicators, remaining useful life (RUL), and any generated warnings to a cloud-based backend server via the onboard telematics unit (T-BOX) via 4G / 5G networks. This data can be used by OEMs for large-scale fleet health monitoring, early fault warning, and provides valuable data support for optimizing future reducer designs.

[0057] Example 1

[0058] In order to verify the practical effect of the technical solution of the present invention, a test based on a hardware-in-the-loop (HIL) simulation platform was carried out.

[0059] Test subject: A pure electric SUV equipped with a single-stage reducer with a final reduction ratio of 9.03 and a drive motor with a peak torque of 320 Nm. The gears were made of 20CrMnTi carburized and hardened steel. The SN curve characteristics, bearing type (the input shaft uses a 32206 tapered roller bearing), and other parameters were all embedded in the control system model as previously described.

[0060] Test conditions: A VCU controller implementing the proposed control method was connected to a HIL simulation platform capable of accurately simulating vehicle dynamics and drive systems. A customized, 30-minute drive cycle with extensive rapid acceleration and deceleration was used to simulate a mix of user experiences in urban congestion and aggressive mountain driving.

[0061] Testing process: The initial damage state of the retarder was set to zero (D_gear = 0, D_bearing = 0). A custom driving cycle was repeated 1000 times, simulating approximately 500 hours of high-intensity operation. The control system of the present invention was continuously operated, with the reliability margin thresholds set to M_th1 = 0.6 and M_th2 = 0.25. During the test, the system recorded changes in D_gear, D_bearing, M_relia, and the activation of the torque modulation strategy in real time.

[0062] Comparative Example 1

[0063] The vehicle model and test conditions were identical to those in Example 1. The difference was that the VCU controller employed a traditional control strategy. This strategy only included a static torque limit based on the motor and battery status (for example, output torque must not exceed 320 Nm under any circumstances) but lacked any dynamic adjustment mechanism based on accumulated fatigue damage. The same custom drive cycle was repeated 1000 times.

[0064] Test results comparison

[0065] After completing 1000 cycles of testing, the key performance indicators of Example 1 and Comparative Example 1 were statistically analyzed and compared, and the results are shown in the following table:

[0066]

[0067] The data in the table above clearly demonstrates the significant superiority of the technical solution of the present invention. After experiencing identical external load inputs, the cumulative damage degree (D_crit) of the key components (gears and bearings) of the reducer using the control system of the present invention was only 0.28, significantly lower than the 0.45 achieved with the conventional strategy. This is directly attributed to the torque modulation module of the present invention, which proactively intervenes when the reliability margin decreases. By limiting the torque change rate and implementing a dynamic torque ceiling, it effectively mitigates the most damaging transient impact loads. While this intervention accounted for a significant portion of the total testing time (a total of 22.7%), its design minimized the impact on the driving experience. Crucially, the final predicted remaining useful life (RUL) comparison shows that the present solution extends the reducer's expected lifespan from approximately 611 hours under the conventional strategy to approximately 1286 hours, a lifespan extension rate exceeding 110%. This fully demonstrates that the present invention, by shifting the paradigm from passive protection to active health management, can significantly improve the reliability and durability of new energy vehicle reducers, providing users with lower lifecycle costs and greater safety.

[0068] In summary, the present invention constructs a multi-dimensional, physical model-based digital twin system to accurately quantify the cumulative fatigue damage of key components inside the reducer in real time, and based on this, makes forward-looking predictions of future risks. Furthermore, it seamlessly feeds back the prediction results to the vehicle's power control loop, and through an intelligent, hierarchical torque modulation strategy, it actively and effectively slows down the cumulative rate of damage without significantly affecting the driving experience. The system and method disclosed in the present invention successfully overcome the limitations of the passive and threshold protection strategies in the prior art, realize the active management and optimization of the health status of the transmission system throughout its life cycle, and has extremely high engineering application value and commercial prospects.

[0069] The above are only preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A control system for improving the reliability of a new energy vehicle reducer, characterized in that: include: A real-time multi-source data acquisition module is configured to acquire multi-channel data streams related to the reducer load and operating environment in real time, wherein the multi-channel data streams include at least the real-time output torque and real-time output speed of the drive motor; The load spectrum generation and cycle counting module is configured to convert the time domain series data composed of the real-time output torque into a structured load spectrum representing the fatigue load characteristics through a preset stress cycle analysis algorithm; A multi-dimensional damage accumulation calculation module is configured to concurrently calculate and update the cumulative damage degree of key components in multiple dimensions based on a structured load spectrum and in combination with a preset physical model of the key components inside the reducer; The multi-dimensional damage accumulation calculation module specifically includes: a gear contact fatigue damage calculation unit configured to calculate the cumulative contact fatigue damage degree of the driving and driven gears of the reducer; a bearing rolling contact fatigue damage calculation unit configured to calculate the cumulative rolling contact fatigue damage degree of key support bearings on the input shaft and output shaft of the reducer; and a lubricant performance degradation calculation unit configured to evaluate the health status of the lubricant inside the reducer and output a lubricant health index; The remaining service life and reliability margin prediction module is configured to dynamically estimate the future damage accumulation rate based on the updated cumulative damage degree, and accordingly predict the remaining service life of the reducer and calculate a dynamic reliability margin; the remaining service life and reliability margin prediction module is specifically configured as follows: Determine a systemic critical damage index, whose value is the maximum of the cumulative contact fatigue damage and the cumulative rolling contact fatigue damage; calculate the dynamic reliability margin as a nonlinear function of the critical damage index; and an active torque strategy generation and modulation module configured to modulate a raw torque request signal from a driver according to a dynamic reliability margin to generate a final torque command and send it to a drive motor controller; The active torque strategy generation and modulation module is internally preset with two reliability margin thresholds, namely a first threshold and a second threshold, and the first threshold is greater than the second threshold, for dividing at least two operating intervals; The active torque strategy generation and modulation module is specifically configured as follows: When the dynamic reliability margin is greater than a first threshold, the original torque request signal is not modified; When the dynamic reliability margin is lower than the first threshold but higher than the second threshold, a first-level active intervention strategy is initiated, which applies a torque change rate limit to the original torque request signal, that is, the original torque request signal is processed through a digital low-pass filter to smooth out the high-frequency variation portion of the torque request; When the dynamic reliability margin is lower than the second threshold, the second level active intervention strategy is initiated on the basis of continuing to implement the torque change rate limitation; The second-level active intervention strategy is to impose a dynamic torque ceiling. The specific steps include: calculating a dynamic, allowable maximum output torque that is a monotonically increasing function of the dynamic reliability margin; comparing the rate-limited torque request signal with the dynamic torque ceiling, and taking the smaller value as the final torque command; At the same time, the active torque strategy generation and modulation module is also configured to send instructions to the vehicle display system through the bus to prompt the user to perform maintenance.

