一种电动汽车功率模块的智能热管理方法

Abstract

This invention discloses an intelligent thermal management method for an electric vehicle power module, comprising: constructing a simulation model of an electric vehicle drive system in a physical simulation platform to obtain a power loss value sequence for each power semiconductor device; constructing a finite element model of the power module, outputting the thermal impedance curves of the power semiconductor devices, and calculating the thermal resistance of the power semiconductor devices; constructing a coupled network model of the power module to obtain a junction temperature estimation model, and outputting a real-time junction temperature time series based on the power loss value sequence; updating the total cumulative damage after a thermal cycling event and calculating the remaining service life of the power module; predicting the future total cumulative damage of the power module, evaluating whether the operating condition of the power module is normal based on the predicted total cumulative damage and the instantaneous cumulative rate, and performing thermal management of the power module.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle power module optimization, and more specifically to an intelligent thermal management method for electric vehicle power modules. Background Technology

[0002] To address the increasingly severe challenges of energy security and environmental protection, developing new energy vehicles has become a global consensus and an important strategic direction. In the core three-electric system (battery, motor, and electronic control) of new energy vehicles, power electronic modules, represented by insulated-gate bipolar transistors (IGBTs), undertake the crucial task of converting DC to AC power, directly determining the performance and efficiency of the motor drive. As electric vehicles develop towards higher power density and lighter weight, power modules generate significant power losses under high-speed, frequent switching conditions, resulting in severe heat loads.

[0003] Existing thermal management technologies for power modules mainly focus on independent loss modeling, thermal simulation analysis, or offline lifetime assessment. For example, loss analysis is often based on empirical models built from device datasheets; thermal simulation analysis uses equivalent thermal network models or finite element methods to estimate junction temperature. While these technologies have achieved some results in their respective areas, they generally suffer from fragmented models and broken data chains, failing to form an organic whole. More importantly, current vehicle thermal management strategies are mostly passive responses, triggering power reduction or enhanced heat dissipation only when the temperature exceeds a safety threshold. This approach cannot effectively mitigate the progressive fatigue damage caused by thermal cycling stress during daily operation, thus becoming one of the main bottlenecks restricting the long-term reliability of power modules. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, this invention provides an intelligent thermal management method for electric vehicle power modules.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0006] A method for intelligent thermal management of an electric vehicle power module is provided, comprising the following steps:

[0007] S1: Construct a simulation model of the electric vehicle drive system in the physical simulation platform to reproduce the operation control logic of the electric vehicle, output the conduction loss and switching loss of each power semiconductor device in the power module, and obtain the power loss value sequence of each power semiconductor device.

[0008] S2: Construct a finite element model of the power module, obtain the three-dimensional partial differential equation of heat conduction characterizing transient thermal behavior, apply a unit power step to the finite element model, output the thermal impedance curve of the power semiconductor device, and calculate the thermal resistance of the power semiconductor device.

[0009] S3: Construct a coupled network model within the power module that reflects all heat conduction paths using thermal resistance, reduce the order of the three-dimensional heat conduction partial differential equation to obtain a junction temperature estimation model, and output a real-time junction temperature time series based on the power loss value sequence.

[0010] S4: Extract key feature vectors characterizing multi-physical damage mechanisms from the real-time junction temperature time series, calculate the fatigue damage increment, thermal shock damage increment, and high-temperature creep damage increment caused by thermal cycling events, fuse them, update the total cumulative damage after thermal cycling events, and calculate the remaining service life of the power module.

[0011] S5: Calculate the instantaneous accumulation rate of fatigue damage, thermal shock damage, and high-temperature creep damage of the power module using fatigue damage increment, thermal shock damage increment, and high-temperature creep damage increment. Predict the total cumulative damage of the power module in the future. Based on the predicted total cumulative damage and instantaneous accumulation rate, assess whether the operating condition of the power module is normal and perform thermal management of the power module.

[0012] Further, step S1 includes:

[0013] S11: Construct a simulation model of the electric vehicle drive system in the physical simulation platform. The simulation model includes a control model, a modulator and inverter model, and a motor model. The simulation model is used to reproduce the operation control logic of the electric vehicle and the conduction loss and switching loss of each power semiconductor device in the output power module.

[0014] S12: Summate the conduction loss and switching loss of each power semiconductor device to obtain the power loss of each power semiconductor device, and output the power loss value sequence of each power semiconductor device in real time. , i This is the designation for power semiconductor devices. t For time.

