A motor controller power module life estimation and adaptive health management and control method and system

Abstract

This application discloses a method and system for predicting the lifespan and adaptive health management of a motor controller power module, relating to the field of new energy vehicle technology. It involves real-time acquisition of electrical and thermal parameters of the power module, as well as the vehicle's overall motor operating parameters. Based on these parameters, the conduction and switching losses of the power module are calculated to obtain the instantaneous power loss sequence. This instantaneous power loss sequence is used as a thermal input, and the thermal parameters are used as thermal boundary conditions input to a third-order Foster electrothermal network model to dynamically calculate the transient thermal response of the power module. This application quantifies cumulative fatigue damage and remaining lifespan through transient junction temperature estimation, rainflow counting, and multivariate lifespan assessment. It also proactively reduces subsequent thermal cycle stress based on the linked motor control parameters according to the health status, while simultaneously performing state storage, fault warning, and maintenance prompts, thereby improving the full life-cycle health management capabilities of new energy vehicle power modules.

Description

Technical Field

[0001] This application relates to the field of new energy vehicle technology, specifically to a method and system for estimating the lifespan of a motor controller power module and for adaptive health management. Background Technology

[0002] The drive motor controller of a new energy vehicle is a key component of the vehicle's electric drive system. The power module is responsible for power conversion and high-power output, typically employing power semiconductor devices such as IGBTs or SiC MOSFETs, and is composed of a multi-layered heterogeneous material structure including chips, bonding wires, solder layers, and a DBC ceramic substrate. During actual driving, new energy vehicles frequently experience various operating conditions such as starting, acceleration, deceleration, high-speed cruising, heavy-load hill climbing, and idling. This causes continuous changes in the drive motor's output torque, speed, and power demand, subjecting the power module to random and periodic electrical load fluctuations, and further resulting in repeated rises and falls in the junction temperature of the power device chips.

[0003] Due to the differences in the coefficients of thermal expansion of different materials inside the power module, long-term temperature cycling will generate repeated thermomechanical stress in the bonding wires, solder layer and substrate connection area, which may lead to failure phenomena such as bonding wire peeling or breakage, solder layer voids or cracks, increased thermal resistance and deterioration of device conduction voltage drop.

[0004] To address the thermal reliability issues of power modules in new energy vehicles, existing technologies typically employ fixed temperature thresholds for over-temperature protection. This involves monitoring the power module housing temperature and implementing power reduction or shutdown protection when the temperature exceeds the set threshold. Some solutions utilize empirical lifespan models based on parameters such as average operating conditions and fixed thermal cycles to perform offline calculations of the power module's lifespan. Other solutions employ fixed thermal resistance models to statically estimate the chip junction temperature based on the module housing temperature and power loss, and then determine the thermal state of the power module accordingly.

[0005] In addition, the life assessment function in existing new energy vehicle motor controllers is usually relatively independent of functions such as motor vector control, PWM wave generation, switching frequency adjustment and torque control. The life assessment results are mainly used for condition monitoring or fault indication.

[0006] The shortcomings of existing technology: While the aforementioned technical solutions can achieve over-temperature protection or lifespan assessment of power modules in new energy vehicles to a certain extent, they still have significant limitations. First, estimating junction temperature solely based on the power module's case temperature or fixed thermal resistance is insufficient to accurately reflect the transient temperature changes of power device chips under dynamic conditions such as frequent acceleration and deceleration, heavy-load hill climbing, and high-speed cruising in new energy vehicles. Furthermore, it easily overlooks thermal cycling, which has a small temperature amplitude but occurs frequently, leading to an underestimation of actual thermal fatigue damage.

[0007] Secondly, existing simplified empirical life models typically consider fewer influencing factors and fail to fully couple the junction temperature variation, peak junction temperature, conduction duration, and electrical stress factors such as operating current and voltage. Therefore, they are difficult to adapt to the complex and randomly changing operating conditions of new energy vehicles, and the accuracy and applicability of life prediction results are limited.

[0008] Furthermore, most existing solutions are passive protection, meaning that power reduction or shutdown measures are only taken after the temperature reaches a preset threshold. The life assessment results lack effective closed-loop linkage with motor control parameters such as PWM switching frequency, torque output limit, and wave generation strategy. Therefore, they cannot proactively reduce subsequent thermal cycle stress and suppress the aging rate based on the actual aging degree of the power module.

[0009] In addition, existing technologies typically lack unified health status classification, remaining service life quantification, historical operation data storage, and vehicle network early warning and maintenance traceability mechanisms, making it difficult to provide effective technical support for preventive maintenance, fault early warning, and full life cycle health management of power modules in new energy vehicles.

[0010] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0011] The purpose of this application is to provide a method and system for estimating the lifespan of a motor controller power module and for adaptive health management, so as to solve the problems in the background art mentioned above.

[0012] To achieve the above objectives, this application provides the following technical solution: a method for estimating the lifespan and adaptive health management of a motor controller power module, comprising the following steps: The electrical and thermal parameters of the power module and the vehicle operating parameters of the motor are collected in real time. The conduction loss and switching loss of the power module are calculated based on the electrical parameters and the vehicle operating parameters of the motor, and the instantaneous power loss sequence of the power module is obtained. The transient thermal response of the power module is dynamically calculated by using the instantaneous power loss sequence as the thermal input and the thermal parameters as the thermal boundary conditions into the third-order Foster electrothermal network model, thereby obtaining the transient junction temperature timing of the power device chip. Rainflow counting is performed on the transient junction temperature time series, the thermal cycles are decomposed, and the junction temperature change amplitude, peak junction temperature and conduction duration corresponding to each thermal cycle are extracted to obtain fatigue characteristic quantities for power module life prediction. The fatigue characteristics and corresponding electrical parameters are input into the multivariable power module life model to calculate the fatigue damage corresponding to each thermal cycle. The Miner linear damage accumulation criterion is used to accumulate the fatigue damage corresponding to each thermal cycle, and the cumulative fatigue damage and remaining service life of the power module are calculated. The power module is classified into multiple health states based on cumulative fatigue damage. The motor vector control parameters, space vector pulse width modulation waveform parameters, and torque output limits are adaptively adjusted according to the multiple health states to reduce the thermal cycle stress of the power module during subsequent operation. At the same time, the cumulative fatigue damage, remaining service life, and multiple health states are stored online, and corresponding fault warning information and operation and maintenance prompts are output.

[0013] Preferably, the electrical parameters, thermal parameters, and motor-vehicle operating condition parameters each have corresponding acquisition cycles, and each parameter is mapped to a unified calculation time according to the time identifier; among them, the electrical parameters include three-phase stator current and bus DC voltage, the thermal parameters include power module shell temperature, cooling medium temperature and ambient temperature, and the motor-vehicle operating condition parameters include motor speed, output torque, pulse width modulation switching frequency, duty cycle and accelerator pedal opening.

[0014] Preferably, the conduction loss is determined according to the current operating current, the conduction characteristics of the power device corresponding to the current operating state, and the duty cycle; the switching loss is determined by the current operating current, the bus DC voltage, the turn-on energy, the turn-off energy, and the pulse width modulation switching frequency. The conduction loss and switching loss generated by the same power device at continuous calculation times constitute the corresponding instantaneous loss power sequence.

[0015] Preferably, the instantaneous junction temperature is used to select the power device conduction characteristics, turn-on energy, and turn-off energy for the next calculation cycle. The current operating current and the instantaneous junction temperature jointly determine the conduction characteristics, and the current operating current, the bus DC voltage, and the instantaneous junction temperature jointly determine the turn-on energy and turn-off energy, thereby updating the conduction loss and switching loss for the next calculation cycle.

[0016] Preferably, the third-order Foster electrothermal network model has three equivalent thermal response branches corresponding to different time scales. Each equivalent thermal response branch is configured with thermal resistance parameters and time constants. The current temperature rise component is updated using the temperature rise component retained in the previous calculation cycle and the current instantaneous power loss. The cumulative result of each current temperature rise component is correlated with the power module shell temperature to obtain the current transient junction temperature.