2. The control system according to claim 1, characterized in that: The real-time multi-source data acquisition module is further configured to: Establish communication with the drive motor controller via a high-speed bus to obtain real-time output torque and real-time output speed data of the drive motor at a sampling frequency of not less than 1kHz; and accessing the vehicle-mounted sensor network via the controller area network bus to obtain the housing temperature of the reducer housing, the lubricating oil temperature of the reducer lubricating oil, and the real-time speed signal of the entire vehicle; Furthermore, the real-time multi-source data acquisition module is further configured to assign a time stamp to all acquired data and store the data in a circular data buffer.

3. The control system according to claim 1, characterized in that: The stress cycle analysis algorithm of the load spectrum generation and cycle counting module is the rain flow counting method; The load spectrum generation and cycle counting module is specifically configured as follows: First, the peak-valley value detection is performed on the received torque time-domain sequence to identify all local maximum and minimum points, thereby forming a torque peak-valley value sequence; Secondly, the four-point method is used to iteratively process the torque peak-valley value sequence, and the torque time domain sequence is decomposed into a series of closed stress cycles defined by stress amplitude and mean stress. Finally, all the counted stress cycles are counted according to the preset stress amplitude interval and average stress interval to generate and update a two-dimensional load spectrum histogram matrix, where the row index of the matrix corresponds to different stress amplitude intervals, the column index corresponds to different average stress intervals, and the value of the matrix cell is the cumulative number of stress cycles falling into the interval.

4. The control system according to claim 1, characterized in that: The gear contact fatigue damage calculation unit has a mathematical model of the SN curve of the gear material solidified inside. This model defines the relationship between stress amplitude and fatigue life cycles. The gear contact fatigue damage calculation unit is specifically configured as follows: For each stress cycle in the load spectrum histogram matrix, the stress cycle is first converted into an equivalent symmetrical cyclic stress according to the preset mean stress correction criterion; Then, the mathematical model of the SN curve is used to calculate the allowable number of cycles under the equivalent symmetrical cyclic stress level; Finally, according to the Palmgren-Miner linear cumulative damage criterion, the cumulative contact fatigue damage degree is calculated and updated by linearly superimposing the damage at all stress levels, that is, summing the ratio of the actual number of cycles to the allowable number of cycles.

5. The control system according to claim 1, characterized in that: The bearing rolling contact fatigue damage calculation unit has a bearing life calculation model defined in accordance with ISO 281 standard, and stores the basic dynamic load rating and life index of key support bearings; The bearing rolling contact fatigue damage calculation unit is specifically configured as follows: First, based on the input torque cycle data and combined with the bearing geometry parameters and gear train transmission parameters, the radial force and axial force cycles acting on the key support bearing are calculated, and the dynamic equivalent load of each cycle is calculated accordingly; Then, the cumulative rolling contact fatigue damage is calculated and updated by summing the damage of each cycle according to the Palmgren-Miner linear cumulative damage criterion, where the damage of a single cycle is determined by the ratio of the dynamic equivalent load to the basic dynamic load rating to the power of p.

6. The control system according to claim 1, characterized in that: The lubricant performance degradation calculation unit is specifically configured as follows: Using a model based on the Arrhenius equation, the cumulative damage to the lubricant caused by thermal oxidation is calculated based on real-time historical lubricant temperature data. Furthermore, the shear stress and shear rate calculated from the real-time output torque and speed data of the drive motor, combined with the shear stability parameters of the lubricant, are used to evaluate the permanent viscosity loss of the lubricant caused by mechanical shear. Finally, the thermal oxidation damage and mechanical shear damage are weightedly fused to output the lubricant health index.

7. The control system according to claim 1, characterized in that: The remaining service life and reliability margin prediction module is further configured as follows: First, a systematic key damage index is determined, whose value is the maximum value of the cumulative contact fatigue damage and the cumulative rolling contact fatigue damage. Secondly, a Kalman filter algorithm is used to process the time series of key damage indicators. The state vector of the Kalman filter contains the current damage value and the damage accumulation rate. Through the prediction step and update step of the filter, the future damage accumulation rate is dynamically estimated. Based on this estimated damage accumulation rate, the predicted remaining service life is calculated.

Images