[0015] Furthermore, the method for outputting conduction loss is as follows:

[0016] The output characteristic curves provided in the datasheets of the power semiconductor devices in the power module are embedded into the simulation model in the form of a lookup table. The simulation model acquires the instantaneous current flowing through each power semiconductor device in real time. And duty cycle signal;

[0017] The instantaneous on-state voltage drop corresponding to the calculated power semiconductor device can be queried. Based on instantaneous on-state pressure drop and instantaneous current Calculate instantaneous conduction power ;

[0018] In a switching cycleT s Internal instantaneous conduction power Integrate and average to obtain the conduction loss of each power semiconductor device.

[0019] Furthermore, the method for outputting switching losses is as follows:

[0020] The switching energy loss curves provided in the datasheets of the power semiconductor devices in the power module are embedded into the simulation model in the form of a lookup table;

[0021] The simulation model queries and calculates the single-turn-on energy based on the real-time electrical boundary conditions at the time of each switching event. E on and shutting off energy E off ;

[0022] Energy will be activated only once. E on and shutting off energy E off Multiply by switching frequency Thus, the switching losses are obtained.

[0023] Further, step S2 includes:

[0024] S21: Construct a finite element model of the power module in terms of geometry and material properties using finite element simulation software in a three-dimensional multiphysics simulation environment; assign corresponding thermal property parameters to each component in the finite element model.

[0025] S22: Constructing three-dimensional partial differential equations of heat conduction to characterize the transient thermal behavior of the finite element model:

[0026] S23: Apply a unit power step to one of the power semiconductor devices in the finite element model, while the power of the other power semiconductor devices is zero; record all... N The temperature of a power semiconductor device over time t The rising curve is used as the thermal impedance curve of power semiconductor devices.

[0027] S24: Use the diagonal element of the thermal impedance curve as the self-thermal impedance curve of the power semiconductor device. Off-diagonal elements as coupling thermal resistance curves between power semiconductor devices , j For power semiconductor devices i Different power semiconductor device designations;

[0028] The self-thermal impedance curve of each power semiconductor device The thermal resistance is fitted to a self-heating RC chain composed of multiple RC units connected in series, thus obtaining the thermal resistance characterizing the main heat dissipation path from the power semiconductor device to the heat sink. and heat capacity , k Number the thermal cycling events;

[0029] The coupling thermal impedance curve of each power semiconductor device By performing fitting, characteristics of power semiconductor devices can be identified. i With power semiconductor devices j Coupling thermal resistance of thermal coupling strength between .

[0030] Further, step S3 includes:

[0031] S31: Each power semiconductor device corresponds to a vertical self-heating RC chain, and different power semiconductor devices are coupled through thermal resistance. Horizontal connections yield a coupled network model within the power module that reflects all heat conduction paths;

[0032] S32: Based on the coupled network model, the three-dimensional heat conduction partial differential equation is reduced in order to obtain a set of first-order ordinary differential equations characterizing the transient thermal behavior of the coupled network model, which serves as a junction temperature estimation model.

[0033] S33: Input the junction temperature estimation model into the control model of the simulation model, and input the power loss value sequence. The reference boundary temperature measured by a physical temperature sensor on the power module housing or cooling system piping. In the input control model, the control model discretizes the junction temperature estimation model, calculates the temperature of all nodes in the heat conduction path at the current moment, and obtains the real-time junction temperature time series of each power semiconductor device in the power module. .

[0034] Further, step S4 includes:

[0035] S41: Rainflow counting method is used to analyze the real-time junction temperature time series. Extracting each complete thermal cycle event

[0036] Item k The basic parameters are used to construct key feature vectors for characterizing multi-physical damage mechanisms. :

[0037] ;

[0038] in, For the first k The junction temperature fluctuation amplitude of a thermal cycling event. For the first kAverage junction temperature of one thermal cycling event , For the first k The maximum rate of change of junction temperature during a thermal cycling event is used to quantify the severity of thermal shock. For the first k The duration of a thermal cycling event near its peak temperature;

[0039] S42: Utilizing junction temperature fluctuation amplitude and the k Average junction temperature of one thermal cycling event Calculate the number of failure cycles under the current loop condition. Then utilize the number of failed loops Calculate the first k Incremental fatigue damage caused by a thermal cycling event ;

[0040] S43: Utilizing the maximum rate of change Calculate the first k Increment of thermal shock damage caused by a thermal cycling event ;

[0041] S44: Utilizing Duration and average junction temperature Calculate the first k Increment of high-temperature creep damage caused by a thermal cycling event ;

[0042] S45: Increase fatigue damage Incremental thermal shock damage and the increase in high-temperature creep damage Perform weight fusion and update the first k Total cumulative damage to the power module after one thermal cycle event The updated total cumulative damage is obtained. ;

[0043] S46: Based on the updated total cumulative damage Calculate the remaining lifespan of the power module RUL .