[0017] Preferably, the power module case temperature in the thermal parameters constitutes the thermal boundary reference for transient junction temperature calculation, and the cooling medium temperature and ambient temperature characterize the current thermal operating conditions; the thermal resistance parameters and time constants of the third-order Foster electrothermal network model are pre-configured according to the transient thermal impedance characteristics of the power device, and the matching thermal parameter group is called according to the current thermal operating conditions.

[0018] Preferably, rainflow counting continuously identifies the direction of temperature change in the transient junction temperature time series, determines the position where the direction of temperature change reverses as candidate peak or candidate valley value, uses the minimum effective junction temperature change threshold to screen out temperature changes that do not meet the fatigue cycle conditions, and then identifies thermal cycles that meet the cycle closure relationship according to the time sequence of retained peaks and valleys.

[0019] Preferably, peak and valley data that have not yet met the cycle closure relationship during thermal cycling decomposition are retained in the peak and valley sequence to be processed. When the vehicle exits the runnable state, the peak and valley sequence to be processed is saved. When the vehicle re-enters the runnable state, the saved peak and valley data is restored, and the newly generated transient junction temperature sequence is continued to complete the thermal cycling identification across the operating cycle.

[0020] Preferably, the junction temperature variation amplitude in the fatigue characteristic quantity is obtained by the difference between the peak junction temperature and the valley junction temperature in the same thermal cycle, and the conduction time characterizes the cumulative time that the power device bears the effective electrical power load in the thermal cycle; each thermal cycle is also associated with the effective value of the operating current and the time average value of the bus DC voltage during its duration, so as to establish the relationship between the thermal stress and electrical stress data corresponding to the same thermal cycle.

[0021] Preferably, the multivariable power module lifetime model is configured with lifetime correction relationships corresponding to junction temperature variation, peak junction temperature, conduction time, operating current, and bus DC voltage. Each lifetime correction relationship is calculated with the lifetime calibration parameters corresponding to the power device to obtain the allowable number of failures corresponding to the current thermal cycle, and the fatigue damage of the current thermal cycle is determined by the allowable number of failures.

[0022] Preferably, the Miner linear damage accumulation criterion characterizes fatigue damage by the ratio between the actual number of occurrences of each thermal cycle and the corresponding allowable number of failures, and adds new fatigue damage to the historical accumulated fatigue damage; the accumulated fatigue damage serves as the irreversible life loss during the service life of the power module, and maintains the already accumulated damage results when the current transient junction temperature drops.

[0023] Preferably, the remaining service life is determined by the cumulative fatigue damage change within a preset effective operating time window and the corresponding effective operating time, and then the equivalent remaining operating time is obtained by using the remaining available damage between the preset service life damage boundary and the current cumulative fatigue damage and the average damage rate per unit time.

[0024] Preferably, the multi-level health status is determined by the cumulative fatigue damage of each power device to be monitored in the power module. Each power device to be monitored corresponds to the instantaneous power loss sequence, the transient junction temperature sequence, and the cumulative fatigue damage, and the largest cumulative fatigue damage is selected as the judgment parameter of the health status of the power module to characterize the power device with the greatest lifespan consumption in the power module.

[0025] Preferably, each of the multiple health states is configured with a corresponding set of health control parameters. The set of health control parameters includes at least the pulse width modulation switching frequency correction parameter, the maximum allowable torque, and the space vector pulse width modulation transmission parameter. When the current health state changes, the corresponding set of health control parameters is called, and the adjusted operating parameters are fed back into the calculation process of the subsequent instantaneous power loss sequence.

[0026] Preferably, the torque output limit is determined by the vehicle's requested torque and the maximum allowable torque corresponding to the current health state, and the torque current target value in the motor vector control is limited according to the maximum allowable torque; the space vector pulse width modulation waveform parameters include the effective vector action time allocation, the zero vector action time allocation, and the switching sequence selection parameters, which are used to adjust the current load and switching state in the subsequent operation process.

[0027] A motor controller power module life prediction and adaptive health management system includes a multi-source parameter acquisition unit, a real-time loss calculation unit, a junction temperature estimation and life calculation control unit, an adaptive control and drive unit, and a storage and controller local area network communication early warning unit. The multi-source parameter acquisition unit is used to acquire the electrical parameters, thermal parameters of the power module, and the vehicle operating parameters of the motor in real time. The electrical parameters include stator current and bus voltage, the thermal parameters include power module temperature, coolant temperature and ambient temperature, and the vehicle operating parameters of the motor include motor speed, output torque, pulse width modulation switching frequency, pulse width modulation duty cycle and accelerator pedal opening. The real-time loss calculation unit is used to calculate the conduction loss and switching loss of the power module based on electrical parameters and motor vehicle operating parameters, respectively, to obtain the instantaneous loss power sequence. The junction temperature estimation and lifetime calculation control unit is used to input the instantaneous power loss sequence as heat input and the thermal parameters as thermal boundary conditions into the third-order Foster electrothermal network model to obtain the transient junction temperature time series of the power device chip. Rainflow counting is performed on the transient junction temperature time series to extract the junction temperature change amplitude, peak junction temperature and conduction duration corresponding to each thermal cycle to form fatigue characteristic quantities. The fatigue characteristic quantities and corresponding electrical parameters are input into the multivariable power module lifetime model, and the Miner linear damage accumulation criterion is used to obtain the cumulative fatigue damage and remaining service life. The adaptive control and drive unit is used to classify the health state into multiple levels according to the cumulative fatigue damage, and adaptively adjust the motor vector control parameters, space vector pulse width modulation waveform parameters and torque output limit according to the multiple health states to reduce the thermal cycle stress of the power module during subsequent operation and drive the motor power module. The storage and controller LAN communication early warning unit is used to store accumulated fatigue damage, remaining service life and multi-level health status online, and output corresponding fault warning information and maintenance prompts to the vehicle controller and instrument.

[0028] The technical effects and advantages provided by this application in the above technical solution are as follows: This application collects electrical and thermal parameters of the power module in real time, converts conduction and switching losses into a continuous instantaneous power loss sequence, and then uses a third-order Foster electrothermal network model to calculate the transient junction temperature sequence of the power device chip online, effectively preserving the dynamic temperature rise changes of new energy vehicles during frequent start-up, acceleration, deceleration, and heavy-load operation. Furthermore, it decomposes the continuous junction temperature sequence into thermal cycles through rainflow counting, extracting the fatigue cycles corresponding to different junction temperature change amplitudes and peak junction temperatures, thereby transforming the actual thermal cycling process experienced by the chip into fatigue characteristic quantities that can be used for life calculation, improving the correspondence between the thermal state characterization of the power module and fatigue damage calculation.

[0029] This application incorporates junction temperature variation, peak junction temperature, conduction time, and corresponding electrical stress into the power module lifespan prediction process. It continuously updates the accumulated fatigue damage using the Miner linear damage accumulation criterion, and then calculates the remaining service life based on the accumulated fatigue damage and the average damage rate per unit time. This effectively reflects the lifespan consumption under the long-term effects of complex random operating conditions in new energy vehicles. On this basis, it divides the health status into multiple levels according to the accumulated fatigue damage, expanding the lifespan status from a single temperature monitoring result to continuous health information that can reflect the degree of historical thermal cycle consumption. This provides a quantitative reference for subsequent fault warning, maintenance timing judgment, and power module full lifespan status management.

[0030] This application further utilizes multi-level health status to adjust motor vector control parameters, space vector pulse width modulation (SVM) waveform parameters, and torque output limits. As the power module ages, the switching frequency adjustment, torque limitation, and waveform parameter control are fed back to subsequent operation, effectively reducing subsequent switching losses, conduction losses, and thermal cycling stress, thereby slowing down the rate of cumulative fatigue damage growth. Simultaneously, the cumulative fatigue damage, remaining service life, and health status are stored online, and fault warning information and maintenance prompts are output. In the preferred implementation, the processing can be completed using existing operating parameters of the new energy vehicle motor controller and the vehicle network, eliminating the need for additional detection devices to directly measure the chip junction temperature, which is beneficial for its widespread application in electric drive platforms for new energy passenger vehicles and commercial vehicles. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0032] Figure 1 This is a block diagram of the overall architecture of a motor controller power module life prediction and adaptive health management system.

[0033] Figure 2 This is a flowchart for estimating the transient junction temperature of a power module.

[0034] Figure 3 The logic flowchart for power module life prediction and fatigue damage accumulation.

[0035] Figure 4 This is a block diagram of adaptive control logic based on the multi-level health status of the power module. Detailed Implementation

[0036] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0037] like Figures 1 to 4 As shown, this application provides a method for estimating the lifespan of a motor controller power module and for adaptive health management, including the following steps: The electrical and thermal parameters of the power module and the vehicle operating parameters of the motor are collected in real time. The conduction loss and switching loss of the power module are calculated based on the electrical parameters and the vehicle operating parameters of the motor, and the instantaneous power loss sequence of the power module is obtained. The transient thermal response of the power module is dynamically calculated by using the instantaneous power loss sequence as the thermal input and the thermal parameters as the thermal boundary conditions into the third-order Foster electrothermal network model, thereby obtaining the transient junction temperature timing of the power device chip. Rainflow counting is performed on the transient junction temperature time series, the thermal cycles are decomposed, and the junction temperature change amplitude, peak junction temperature and conduction duration corresponding to each thermal cycle are extracted to obtain fatigue characteristic quantities for power module life prediction. The fatigue characteristics and corresponding electrical parameters are input into the multivariable power module life model to calculate the fatigue damage corresponding to each thermal cycle. The Miner linear damage accumulation criterion is used to accumulate the fatigue damage corresponding to each thermal cycle, and the cumulative fatigue damage and remaining service life of the power module are calculated. The power module is classified into multiple health states based on cumulative fatigue damage. The motor vector control parameters, space vector pulse width modulation waveform parameters, and torque output limits are adaptively adjusted according to the multiple health states to reduce the thermal cycle stress of the power module during subsequent operation. At the same time, the cumulative fatigue damage, remaining service life, and multiple health states are stored online, and corresponding fault warning information and operation and maintenance prompts are output.

[0038] A method for estimating the lifespan of a motor controller power module and for adaptive health management is applied during the operation of a new energy vehicle motor controller, and is executed cyclically according to a continuous calculation cycle. (See also...) Figures 1 to 4 The entire process begins with real-time operating data from the power module, converting electrical parameters, thermal parameters, and motor-vehicle operating parameters into an instantaneous power loss sequence. This sequence is then used to calculate the transient junction temperature timing of the power device chip, decomposing the thermal cycle within the transient junction temperature timing to further calculate cumulative fatigue damage and remaining service life. Subsequently, multiple health states are defined based on cumulative fatigue damage, and the motor vector control parameters, space vector pulse width modulation (SPWM) waveform parameters, and torque output limits are adaptively adjusted according to these health states. The adjusted control parameters continue to act on the next operating cycle, which then regenerates the corresponding electrical parameters and motor-vehicle operating parameters, thus establishing a continuous feedback relationship between the lifespan status and subsequent thermal loads.

[0039] Once a new energy vehicle enters an operational state, the system first reads the basic configuration parameters corresponding to the current power module. These parameters include device conduction characteristic parameters, switching energy characteristic parameters, three sets of thermal resistance parameters and time constants in the third-order Foster electrothermal network model, life calibration parameters corresponding to the multivariable power module life model, the minimum effective junction temperature change threshold used for rainflow counting, the damage threshold corresponding to the health state, pulse width modulation switching frequency adjustment parameters corresponding to different health states, torque output limits, and space vector pulse width modulation waveform generation parameters. For power modules put into operation for the first time, the accumulated fatigue damage is initialized to zero. For power modules with historical operating records, the system reads the most recently saved accumulated fatigue damage, remaining service life, multi-level health states, and junction temperature peak-valley sequences that have not yet closed, and continues to estimate life based on the historical data. Accumulated fatigue damage is not reset to zero during normal vehicle power-on and power-off cycles; it is only re-initialized after the power module is replaced and a new life record is established.

[0040] The system collects real-time electrical and thermal parameters of the power module, as well as vehicle operating parameters of the motor. Electrical parameters include at least three-phase stator current and bus DC voltage; thermal parameters include at least power module case temperature, cooling medium temperature, and ambient temperature; and vehicle operating parameters include at least motor speed, output torque, pulse width modulation (PWM) switching frequency, duty cycle, and accelerator pedal opening. Three-phase stator current is used to determine the current operating current; bus DC voltage characterizes the voltage stress currently borne by the power devices; PWM switching frequency and duty cycle are used in the calculation of conduction and switching losses; power module case temperature serves as the primary thermal boundary condition for the third-order Foster electrothermal network model; cooling medium temperature and ambient temperature characterize the current thermal operating environment; and motor speed, output torque, and accelerator pedal opening record the vehicle operating conditions corresponding to the occurrence of thermal cycles.

[0041] Different types of parameters can be collected according to different update cycles. Three-phase stator current, bus DC voltage, pulse width modulation switching frequency, and duty cycle are updated using the first acquisition cycle, while power module case temperature, cooling medium temperature, and ambient temperature are updated using the second acquisition cycle. The first acquisition cycle is shorter than the second acquisition cycle. At each unified calculation moment, the most recent valid sampled value is obtained based on the time identifier corresponding to each parameter, and these values ​​are used to form the data set for the current calculation cycle according to a unified time base. For current, voltage, and temperature sampled data, low-pass filtering or sliding time window averaging can be performed before entering subsequent calculations. The filtering parameters are set to retain the actual load changes caused by vehicle start-up, acceleration, deceleration, cruising, and heavy-load hill climbing, while suppressing isolated spikes caused by sampling noise.

[0042] For the three-phase stator current, calculations can be performed separately for each corresponding power device, or the power device with the highest current load can be selected as the key monitoring target. In a preferred embodiment, instantaneous power loss sequences, transient junction temperature sequences, and cumulative fatigue damage are generated for multiple power devices to be monitored. After calculating the lifetime of each power device, the power device with the maximum cumulative fatigue damage is selected as the lifetime control target of the current power module. This avoids using the average loss of the entire power module to mask the local thermal stress borne by a single power device.

[0043] After obtaining the electrical parameters and the motor's overall vehicle operating parameters, the conduction loss and switching loss are calculated separately. For the IGBT implementation, the on-state voltage drop is obtained based on the device's conduction characteristic data corresponding to the current operating current and current junction temperature. Then, the conduction loss in the current calculation cycle is calculated using the on-state voltage drop, operating current, and the duty cycle corresponding to the current power device. The calculation relationship can be expressed as the conduction loss equal to the product of the on-state voltage drop, operating current, and conduction duty cycle. When the device's on-state voltage drop changes with the operating current and junction temperature, the on-state voltage drop data corresponding to different operating currents and junction temperatures are pre-saved. During operation, the corresponding data is selected according to the current operating current and the transient junction temperature of the previous calculation cycle. When the current operating point is between two pre-saved operating points, linear interpolation can be performed based on the characteristic data of adjacent operating points.

[0044] For implementations using SiC MOSFETs, the conduction loss is determined based on the on-resistance corresponding to the current operating current and junction temperature. The conduction loss can be calculated as the product of the square of the operating current, the on-resistance, and the duty cycle. Similarly, the on-resistance can be based on preset characteristic data established according to the operating current and junction temperature. During operation, the corresponding on-resistance is obtained based on the current operating current and the transient junction temperature of the previous calculation cycle. This allows changes in conduction characteristics caused by an increase in device temperature to be re-included in the loss calculation of the next calculation cycle, achieving continuous updating between electrical losses and thermal conditions.

[0045] During the calculation of switching losses, the turn-on energy and turn-off energy corresponding to the current operating point are obtained. The turn-on energy and turn-off energy are jointly determined by the current operating current, the DC bus voltage, and the transient junction temperature of the previous calculation cycle. Characteristic data on the device's turn-on energy and turn-off energy as a function of operating current, DC bus voltage, and junction temperature are pre-established. In each calculation cycle, the corresponding turn-on energy and turn-off energy are obtained according to the current operating point. When the current operating point is between multiple calibrated operating points, interpolation can be performed in the current direction, voltage direction, and temperature direction to obtain the turn-on energy and turn-off energy corresponding to the current operating point.