[0044] Further, step S5 includes:

[0045] S51: Fatigue Damage Increment Incremental thermal shock damage and the increase in high-temperature creep damage The output power module is in the first k Real-time fatigue damage after a thermal cycling event Thermal shock damage and high temperature creep damage Calculate the instantaneous cumulative rate of fatigue damage, thermal shock damage, and high-temperature creep damage;

[0046] S52: Constructing a fatigue damage-based system Thermal shock damage and high temperature creep damage Short-term lifespan consumption prediction function Predict a preset time window in the future Total cumulative damage to internal power modules ;

[0047] S53: Set the threshold for damage rate ,like or If the power module is in the initial stage of a period of intense acceleration or deceleration or prolonged high load on the electric vehicle, it will send a command to the cooling system of the electric vehicle to improve the cooling effect; otherwise, the power module load will be considered normal.

[0048] Set a moderate damage warning threshold ,like If the power module is under unacceptable rapid lifespan depletion conditions, a torque request limit command is sent to the vehicle controller of the electric vehicle to reduce power loss, and the cooling system operates at maximum power; otherwise, the power module load is normal.

[0049] Set a high damage risk threshold ,like If the total accumulated damage to the power module is about to become excessive, a power limiting command is sent to the vehicle controller of the electric vehicle to reduce the output power, and a system protection warning is displayed to the driver through the instrument panel, while the cooling system operates at maximum power; otherwise, the power module load is normal.

[0050] The beneficial effects of this invention are as follows: By constructing a high-fidelity model chain that accurately reflects the thermal coupling effect, the invention obtains the true internal temperature field of the power module; then, based on this accurate data, an innovative multi-physics mechanism coupled damage model is adopted to comprehensively and quantitatively evaluate fatigue, thermal shock, and creep. Finally, based on this multi-dimensional and accurate state assessment, a set of forward-looking hierarchical feedback control logic based on damage trend prediction is established.

[0051] This invention accurately calculates the instantaneous power loss of each independent chip under real PWM waveform driving and precisely captures the temperature interaction between chips caused by thermal crosstalk by solving the three-dimensional heat conduction equation. This ensures that the junction temperature data obtained by this invention is no longer a vague average value, but a high-resolution spatiotemporal sequence reflecting the real physical process. This precision fundamentally solves the problem of inaccurate lifetime prediction caused by information distortion in traditional models. Because the accuracy of all subsequent lifetime calculations depends on the authenticity of the input junction temperature data. This invention provides an unprecedented high-confidence data foundation for reliability assessment, significantly improving the prediction accuracy of the entire method.

[0052] This invention constructs a unified framework for multi-physical damage calculation, unifying fatigue damage caused by thermal cycling amplitude, thermal shock damage caused by junction temperature changes, and creep damage caused by prolonged high temperatures within a dynamically weighted mathematical framework for the first time. This multi-dimensional damage assessment system can identify and quantify the dominant failure mechanisms under different operating conditions. For example, under frequent start-stop conditions in urban environments, the system can identify accelerated accumulation of thermal shock damage; under long-term uphill conditions in mountainous areas, the system can identify the significant impact of creep damage. This allows the invention to detect potential risks that traditional models cannot uncover, thus making a far more comprehensive and profound judgment on the true health status of power modules than existing technologies. The "predictive hierarchical control logic" constructed in this invention is based on the prediction of future short-term damage trends and real-time analysis of the current damage rate, intervening in advance before damage causes serious consequences. Its three-level control strategy of "preventive fine-tuning," "active intervention," and "protective limitation" achieves a seamless transition from imperceptible smooth adjustment to mandatory protection when necessary; it can achieve maximum lifespan extension with minimal performance cost.

[0053] By effectively mitigating the accumulation of chronic damage during daily operation, the actual service life of the power module is significantly extended, reducing the risk of vehicle breakdown due to electronic control system failure, and comprehensively improving the long-term reliability of the entire vehicle.

[0054] This invention provides proactive lifespan management capabilities, eliminating the need for OEMs to select excessively redundant (i.e., more expensive) power modules or cooling systems during the design phase to address uncertainties. It allows for the selection of more cost-effective hardware while maintaining equal or even higher reliability, directly reducing bill of materials (BOM) costs. This resolves the core conflict between reliability and cost, providing crucial technological support for the widespread adoption of new energy vehicles. Attached Figure Description

[0055] Figure 1 This is a flowchart of an intelligent thermal management method for the power module of an electric vehicle. Detailed Implementation

[0056] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0057] like Figure 1 As shown, an intelligent thermal management method for an electric vehicle power module includes the following steps:

[0058] S1: Construct a simulation model of the electric vehicle drive system in a physical simulation platform to reproduce the operation and control logic of the electric vehicle. Output the conduction and switching losses of each power semiconductor device in the power module to obtain a power loss value sequence for each power semiconductor device. Step S1 specifically includes:

[0059] S11: Construct a simulation model of the electric vehicle drive system in the physical simulation platform. The simulation model includes a control model, a modulator and inverter model, and a motor model. The simulation model is used to reproduce the operation control logic of the electric vehicle and the conduction loss and switching loss of each power semiconductor device in the output power module.