[0046] After obtaining the turn-on energy and turn-off energy, add them together and multiply by the current pulse width modulation switching frequency to obtain the switching loss per unit time. When a switching cycle in the actual control process contains multiple switching actions, accumulate the corresponding turn-on energy and turn-off energy according to the actual number of turn-on and turn-off times within one pulse width modulation cycle, and then convert them into the switching loss per unit time. For operating states with freewheeling, calculate the conduction loss and recovery-related loss of the corresponding power devices participating in freewheeling according to the same time base, and then add the total loss of the corresponding power devices.

[0047] Within the same calculation cycle, the conduction loss and switching loss of the current power device are superimposed to obtain the instantaneous power loss at the current calculation moment. If freewheeling loss is further included, then freewheeling-related losses are added to the conduction and switching losses. The instantaneous power loss obtained from multiple consecutive calculation moments is arranged in chronological order to form the instantaneous power loss sequence of the power device corresponding to the power module. Thus, the instantaneous power loss sequence changes with the real-time variations of the three-phase stator current, bus DC voltage, output torque, duty cycle, and pulse width modulation switching frequency, without using a fixed average power to replace the dynamic operating conditions.

[0048] See Figure 2 After obtaining the instantaneous power loss sequence, this sequence is used as the thermal input, and the power module case temperature is used as the primary thermal boundary condition in a third-order Foster electrothermal network model to continuously calculate the transient thermal response of the power device chip. The third-order Foster electrothermal network model includes three equivalent thermal response branches, each with corresponding thermal resistance parameters and time constants. These three branches describe the transient temperature rise response at different time scales, and together they characterize the transient thermal response features between the power device chip and the power module case.

[0049] The thermal resistance parameters corresponding to the three equivalent thermal response branches are denoted as the first thermal resistance parameter, the second thermal resistance parameter, and the third thermal resistance parameter, respectively, and their corresponding time constants are denoted as the first time constant, the second time constant, and the third time constant, respectively. Each thermal resistance parameter and time constant can be predetermined based on the publicly available transient thermal impedance characteristic data of the corresponding power device, or it can be pre-calibrated based on the temperature response data obtained by the power device under different thermal input conditions. Different power device models have their corresponding parameter sets stored separately, and the corresponding parameters are loaded according to the current power device model before vehicle operation.

[0050] For any given calculation cycle, the current temperature rise component of the first equivalent thermal response branch is determined by the first temperature rise component saved from the previous calculation cycle and the current instantaneous power loss. The first temperature rise component saved from the previous calculation cycle decays according to the exponential decay relationship between the current calculation cycle time interval and the first time constant. Then, the current instantaneous power loss, the first thermal resistance parameter, and the thermal response increment corresponding to the current calculation cycle are superimposed to obtain the first temperature rise component of the current calculation cycle. The second and third temperature rise components are updated using the second thermal resistance parameter and the second time constant, and the third thermal resistance parameter and the third time constant, respectively, according to the same processing relationship.

[0051] When using discrete recursion, the current temperature rise component of the i-th equivalent thermal response branch can be expressed as: the temperature rise component of the previous calculation cycle multiplied by an exponential decay term, plus the current instantaneous power loss multiplied by the corresponding thermal resistance parameter, and the additional temperature rise obtained by subtracting the exponential decay term. The exponential decay term is obtained by dividing the current calculation cycle time interval by the corresponding time constant and taking the negative exponent. Through this recursive method, the thermal state of the previous calculation cycle is retained in each calculation cycle, and the heat accumulated in the previous calculation cycle will not immediately disappear due to changes in the current instantaneous power loss.

[0052] After the three temperature rise components are updated, the first, second, and third temperature rise components are added together to obtain the dynamic temperature rise of the power device chip relative to the power module case temperature at the current calculation moment. This dynamic temperature rise is then added to the current power module case temperature to obtain the transient junction temperature of the power device chip at the current calculation moment. The next calculation cycle continues to use the new instantaneous power loss and the three temperature rise components saved from the previous calculation cycle for recursion, thus obtaining a continuous transient junction temperature timing sequence for the power device chip.

[0053] The cooling medium temperature and ambient temperature do not directly replace the power module case temperature as the chip junction temperature reference. Instead, they are recorded together with the power module case temperature to characterize the current thermal operating conditions of the new energy vehicle. When the same power device stores multiple sets of thermal parameters for different cooling conditions, the corresponding thermal parameter set can be selected based on the current cooling medium temperature. If a fixed thermal parameter set is used, the cooling medium temperature and ambient temperature are used for operating condition recording and subsequent maintenance information traceability, without changing the current recursive relationship of the third-order Foster thermoelectric network.

[0054] As new energy vehicles transition from cruising to acceleration or heavy-load hill climbing, the operating current and output torque increase, leading to changes in conduction and switching losses, and consequently altering the instantaneous power loss sequence. The third-order Foster electrothermal network model retains historical thermal responses at three different time scales, outputting a continuous transient junction temperature sequence. Even after the vehicle transitions from heavy-load to deceleration, and the current instantaneous power loss decreases, the temperature rise component saved from the previous calculation cycle continues to participate in the current junction temperature calculation according to its corresponding time constant. Therefore, it effectively reflects the time lag between the power device chip temperature and changes in electrical power.

[0055] See Figure 3After obtaining the transient junction temperature time series, peak-valley identification is continuously performed on the transient junction temperature time series. The direction of change of continuous transient junction temperature data is compared sequentially. When the junction temperature change direction changes from rising to falling, the corresponding turning point is recorded as a candidate peak value; when the junction temperature change direction changes from falling to rising, the corresponding turning point is recorded as a candidate valley value. To prevent temperature sampling fluctuations from being misidentified as fatigue thermal cycles, a minimum effective junction temperature change threshold is set between the candidate peak and candidate valley values. When the temperature difference between adjacent candidate extreme values ​​is less than the minimum effective junction temperature change threshold, the temperature change is not considered an effective thermal cycle turning point; when the minimum effective junction temperature change threshold is reached, the corresponding peak or valley value is added to the peak-valley sequence to be processed.

[0056] The minimum effective junction temperature change threshold is saved as a calibrable parameter and can be set according to the lifespan characteristics of the power device and the temperature acquisition resolution. Preferably, in actual vehicle applications, the minimum effective junction temperature change threshold is set to a temperature variation range that can eliminate sampling noise while retaining the temperature variation range that contributes to fatigue. If a fixed value is used, a specific threshold can be given during the calibration stage of the corresponding power device; if different power devices use different values, the corresponding threshold is called according to the power device model.

[0057] The peak-valley sequence to be processed performs rainflow counting in chronological order, comparing the temperature change intervals between consecutive peaks and valleys. When adjacent temperature change intervals satisfy a thermal cycle closure relationship, the corresponding peak-valley data are paired into a complete thermal cycle, and the intermediate extreme value that has been counted is removed from the peak-valley sequence to be processed. Peak-valley data that do not satisfy the closure relationship are retained in the peak-valley sequence to be processed, waiting for subsequent new transient junction temperature peaks or valleys to enter and continue to participate in the judgment. In this way, junction temperature timing can be processed online during the continuous operation of new energy vehicles, without waiting for the entire vehicle operation cycle to end before calculating the thermal cycle.

[0058] When a vehicle exits the operational state, closed thermal cycles are counted as complete cycles, while unclosed peak-valley sequences are processed using either of two methods. In a preferred embodiment, the unclosed peak-valley sequences are saved along with the current accumulated fatigue damage. Upon the vehicle re-entering the operational state, the unclosed peak-valley sequences are loaded again, and the cycle is re-evaluated against the new transient junction temperature extremes. In another embodiment, when the vehicle ends its current lifespan statistical cycle, unclosed thermal cycles are counted as half-cycles, with a counting weight of 0.5 for half-cycles and a counting weight of 1 for complete thermal cycles. The processing method used remains consistent throughout the same vehicle lifespan record.

[0059] For each identified effective thermal cycle, the junction temperature variation amplitude, peak junction temperature, valley junction temperature, and conduction duration corresponding to that thermal cycle are extracted. The junction temperature variation amplitude is equal to the difference between the peak junction temperature and the valley junction temperature within the same thermal cycle; the peak junction temperature is the maximum transient junction temperature reached within that thermal cycle; and the valley junction temperature is the minimum transient junction temperature within that thermal cycle. In this embodiment, conduction duration is defined as the effective thermal load duration corresponding to the current thermal cycle, that is, the cumulative time during which the power device continuously bears the effective electrical power load from the start of the temperature rise phase of the thermal cycle to the completion of the corresponding temperature drop phase, rather than the instantaneous electrical conduction time within a single pulse width modulation switching cycle. This definition ensures that the conduction duration is consistent with the meaning of the thermal load duration in power cycle lifetime calculation.

[0060] For a complete thermal cycle, the operating current and bus DC voltage are recorded simultaneously during the cycle. The operating current can be the effective value over the current thermal cycle duration, or the time-weighted average current over the thermal cycle duration; the effective value is preferred as it reflects the electrical load during the thermal cycle. The bus DC voltage can be the time average over the same thermal cycle duration. Thus, each effective thermal cycle corresponds to a set of fatigue characteristic quantities, including junction temperature variation, peak junction temperature, valley junction temperature, conduction duration, operating current, and bus DC voltage.

[0061] After fatigue characteristics are established, the junction temperature variation amplitude, peak junction temperature, conduction time, and corresponding electrical parameters are input into the multivariate power module life model. The multivariate power module life model employs a product form that simultaneously describes the impact of junction temperature variation amplitude, temperature level, thermal load duration, and electrical stress on life. Specifically, the allowable number of failures can be expressed as the product of a baseline life coefficient and multiple stress correction terms. The junction temperature variation amplitude is used in life correction using a power function, the peak junction temperature is used in life correction using an exponential temperature correction term, the conduction time is used in life correction using a power function, and the operating current and bus DC voltage are each used in life correction using normalized electrical stress correction terms.

[0062] In practice, the permissible number of failures can be calculated according to the following parameter relationships: First, establish a temperature change amplitude correction term based on the ratio of the junction temperature change amplitude to the reference junction temperature change amplitude; then, establish a temperature stress correction term based on the absolute temperature corresponding to the peak junction temperature; establish a duration correction term based on the ratio of the conduction duration to the reference conduction duration; establish a current stress correction term based on the ratio of the operating current to the rated reference current; establish a voltage stress correction term based on the ratio of the bus DC voltage to the rated reference voltage; then, multiply the above correction terms by the reference life coefficient to obtain the permissible number of failures corresponding to the current thermal cycle.

[0063] The baseline lifetime factor, junction temperature variation index, temperature stress factor, conduction time index, current stress index, and voltage stress index are all pre-saved as lifetime calibration parameters. Different IGBTs or SiC MOSFETs use different sets of lifetime calibration parameters. Each lifetime calibration parameter can be determined by fitting the power cycle lifetime data of the corresponding power device, or it can be established based on the lifetime curve provided by the device and the known operating point lifetime data. As long as the junction temperature variation, peak junction temperature, conduction time, operating current, and bus DC voltage can be converted into the allowable number of failures for the corresponding thermal cycles, they can be used for subsequent fatigue damage calculations.

[0064] For power devices that use the average junction temperature as the lifetime temperature input, the average value can be calculated from the peak junction temperature and the valley junction temperature, and this average value can be used to replace the peak junction temperature in the temperature stress correction term. In the current implementation, it is preferable to directly use the peak junction temperature in the lifetime calculation, so that the highest temperature level reached in each thermal cycle directly participates in the allowable failure count correction. The same temperature input definition is used throughout the entire operating cycle to avoid mixing the peak junction temperature and the average junction temperature in different calculation cycles.

[0065] After obtaining the permissible number of failures for the current thermal cycle, the fatigue damage for that thermal cycle is calculated. For a complete thermal cycle, the fatigue damage per cycle equals 1 divided by the permissible number of failures for the current thermal cycle; for an unclosed thermal cycle recorded in a half-cycle manner, the fatigue damage per cycle equals 0.5 divided by the corresponding permissible number of failures. When multiple thermal cycles exist within the same fatigue characteristic range, the actual number of cycles can be divided by the permissible number of failures corresponding to that fatigue characteristic range to obtain the cumulative fatigue contribution of that type of thermal cycle.

[0066] Subsequently, the Miner linear damage accumulation criterion is adopted to add new thermal cycle fatigue damage to the historical accumulated fatigue damage. The accumulated fatigue damage is denoted as D. Each time a new complete thermal cycle or half-cycle occurs, D is updated according to the corresponding fatigue damage. The theoretical lifetime damage boundary DL is used as the upper limit of damage corresponding to lifetime depletion, preferably with DL set to 1. When D is less than DL, it indicates that there is still usable lifetime; when D reaches or exceeds DL, it indicates that the accumulated fatigue consumption calculated according to the current lifetime model has reached the theoretical lifetime boundary. During operation, D is stored as historical damage and does not decrease due to a drop in the current temperature of the power device.

[0067] After calculating the cumulative fatigue damage, the remaining service life is further calculated. The remaining service life is obtained by combining the current cumulative fatigue damage and the average damage rate per unit time. First, an effective operating time statistics window is set. Within this window, the cumulative fatigue damage at the start and end times is recorded. The difference between the two is divided by the effective time the power module is actually in operation within this statistics window to obtain the average damage rate per unit time. The time spent when the vehicle is stopped, powered off, or when the motor controller is not in an effective operating state is not included in the effective operating time.

[0068] The statistical window can be a fixed operating time window or a cumulative operating time window corresponding to the most recent few effective thermal cycles. A fixed effective operating time window is preferred, and it should be updated on a rolling basis in each calculation cycle. To avoid sudden changes in the average damage rate per unit time due to a single special operating condition within a short period, the average damage rate can also be updated using a weighted average of the damage rates from the current and previous statistical windows. This processing only changes the statistical method of the damage rate and does not change the cumulative fatigue damage itself.

[0069] After obtaining the average damage rate per unit time, the current accumulated fatigue damage D is subtracted from the theoretical life damage boundary DL to obtain the remaining usable damage. Then, the remaining usable damage is divided by the average damage rate per unit time to obtain the equivalent remaining operating time under the current operating characteristics. The remaining operating time is calculated when the average damage rate per unit time is greater than zero; when the average damage rate per unit time is equal to zero, no division by zero is performed, and the effective remaining service life result from the previous cycle is retained, or the remaining service life status is set to exceed the preset display limit. For new power modules that have not yet established an effective statistical window, the remaining service life can be output for the first time after the accumulated effective operating time reaches the preset limit.

[0070] If the equivalent remaining cycle life needs to be output, the average single-cycle fatigue damage corresponding to all effective thermal cycles is calculated within the current statistical window. Then, the remaining available damage is divided by the average single-cycle fatigue damage to obtain the equivalent remaining cycle count. The remaining service life can be output in hours, and the equivalent remaining cycle life can be output in cycles. Both results are continuously updated with new thermal cycles and new accumulated fatigue damage.

[0071] See Figure 4After calculating the cumulative fatigue damage and remaining service life, the power module is classified into multiple health states based on the cumulative fatigue damage. Preferably, three health states are used: healthy level, early warning level, and fault alarm level. The first damage threshold is preferably set to 0.3, the second damage threshold is preferably set to 0.7, and the theoretical life damage boundary is preferably set to 1. When D is not greater than 0.3, the current power module is classified as healthy; when D is greater than 0.3 but not greater than 0.7, the current power module is classified as early warning level; when D is greater than 0.7, the current power module is classified as fault alarm level.