[0060] This embodiment builds a high-fidelity simulation model of an electric vehicle drive system within a system-level multi-domain physical simulation platform. This model accurately reproduces the actual operation and control logic of the electric vehicle, providing precise electrical parameter inputs for real-time power loss calculation. The simulation model mainly includes the following core components:

[0061] Controller Model: Implements the electric vehicle motor control strategy and incorporates a field-oriented control algorithm. It receives commands (such as target torque and speed) from the upper-level vehicle controller (VCU) and combines them with real-time feedback signals from the motor model (such as the dq-axis current transformed by the abc2dq module). Through a PI closed-loop regulator, it generates the voltage commands (ud, uq) required to control the inverter.

[0062] Modulator and inverter model: Includes a pulse width modulation module and a space vector pulse width modulation (SVPWM) module, which calculates in real time the gate drive signal or equivalent duty cycle signal of each power semiconductor switching device in the power module based on the voltage command output by the control module. D u , D v , D w Meanwhile, the modulator and inverter model also includes a topology circuit representing a three-phase bridge inverter.

[0063] Motor model: A mathematical model that accurately characterizes the dynamic behavior of permanent magnet synchronous motors or other types of motors, used to simulate the electromagnetic response of the motor under inverter output voltage drive, and to provide phase current ( i abc Key feedback signals such as rotational speed.

[0064] The output method for conduction loss is as follows:

[0065] The output characteristic curves (i.e., the relationship between on-state voltage drop and current and temperature) provided in the datasheets of the power semiconductor devices (IGBTs and freewheeling diodes) in the power module are embedded into the simulation model in the form of a lookup table. The simulation model obtains the instantaneous current flowing through each power semiconductor device in real time. And duty cycle signal;

[0066] The instantaneous on-state voltage drop corresponding to the calculated power semiconductor device can be queried. Based on instantaneous on-state pressure drop and instantaneous current Calculate instantaneous conduction power ;

[0067] In a switching cycle T s Internal instantaneous conduction power Integrate and average to obtain the conduction loss of each power semiconductor device.

[0068] The method for outputting switching losses is as follows:

[0069] The switching energy loss curves (i.e., the relationship curves between single turn-on / turn-off energy and current, bus voltage, and temperature) provided in the datasheets of the power semiconductor devices in the power module are embedded into the simulation model in the form of a lookup table.

[0070] The simulation model is based on the real-time electrical boundary conditions (including DC bus voltage) at the occurrence of each switching event. V bus collector current of the switching transistor i c (etc.), query the calculated single activation energy E on and shutting off energy E off ;

[0071] Energy will be activated only once. E on and shutting off energy E off Multiply by switching frequency Thus, the switching losses are obtained.

[0072] Using the above method, this invention can accurately and in real-time analyze the conduction and switching losses of each independent power semiconductor device (six IGBTs and six freewheeling diodes in the three-phase bridge arm) within the power module under specific control strategies and dynamic operating conditions. Ultimately, the model outputs the total power loss of each device, providing a high-resolution, high-precision distributed heat source input for subsequent junction temperature estimation. Its accuracy far surpasses offline estimation methods based on empirical formulas or average current.

[0073] S12: Summate the conduction loss and switching loss of each power semiconductor device to obtain the power loss of each power semiconductor device, and output the power loss value sequence of each power semiconductor device in real time. , i This is the designation for power semiconductor devices. t For time.

[0074] S2: Construct a finite element model of the power module, obtain the three-dimensional partial differential equation of heat conduction characterizing the transient thermal behavior, apply a unit power step to the finite element model, output the thermal impedance curve of the power semiconductor device, and calculate the thermal resistance of the power semiconductor device. Step S2 specifically includes:

[0075] S21: Construct a finite element model of the power module in terms of geometry and material properties using finite element simulation software in a three-dimensional multiphysics simulation environment; assign corresponding thermophysical parameters (density) to each component in the finite element model. Specific heat capacity and thermal conductivity wait);

[0076] The finite element model includes all the key components of the power module, including power semiconductor devices (semiconductor chips), multilayer solder, direct copper-clad ceramic substrate (DBC), copper substrate, and thermal interface materials.