[0072] Cumulative fatigue damage reflects irreversible lifespan loss. Therefore, within the same power module's service life, multiple health states change according to the order of healthy level, warning level, and fault alarm level, and do not automatically revert to the previous healthy state due to a decrease in junction temperature or the vehicle switching to a light-load state. Only after the power module is replaced and a new lifespan record is established will the health state be reclassified according to the new cumulative fatigue damage.

[0073] In the implementation method of estimating the lifespan of multiple power devices separately, the cumulative fatigue damage of each power device is calculated, and the maximum cumulative fatigue damage is selected as the representative damage value used when classifying the current power module into healthy states. This avoids situations where some power devices have already suffered significant fatigue damage while the overall average remains within the healthy range.

[0074] When the power module is currently in a healthy state, it executes motor control according to the vehicle's normal power requirements, maintaining normal motor vector control parameters, normal space vector pulse width modulation waveform parameters, and normal torque output limits. In this state, the driver's normal power request is not additionally restricted due to the lifespan status, but electrical parameter acquisition, thermal parameter acquisition, instantaneous power loss sequence generation, transient junction temperature timing calculation, thermal cycle decomposition, cumulative fatigue damage update, and remaining service life calculation are still continuously performed.

[0075] Once the power module enters the warning level, the pulse width modulation (PWM) switching frequency is adaptively adjusted based on the corresponding health control parameters. Specifically, the target PWM switching frequency under normal control is first obtained, then multiplied by a first switching frequency correction coefficient to obtain the target switching frequency for the warning level. The first switching frequency correction coefficient is set between 0 and 1, with its specific value calibrated according to motor operating performance and device thermal state. By reducing the number of switching actions per unit time, switching losses are reduced, limiting the instantaneous power loss generated in the next calculation cycle.

[0076] Simultaneously, a first maximum permissible torque is generated based on the torque limit parameters corresponding to the warning level. When the vehicle's normal driving needs generate a requested torque, this requested torque is compared with the first maximum permissible torque, and a torque target whose value does not exceed the first maximum permissible torque is selected for subsequent motor control. This limits the operating current corresponding to continuous high torque conditions, reducing conduction losses and continuous thermal load.

[0077] The space vector pulse width modulation (SPWM) waveform generation parameters are adjusted at the warning level according to the first waveform parameter set. This first set of parameters includes at least the effective vector action time allocation, the zero vector action time allocation, and the switching sequence selection parameters. During adjustment, the motor target voltage vector is maintained to meet the current torque demand constraints, while reducing waveform combinations that could cause current spikes and frequent switching operations. Specifically, during calibration, waveform parameter sets corresponding to the warning level can be pre-saved for different speeds, torques, and bus DC voltage regions. During operation, the corresponding parameters are called based on the current motor speed, current torque target, and bus DC voltage.

[0078] The motor vector control parameters can limit the target torque and current values ​​based on the current health status. In the warning level, the corresponding upper limit of torque and current is calculated based on the first maximum allowable torque, and the current target torque and current value must not exceed this upper limit. In the field weakening operation region, a continuous current limit can also be set for the warning level to prevent the motor from maintaining a high current state for extended periods during high-speed operation. Through the combined effect of torque and current target limiting and switching frequency adjustment, both conduction losses and switching losses can be brought within a controllable range.

[0079] When accumulated fatigue damage reaches the fault alarm level, the second switching frequency correction coefficient, the second maximum permissible torque, and the second set of waveform parameters corresponding to the fault alarm level are read. The second switching frequency correction coefficient is less than or equal to the first switching frequency correction coefficient, and the second maximum permissible torque is less than or equal to the first maximum permissible torque, thereby further limiting the continuous thermal load at the fault alarm level. For heavy-load uphill driving, continuous high-speed driving, and long-term high-torque operation, the actual torque target is limited according to the second maximum permissible torque to avoid power devices operating in the corresponding high thermal load area for extended periods.

[0080] Multiple control actions are executed according to a unified set of health status parameters. The health level calls the normal parameter set, the warning level calls the first health control parameter set, and the fault alarm level calls the second health control parameter set. Each parameter set includes pulse width modulation (PWM) switching frequency correction parameters, maximum permissible torque, and space vector PWM waveform generation parameters, thus avoiding inconsistent health states for different control actions. If basic vehicle safety control, braking control, or other higher-priority vehicle control requirements conflict with health control, basic vehicle safety control takes priority, and health control executes within the limits of meeting vehicle safety constraints.

[0081] In continuous-time processing, the k-th calculation cycle obtains the cumulative fatigue damage, remaining service life, and multi-level health status based on the current electrical parameters, thermal parameters, and motor-vehicle operating parameters. Then, based on the multi-level health status of the k-th calculation cycle, it generates motor vector control parameters, space vector pulse width modulation (PWM) waveform parameters, and torque output limits for subsequent control cycles. After the control parameters are applied, the next operating cycle generates new three-phase stator current, bus DC voltage, output torque, duty cycle, and PWM switching frequency. New conduction losses and switching losses are then calculated based on these new operating parameters.

[0082] Thus, as the pulse width modulation (PWM) switching frequency decreases, the number of switching operations per unit time decreases, correspondingly reducing switching losses. With the torque output limit imposed, sustained high-current operation is restricted, thus limiting conduction losses and sustained thermal load. Adjustments to the space vector PWM transmission parameters suppress current spikes and some frequent switching states. The new conduction and switching losses regenerate a new instantaneous power loss sequence, which is then re-entered into the third-order Foster electrothermal network model to calculate the new transient junction temperature sequence, thereby continuing to update subsequent thermal cycles, accumulated fatigue damage, and remaining service life.

[0083] Therefore, this implementation does not reduce power all at once after the temperature exceeds a single threshold. Instead, it uses accumulated fatigue damage as a state variable reflecting historical lifespan consumption, and then controls subsequent motor operating parameters through multi-level health status control. The more thermal cycles the current power module has experienced or the stronger the thermal cycle stress, the greater the accumulated fatigue damage, and the multi-level health status changes accordingly. Subsequent control parameters then enter the corresponding health management range, and the thermal load in the next stage is re-involved in the lifespan calculation, thus constituting a continuous adaptive health management process.

[0084] Accumulated fatigue damage, remaining service life, and multi-level health status are stored online. Storage trigger conditions include at least periodic storage conditions, cumulative fatigue damage changes reaching a preset increment, changes in health status, and the vehicle exiting an operational state. When any of these conditions are met, the current accumulated fatigue damage, remaining service life, multi-level health status, corresponding operating time, and the current unclosed peak-valley sequence are written to a non-volatile storage area. Upon the vehicle re-entering an operational state, the most recently saved results are read, and subsequent lifespan predictions are performed.

[0085] To control the frequency of storage updates, a "periodic write + event write" approach can be used. Periodic writes save the lifetime status according to a preset effective running time interval; event writes are executed immediately when the health status changes from healthy to warning level, from warning level to fault alarm level, or when the vehicle is about to exit the operational state. This method can maintain the continuity of historical lifetime status while reducing redundant writes when there are no status changes.

[0086] Historical data can also store transient junction temperature characteristics, cumulative fatigue damage increments, motor speed, output torque, bus DC voltage, operating current, power module case temperature, and thermal cycling fatigue characteristics related to the lifespan of the device. In actual storage, it is not necessary to permanently save all high-frequency original transient junction temperature points. Instead, the most recent transient junction temperature sequence can be retained according to a preset time window, and the maximum junction temperature, junction temperature variation distribution, cumulative fatigue damage, and health status corresponding to each statistical period can be stored long-term, thus supporting subsequent operational status traceability.

[0087] Fault warning and maintenance prompts are generated based on multi-level health status and remaining service life, and sent to the vehicle controller and instrument cluster via the controller area network. The health level outputs normal service life status; the warning level outputs power module aging warnings, current remaining service life, and recommended maintenance prompts; the fault alarm level outputs fault alarms, remaining service life information, and maintenance prompts for power module inspection or replacement. New status information is immediately sent when the health status changes, and remaining service life and accumulated fatigue damage are updated and sent according to a preset communication cycle.