[0077] S22: Constructing three-dimensional partial differential equations of heat conduction to characterize the transient thermal behavior of the finite element model:

[0078] ;

[0079] in, The density of the component material in the finite element model. The specific heat capacity of the component material, K The thermal conductivity of the component material. T Spatial coordinates in the finite element model With time t The function, Q The heat source density output by the simulation model;

[0080] Heat source densityQ Heat source density output from simulation model Q It is applied as a distributed heat source, meaning that each semiconductor chip within the power module is defined as an independent heat source with power varying over time.

[0081] S23: Apply a unit power step to one of the power semiconductor devices in the finite element model. P i =1W), the power of other power semiconductor devices is zero; record all N The temperature of a power semiconductor device over time t The rising curve is used as the thermal impedance curve of power semiconductor devices.

[0082] S24: Use the diagonal element of the thermal impedance curve as the self-thermal impedance curve of the power semiconductor device. Off-diagonal elements as coupling thermal resistance curves between power semiconductor devices , j For power semiconductor devices i Different power semiconductor device designations;

[0083] Diagonal elements represent power semiconductor devices i The self-heating resistance describes the temperature rise process caused by self-heating. Off-diagonal elements (when...) i ≠ j (Time) represents power semiconductor devices i With power semiconductor devices j The coupling thermal impedance between them, which precisely quantifies the power semiconductor device. i As a heat source for nearby power semiconductor devices j The degree of crosstalk impact.

[0084] The self-thermal impedance curve of each power semiconductor device The thermal resistance is fitted to a self-heating RC chain composed of multiple RC units connected in series, thus obtaining the thermal resistance characterizing the main heat dissipation path from the power semiconductor device to the heat sink. and heat capacity , k Number the thermal cycling events;

[0085] In this embodiment, the self-heating resistance curve The fitting process specifically includes:

[0086] Based on thermoelectric simulation theory, the transient thermal impedance of power semiconductor devices It can be equivalently described by the linear superposition of multiple exponential functions (i.e., the Foster thermal network model). And for the self-heating resistance curve... Its fitting function expression is:

[0087] ;

[0088] in, n The fitting order is... R k For the first k thermal resistance coefficient of order, For the first k The thermal time constant of the order, For the first k The heat of the stage.

[0089] The fitting process employs the nonlinear least squares method, by constructing an objective function. J This allows the discrete temperature rise response data points obtained from finite element simulation to be displayed. Minimize the mean square error between the calculated value and the fitted function:

[0090] ;

[0091] in, M The number of sampling points. m Self-heating resistance curve Sampling points on, For sampling point time, Sampling points m The corresponding self-heating resistance, Sampling point time The corresponding discrete temperature rise response, The applied unit step power.

[0092] Based on the objective function J By iteratively solving the fitted function, the parameters of the Foster model can be identified. This allows us to obtain the thermal resistance characterizing the main heat dissipation path from the power semiconductor device to the heat sink. and heat capacity .

[0093] The coupling thermal impedance curve of each power semiconductor device By performing fitting, characteristics of power semiconductor devices can be identified. i With power semiconductor devices j Coupling thermal resistance of thermal coupling strength between .

[0094] In this embodiment, the coupling thermal resistance is identified. The specific method is as follows:

[0095] For coupling thermal resistance The identification is based on the principle of steady-state heat conduction. The off-diagonal element curves obtained using finite element simulation are used. Extract its steady-state value when it reaches thermal equilibrium. This steady-state coupling thermal impedance is the characteristic of the power semiconductor device in the multi-port thermal network formed by the device. i With power semiconductor devices j The direct thermal coupling strength between them will affect the steady-state value. As a coupling thermal resistance Embedded into two self-heating RC Between corresponding nodes in the chain.

[0096] S3: Construct a coupled network model reflecting all heat conduction paths within the power module using thermal resistance, reduce the order of the three-dimensional heat conduction partial differential equation to obtain a junction temperature estimation model, and output a real-time junction temperature time series based on the power loss value sequence. Step S3 specifically includes:

[0097] S31: Each power semiconductor device corresponds to a vertical self-heating RC chain, and different power semiconductor devices are coupled through thermal resistance. Horizontal connections yield a coupled network model within the power module that reflects all heat conduction paths;

[0098] S32: Based on the coupled network model, the three-dimensional heat conduction partial differential equations are reduced in order to obtain a set of first-order ordinary differential equations characterizing the transient thermal behavior of the coupled network model. This set of first-order ordinary differential equations serves as the junction temperature estimation model. The matrix form of the set of first-order ordinary differential equations is as follows:

[0099] ;

[0100] in, Let be the column vector of temperatures of all nodes in the heat conduction path in the coupled network model. This is the input vector for the real-time power loss of each power semiconductor device. This is a diagonal matrix formed by the heat capacity in the coupled network model. The conductance matrix is ​​formed by the reciprocals of the coupling thermal resistance in the coupled network model;

[0101] S33: Input the junction temperature estimation model into the control model of the simulation model, and input the power loss value sequence. The reference boundary temperature measured by a physical temperature sensor on the power module housing or cooling system piping. In the input control model, the control model discretizes the junction temperature estimation model, calculates the temperature of all nodes in the heat conduction path at the current moment, and obtains the real-time junction temperature time series of each power semiconductor device in the power module. .