[0088] See Figure 1 , Figure 1 This corresponds to the overall data transfer relationship from multi-source operating parameters to instantaneous power loss calculation, transient junction temperature calculation, lifetime calculation, multi-level health state classification, and subsequent control parameter adjustment; see [link to relevant documentation]. Figure 2 , Figure 2 The process of using the instantaneous power loss sequence as heat input into a third-order Foster electrothermal network model to continuously obtain the transient junction temperature timing of the power device chip; see [link to relevant documentation]. Figure 3 , Figure 3The process of obtaining cumulative fatigue damage and remaining service life after peak-valley identification, rainflow counting, fatigue characteristic extraction, multivariate power module lifetime modeling, and Miner linear damage accumulation corresponding to transient junction temperature time series; see [link to relevant documentation]. Figure 4 , Figure 4 This process involves classifying health states into multiple levels based on cumulative fatigue damage, and then adaptively adjusting motor vector control parameters, space vector pulse width modulation waveform parameters, and torque output limits according to these multiple health states.

[0089] To further illustrate the continuous operation process, the following implementation scenario can be adopted. During a certain operating cycle, the new energy vehicle continuously collects data on three-phase stator current, bus DC voltage, power module case temperature, cooling medium temperature, ambient temperature, motor speed, output torque, pulse width modulation switching frequency, duty cycle, and accelerator pedal opening. First, conduction losses are obtained based on the current operating current, device conduction characteristics, and duty cycle. Switching losses are obtained based on turn-on energy, turn-off energy, and pulse width modulation switching frequency. These two are then superimposed and entered into the current instantaneous power loss sequence. Subsequently, transient junction temperature timing is continuously obtained through a third-order Foster electrothermal network model. After peak-valley changes occur in the transient junction temperature timing, rainflow counting outputs the corresponding thermal cycle, and the junction temperature change amplitude, peak junction temperature, valley junction temperature, and conduction duration are extracted. Then, fatigue characteristic quantities and corresponding operating current and bus DC voltage are input into a multivariable power module life model to obtain the allowable number of failures for that thermal cycle and convert it into fatigue damage. Fatigue damage is added to historical accumulated fatigue damage and then updated to D. Simultaneously, the remaining service life is updated using the unit-time average damage rate in the statistical window.

[0090] When the updated D is within the healthy range, normal control parameters continue to be used. When D enters the warning range, the system switches to the first health control parameter set, adjusting the pulse width modulation switching frequency, maximum allowable torque, upper limit of torque current, and space vector pulse width modulation waveform parameters. When D enters the fault alarm range, the system switches to the second health control parameter set, further limiting continuous high-power operation and simultaneously outputting fault warning information and maintenance prompts. The actual operating parameters after the control parameter adjustment re-enter the next calculation cycle, thus continuously completing the closed-loop process of "real-time acquisition - instantaneous power loss sequence - transient junction temperature sequence - rainflow counting - fatigue characteristic quantities - cumulative fatigue damage and remaining service life - multi-level health status - adaptive control - next operating cycle".

[0091] In preferred applications, the first damage threshold is 0.3, the second damage threshold is 0.7, and the theoretical lifetime damage boundary is 1. The thermal resistance parameters and time constants in the third-order Foster electrothermal network model, the lifetime calibration parameters corresponding to the multivariable power module lifetime model, the minimum effective junction temperature change threshold, the first switching frequency correction coefficient, the second switching frequency correction coefficient, the first maximum allowable torque, the second maximum allowable torque, and the space vector pulse width modulation (SPWM) generation parameters at each stage are all pre-configured according to the specific IGBT or SiC MOSFET and the corresponding new energy vehicle electric drive control requirements. This approach maintains compatibility between different power devices and ensures that all critical processing steps have clear data inputs, processing relationships, and output results.

[0092] Through the above implementation methods, electrical parameters, thermal parameters, and motor-vehicle operating parameters can be continuously transmitted to subsequent processes along a unified time base. Conduction losses and switching losses can be converted into instantaneous power loss sequences in real time. The third-order Foster electrothermal network model can continuously calculate the transient junction temperature sequence of power device chips using current losses and historical thermal states. Rainflow counting can perform online thermal cycle decomposition of the transient junction temperature sequence under random operating conditions. The multivariate power module life model and Miner linear damage accumulation criterion can convert each thermal cycle into cumulative fatigue damage and remaining service life. Multi-level health states further control the motor vector control parameters, space vector pulse width modulation waveform parameters, and torque output limits, and re-input the new operating state after control into the next calculation cycle, thereby completing the continuous implementation process of motor controller power module life prediction and adaptive health management.

[0093] A power module life prediction and adaptive health management system for a motor controller is characterized by comprising a multi-source parameter acquisition unit, a real-time loss calculation unit, a junction temperature estimation and life calculation control unit, an adaptive control and drive unit, and a storage and controller local area network communication early warning unit. The multi-source parameter acquisition unit is used to acquire the electrical parameters, thermal parameters of the power module, and the vehicle operating parameters of the motor in real time. The electrical parameters include stator current and bus voltage, the thermal parameters include power module temperature, coolant temperature and ambient temperature, and the vehicle operating parameters of the motor include motor speed, output torque, pulse width modulation switching frequency, pulse width modulation duty cycle and accelerator pedal opening. The real-time loss calculation unit is used to calculate the conduction loss and switching loss of the power module based on electrical parameters and motor vehicle operating parameters, respectively, to obtain the instantaneous loss power sequence. The junction temperature estimation and lifetime calculation control unit is used to input the instantaneous power loss sequence as heat input and the thermal parameters as thermal boundary conditions into the third-order Foster electrothermal network model to obtain the transient junction temperature time series of the power device chip. Rainflow counting is performed on the transient junction temperature time series to extract the junction temperature change amplitude, peak junction temperature and conduction duration corresponding to each thermal cycle to form fatigue characteristic quantities. The fatigue characteristic quantities and corresponding electrical parameters are input into the multivariable power module lifetime model, and the Miner linear damage accumulation criterion is used to obtain the cumulative fatigue damage and remaining service life. The adaptive control and drive unit is used to classify the health state into multiple levels according to the cumulative fatigue damage, and adaptively adjust the motor vector control parameters, space vector pulse width modulation waveform parameters and torque output limit according to the multiple health states to reduce the thermal cycle stress of the power module during subsequent operation and drive the motor power module. The storage and controller LAN communication early warning unit is used to store accumulated fatigue damage, remaining service life and multi-level health status online, and output corresponding fault warning information and maintenance prompts to the vehicle controller and instrument.

[0094] The present invention provides a method for estimating the lifespan of a motor controller power module and adaptive health management, which is implemented by the aforementioned system for estimating the lifespan of a motor controller power module and adaptive health management. For details of the specific method and process of the system for estimating the lifespan of a motor controller power module and adaptive health management, please refer to the embodiment of the above-mentioned method for estimating the lifespan of a motor controller power module and adaptive health management, which will not be repeated here.

[0095] The foregoing has only described certain exemplary embodiments of this application by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of this application. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of this application.

Claims

1. A method for estimating the lifespan and adaptive health management of a motor controller power module, characterized in that, Includes the following steps: The electrical and thermal parameters of the power module and the vehicle operating parameters of the motor are collected in real time. The conduction loss and switching loss of the power module are calculated based on the electrical parameters and the vehicle operating parameters of the motor, and the instantaneous power loss sequence of the power module is obtained. The transient thermal response of the power module is dynamically calculated by using the instantaneous power loss sequence as the thermal input and the thermal parameters as the thermal boundary conditions into the third-order Foster electrothermal network model, thereby obtaining the transient junction temperature timing of the power device chip. Rainflow counting is performed on the transient junction temperature time series, the thermal cycles are decomposed, and the junction temperature change amplitude, peak junction temperature and conduction duration corresponding to each thermal cycle are extracted to obtain fatigue characteristic quantities for power module life prediction. The fatigue characteristics and corresponding electrical parameters are input into the multivariable power module life model to calculate the fatigue damage corresponding to each thermal cycle. The Miner linear damage accumulation criterion is used to accumulate the fatigue damage corresponding to each thermal cycle, and the cumulative fatigue damage and remaining service life of the power module are calculated. The power module is classified into multiple health states based on cumulative fatigue damage. The motor vector control parameters, space vector pulse width modulation waveform parameters, and torque output limits are adaptively adjusted according to the multiple health states to reduce the thermal cycle stress of the power module during subsequent operation. At the same time, the cumulative fatigue damage, remaining service life, and multiple health states are stored online, and corresponding fault warning information and operation and maintenance prompts are output.

2. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, Electrical parameters, thermal parameters, and motor-vehicle operating condition parameters each have corresponding acquisition cycles, and each parameter is mapped to a unified calculation time according to the time identifier. Among them, electrical parameters include three-phase stator current and bus DC voltage, thermal parameters include power module shell temperature, cooling medium temperature, and ambient temperature, and motor-vehicle operating condition parameters include motor speed, output torque, pulse width modulation switching frequency, duty cycle, and accelerator pedal opening.

3. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, Conduction losses are determined by the current operating current, the conduction characteristics of the power device corresponding to the current operating state, and the duty cycle; switching losses are determined by the current operating current, the bus DC voltage, the turn-on energy, the turn-off energy, and the pulse width modulation switching frequency. The conduction losses and switching losses generated by the same power device at continuous calculation times constitute the corresponding instantaneous power loss sequence.

4. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 3, characterized in that, The instantaneous junction temperature is used to select the power device's conduction characteristics, turn-on energy, and turn-off energy for the next calculation cycle. The current operating current and instantaneous junction temperature jointly determine the conduction characteristics, while the current operating current, bus DC voltage, and instantaneous junction temperature jointly determine the turn-on energy and turn-off energy. This updates the conduction loss and switching loss for the next calculation cycle.

5. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, The third-order Foster electrothermal network model has three equivalent thermal response branches corresponding to different time scales. Each equivalent thermal response branch is configured with thermal resistance parameters and time constants. The current temperature rise component is updated using the temperature rise component retained from the previous calculation cycle and the current instantaneous power loss. The cumulative result of each current temperature rise component is correlated with the power module case temperature to obtain the current transient junction temperature.

6. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 5, characterized in that, The power module case temperature in the thermal parameters constitutes the thermal boundary reference for transient junction temperature calculation, while the cooling medium temperature and ambient temperature characterize the current thermal operating conditions. The thermal resistance parameters and time constants of the third-order Foster electrothermal network model are pre-configured according to the transient thermal impedance characteristics of the power devices, and the matching thermal parameter set is called according to the current thermal operating conditions.

7. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, Rainflow counting continuously identifies the direction of temperature change in the transient junction temperature time series, determines the position where the direction of temperature change reverses as candidate peak or candidate valley value, uses the minimum effective junction temperature change threshold to screen out temperature changes that do not meet the fatigue cycle conditions, and then identifies thermal cycles that meet the cycle closure relationship according to the time sequence of retained peaks and valleys.

8. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 7, characterized in that, Peak and valley data that have not yet met the cycle closure relationship during thermal cycling decomposition are retained in the peak and valley sequence to be processed. When the vehicle exits the runnable state, the peak and valley sequence to be processed is saved. When the vehicle re-enters the runnable state, the saved peak and valley data is restored, and the newly generated transient junction temperature sequence is continued to complete the thermal cycling identification across the operating cycle.

9. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, The junction temperature variation amplitude in the fatigue characteristic quantities is obtained by the difference between the peak junction temperature and the valley junction temperature in the same thermal cycle. The conduction time characterizes the cumulative time that the power device bears the effective electrical power load in the thermal cycle. Each thermal cycle is also associated with the effective value of the operating current and the time average value of the bus DC voltage during its duration, so as to establish the relationship between the thermal stress and electrical stress data corresponding to the same thermal cycle.

10. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 9, characterized in that, The multivariable power module lifetime model is configured with lifetime correction relationships corresponding to junction temperature variation, peak junction temperature, conduction time, operating current, and bus DC voltage. Each lifetime correction relationship is calculated with the lifetime calibration parameters corresponding to the power device to obtain the allowable number of failures for the current thermal cycle, and the fatigue damage of the current thermal cycle is determined by the allowable number of failures.

11. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, Miner's linear damage accumulation criterion characterizes fatigue damage by the ratio between the actual number of occurrences of each thermal cycle and the corresponding allowable number of failures, and adds new fatigue damage to the historical cumulative fatigue damage. Accumulated fatigue damage, as an irreversible loss of life during the service life of the power module, retains the accumulated damage even as the current transient junction temperature drops.

12. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 11, characterized in that, The remaining service life is determined by the cumulative fatigue damage change within the preset effective operating time window and the corresponding effective operating time, and then the equivalent remaining operating time is obtained by using the remaining available damage between the preset service life damage boundary and the current cumulative fatigue damage and the average damage rate per unit time.

13. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, The multi-level health status is determined by the cumulative fatigue damage of each power device to be monitored in the power module. Each power device to be monitored corresponds to the instantaneous power loss sequence, transient junction temperature sequence and cumulative fatigue damage, and the largest cumulative fatigue damage is selected as the judgment parameter of the power module health status to characterize the power device with the greatest life consumption in the power module.

14. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 1, characterized in that, Each health status level is configured with a corresponding set of health control parameters. The set of health control parameters includes at least the pulse width modulation switching frequency correction parameter, the maximum allowable torque, and the space vector pulse width modulation transmission parameter. When the current health status changes, the corresponding set of health control parameters is called, and the adjusted operating parameters are fed back into the calculation process of the subsequent instantaneous power loss sequence.

15. The method for estimating the lifespan and adaptive health management of a motor controller power module according to claim 14, characterized in that, The torque output limit is determined by the vehicle's requested torque and the maximum allowable torque corresponding to the current health state, and the target value of torque current in motor vector control is limited according to the maximum allowable torque; the space vector pulse width modulation waveform parameters include effective vector action time allocation, zero vector action time allocation and switching sequence selection parameters, which are used to adjust the current load and switching state in subsequent operation.

16. A power module lifespan prediction and adaptive health management system for a motor controller, characterized in that, It includes a multi-source parameter acquisition unit, a real-time loss calculation unit, a junction temperature estimation and lifetime calculation control unit, an adaptive control and drive unit, and a storage and controller LAN communication early warning unit; The multi-source parameter acquisition unit is used to acquire the electrical parameters, thermal parameters of the power module, and the vehicle operating parameters of the motor in real time. The electrical parameters include stator current and bus voltage, the thermal parameters include power module temperature, coolant temperature and ambient temperature, and the vehicle operating parameters of the motor include motor speed, output torque, pulse width modulation switching frequency, pulse width modulation duty cycle and accelerator pedal opening. The real-time loss calculation unit is used to calculate the conduction loss and switching loss of the power module based on electrical parameters and motor vehicle operating parameters, respectively, to obtain the instantaneous loss power sequence. The junction temperature estimation and lifetime calculation control unit is used to input the instantaneous power loss sequence as heat input and the thermal parameters as thermal boundary conditions into the third-order Foster electrothermal network model to obtain the transient junction temperature time series of the power device chip. Rainflow counting is performed on the transient junction temperature time series to extract the junction temperature change amplitude, peak junction temperature and conduction duration corresponding to each thermal cycle to form fatigue characteristic quantities. The fatigue characteristic quantities and corresponding electrical parameters are input into the multivariable power module lifetime model, and the Miner linear damage accumulation criterion is used to obtain the cumulative fatigue damage and remaining service life. The adaptive control and drive unit is used to classify the health state into multiple levels according to the cumulative fatigue damage, and adaptively adjust the motor vector control parameters, space vector pulse width modulation waveform parameters and torque output limit according to the multiple health states to reduce the thermal cycle stress of the power module during subsequent operation and drive the motor power module. The storage and controller LAN communication early warning unit is used to store accumulated fatigue damage, remaining service life and multi-level health status online, and output corresponding fault warning information and maintenance prompts to the vehicle controller and instrument.

Images