[0102] Within each sampling period, the control model calculates the temperature of all nodes at the current moment through a single, rapid matrix operation, based on the real-time power loss input of each power semiconductor device and the node temperature of the previous moment. This yields a high-resolution, high-precision junction temperature time series for each power semiconductor device, forming the core data foundation of the intelligent thermal management method of this invention, and serving as the direct input to the lifetime prediction model in the next stage.

[0103] S4: Extract key feature vectors characterizing multi-physical damage mechanisms from the real-time junction temperature time series, calculate the fatigue damage increment, thermal shock damage increment, and high-temperature creep damage increment caused by thermal cycling events, fuse them, update the total cumulative damage after thermal cycling events, and calculate the remaining service life of the power module. Step S4 specifically includes:

[0104] S41: Rainflow counting method is used to analyze the real-time junction temperature time series. Extract each complete thermal cycle

[0105] event k The basic parameters are used to construct key feature vectors for characterizing multi-physical damage mechanisms. :

[0106] ;

[0107] in, For the first k The junction temperature fluctuation amplitude of a thermal cycling event. For the first k Average junction temperature of one thermal cycling event , For the first k The maximum rate of change of junction temperature during a thermal cycling event is used to quantify the severity of thermal shock. For the first k The duration of a thermal cycling event near its peak temperature;

[0108] S42: Utilizing junction temperature fluctuation amplitude and the k Average junction temperature of one thermal cycling event Calculate the number of failure cycles under the current loop condition. Then utilize the number of failed loops Calculate the first k Incremental fatigue damage caused by a thermal cycling event ;

[0109] ;

[0110] ;

[0111] in, , The fatigue characteristic coefficient of the packaging material. For activation energy, Boltzmann constant; fatigue damage increment It is caused by the propagation of microcracks in the material under cyclic thermal strain.

[0112] S43: Utilizing the maximum rate of change Calculate the first k Increment of thermal shock damage caused by a thermal cycling event ;

[0113] ;

[0114] in, , The thermal shock damage coefficient, calibrated experimentally, characterizes the material's sensitivity to the rate of temperature change. Thermal shock damage increment. This damage is caused by instantaneous mechanical stress generated under rapid temperature changes due to the mismatch in the coefficients of thermal expansion (CTE) of different materials within the power module. The incremental damage is proportional to a power of the junction temperature change rate. S44: Utilizing duration and average junction temperature Calculate the first k Increment of high-temperature creep damage caused by a thermal cycling event ;

[0115] ;

[0116] in, The creep damage coefficient is... The activation energy of the creep process; the increment of high-temperature creep damage. Caused by plastic deformation and lattice diffusion of materials such as solder under prolonged stress at high temperatures, this invention addresses the increase in high-temperature creep damage. With duration Average junction temperature Related.

[0117] S45: Increase fatigue damage Incremental thermal shock damage and the increase in high-temperature creep damage Perform weight fusion and update the first k Total cumulative damage to the power module after one thermal cycle event The updated total cumulative damage is obtained. ;

[0118] ;

[0119] in, This represents the total cumulative damage before the update. , , These are the weighting factors for fatigue damage, thermal shock damage, and high-temperature creep damage, respectively, and they satisfy the following conditions: .

[0120] Weighting factors , , Values ​​can be assigned to power modules at different aging stages; for example, in the early stages of power module use, fatigue damage is dominant (weighting factor). The value is relatively large, so it can be taken as follows: As damage accumulates, the material degrades and becomes sensitive to thermal shock and creep (weighting factor). , The value increases, at which point it can be taken. ).

[0121] S46: Based on the updated total cumulative damage Calculate the remaining lifespan of the power module RUL ;

[0122] ;

[0123] in, Rated lifespan designed for power modules.

[0124] S5: Calculate the instantaneous accumulation rate of fatigue damage, thermal shock damage, and high-temperature creep damage in the power module using the fatigue damage increment, thermal shock damage increment, and high-temperature creep damage increment. Predict the total cumulative damage of the power module in the future. Based on the predicted total cumulative damage and instantaneous accumulation rate, assess whether the power module's operating condition is normal and perform thermal management of the power module. Step S5 specifically includes:

[0125] S51: Based on fatigue damage increment Incremental thermal shock damage and the increase in high-temperature creep damage The output power module is in the first k Real-time fatigue damage after a thermal cycling event Thermal shock damage and high temperature creep damage Calculate the instantaneous cumulative rate of fatigue damage, thermal shock damage, and high-temperature creep damage;

[0126] , , ;

[0127] in, , , These represent the instantaneous cumulative rates of fatigue damage, thermal shock damage, and high-temperature creep damage, respectively.

[0128] S52: Constructing a fatigue damage-based system Thermal shock damage and high temperature creep damage Short-term lifespan consumption prediction function Used to characterize the overall predicted damage rate:

[0129] ;

[0130] in, , , The weighting coefficients for the instantaneous cumulative rates of fatigue damage, thermal shock damage, and high-temperature creep damage, respectively, can be determined by conducting accelerated life tests on the power module and analyzing the contribution of each damage mechanism to the total lifespan consumption under different operating conditions.

[0131] Predict a preset time window in the future Total cumulative damage to internal power modules ;

[0132] ;

[0133] S53: Set the threshold for damage rate ,like or If the power module is in the initial stage of a period of intense acceleration or deceleration or prolonged high load on the electric vehicle, it will send a command to the cooling system of the electric vehicle to improve the cooling effect; otherwise, the power module load will be considered normal.

[0134] Set a moderate damage warning threshold ,like If the power module is under unacceptable rapid lifespan depletion conditions, a torque request limit command is sent to the vehicle controller of the electric vehicle to reduce power loss, and the cooling system operates at maximum power; otherwise, the power module load is normal.

[0135] Set a high damage risk threshold ,like If the total accumulated damage to the power module is about to become excessive, a power limiting command is sent to the vehicle controller of the electric vehicle to reduce the output power, and a system protection warning is displayed to the driver through the instrument panel, while the cooling system operates at maximum power; otherwise, the power module load is normal.

[0136] This invention establishes feedback control logic through cumulative damage prediction to achieve closed-loop control of the entire intelligent thermal management system. It constructs a predictive hierarchical control architecture based on multi-dimensional damage state perception. This architecture no longer relies on passive responses to the current junction temperature or a single cumulative damage value, but actively analyzes multiple damage increments output from the previous stage. , , The real-time rate, combined with the total cumulative damage. It can predict the lifespan consumption trend in the near future, thereby enabling the graded and precise execution of adjustment instructions before significant damage accumulates.

[0137] This invention defines a multi-level control strategy, with each level corresponding to a different set of damage state thresholds and a specific set of adjustment instructions. This hierarchical mechanism replaces the traditional single "alarm-execution" mode, achieving smoother and more refined intervention in the system.

[0138] When the triggering conditions at any level are met, the control module generates the corresponding digital command signal and sends it to the vehicle controller (VCU) and cooling system via the vehicle's internal bus (such as the CAN bus).

[0139] The execution status of control commands forms a closed loop, with the control module continuously monitoring the damage rate and predicting damage accumulation. Once these indicators fall below the safety threshold, indicating that the operating condition has been mitigated or adjustment measures have taken effect, the system will automatically and smoothly revert from a high-level control state to a low-level or normal state, thereby maximizing the recovery of vehicle performance while ensuring the lifespan of the power module. Through predictive hierarchical control logic, this invention elevates the health management of the power module from "post-event remediation" to a new level of "prevention" and "precise control during operation."

Claims

1. A method for intelligent thermal management of an electric vehicle power module, characterized in that, Includes the following steps: S1: Construct a simulation model of the electric vehicle drive system in the physical simulation platform to reproduce the operation control logic of the electric vehicle, output the conduction loss and switching loss of each power semiconductor device in the power module, and obtain the power loss value sequence of each power semiconductor device. S2: Construct a finite element model of the power module, obtain the three-dimensional partial differential equation of heat conduction characterizing transient thermal behavior, apply a unit power step to the finite element model, output the thermal impedance curve of the power semiconductor device, and calculate the thermal resistance of the power semiconductor device. Step S2 includes: S21: Construct a finite element model of the power module in terms of geometry and material properties using finite element simulation software in a three-dimensional multiphysics simulation environment; assign corresponding thermal property parameters to each component in the finite element model. S22: Constructing three-dimensional partial differential equations of heat conduction to characterize the transient thermal behavior of the finite element model: ； in, The density of the component material in the finite element model. The specific heat capacity of the component material, K The thermal conductivity of the component material. T Spatial coordinates in the finite element model With time t The function, Q The heat source density output by the simulation model; S23: Apply a unit power step to one of the power semiconductor devices in the finite element model, while the power of the other power semiconductor devices is zero; record all... N The temperature of a power semiconductor device over time t The rising curve is used as the thermal impedance curve of power semiconductor devices. S24: Use the diagonal element of the thermal impedance curve as the self-thermal impedance curve of the power semiconductor device. Off-diagonal elements as coupling thermal resistance curves between power semiconductor devices , j For power semiconductor devices i Different power semiconductor device designations; Diagonal elements represent power semiconductor devices i The self-thermal impedance describes the temperature rise process caused by self-heating; off-diagonal elements represent power semiconductor devices. i With power semiconductor devices j The coupling thermal impedance between them quantifies the power semiconductor device. i As a heat source for nearby power semiconductor devices j The degree of crosstalk impact; The self-thermal impedance curve of each power semiconductor device The thermal resistance is fitted to a self-heating RC chain composed of multiple RC units connected in series, thus obtaining the thermal resistance characterizing the main heat dissipation path from the power semiconductor device to the heat sink. and heat capacity , k Number the thermal cycling events; The coupling thermal impedance curve of each power semiconductor device By performing fitting, characteristics of power semiconductor devices can be identified. i With power semiconductor devices j Coupling thermal resistance of thermal coupling strength between ; S3: Construct a coupled network model within the power module that reflects all heat conduction paths using thermal resistance, reduce the order of the three-dimensional heat conduction partial differential equation to obtain a junction temperature estimation model, and output a real-time junction temperature time series based on the power loss value sequence. Step S3 includes: S31: Each power semiconductor device corresponds to a vertical self-heating RC chain, and different power semiconductor devices are coupled through thermal resistance. Horizontal connections yield a coupled network model within the power module that reflects all heat conduction paths; S32: Based on the coupled network model, the three-dimensional heat conduction partial differential equation is reduced in order to obtain a set of first-order ordinary differential equations characterizing the transient thermal behavior of the coupled network model, which serves as a junction temperature estimation model. The matrix form of the first-order ordinary differential equation system is: ； in, Let be the column vector of temperatures of all nodes in the heat conduction path in the coupled network model. This is the input vector for the real-time power loss of each power semiconductor device. This is a diagonal matrix formed by the heat capacity in the coupled network model. The conductance matrix is ​​formed by the reciprocals of the coupling thermal resistance in the coupled network model; S33: Input the junction temperature estimation model into the control model of the simulation model, and input the power loss value sequence. The reference boundary temperature measured by a physical temperature sensor on the power module housing or cooling system piping. In the input control model, the control model discretizes the junction temperature estimation model, calculates the temperature of all nodes in the heat conduction path at the current moment, and obtains the real-time junction temperature time series of each power semiconductor device in the power module. ; Within each sampling period, the control model calculates the temperature of all nodes at the current moment through a fast matrix operation based on the real-time power loss input of each power semiconductor device and the node temperature at the previous moment, thus obtaining the junction temperature time series of each power semiconductor device. S4: Extract key feature vectors characterizing multi-physical damage mechanisms from the real-time junction temperature time series, calculate the fatigue damage increment, thermal shock damage increment, and high-temperature creep damage increment caused by thermal cycling events, fuse them, update the total cumulative damage after thermal cycling events, and calculate the remaining service life of the power module. Step S4 includes: S41: Rainflow counting method is used to analyze the real-time junction temperature time series. Extracting each complete thermal cycle event Item k The basic parameters are used to construct key feature vectors for characterizing multi-physical damage mechanisms. : ； in, For the first k The junction temperature fluctuation amplitude of a thermal cycling event. For the first k Average junction temperature of one thermal cycling event , For the first k The maximum rate of change of junction temperature during a thermal cycling event is used to quantify the severity of thermal shock. For the first k The duration of a thermal cycling event near its peak temperature; S42: Utilizing junction temperature fluctuation amplitude and the k Average junction temperature of one thermal cycling event Calculate the number of failure cycles under the current loop condition. Then utilize the number of failed loops Calculate the first k Incremental fatigue damage caused by a thermal cycling event ; ； ； in, , The fatigue characteristic coefficient of the packaging material. For activation energy, Boltzmann's constant; S43: Utilizing the maximum rate of change Calculate the first k Increment of thermal shock damage caused by a thermal cycling event ; ； in, , The thermal shock damage coefficient, calibrated experimentally, characterizes the material's sensitivity to the rate of temperature change. S44: Utilizing Duration and average junction temperature Calculate the first k Increment of high-temperature creep damage caused by a thermal cycling event ; ； in, The creep damage coefficient is... The activation energy for the creep process; S45: Increase fatigue damage Incremental thermal shock damage and the increase in high-temperature creep damage Perform weight fusion and update the first k Total cumulative damage to the power module after one thermal cycle event The updated total cumulative damage is obtained. ; ； in, This represents the total cumulative damage before the update. , , These are the weighting factors for fatigue damage, thermal shock damage, and high-temperature creep damage, respectively, and they satisfy the following conditions: ; S46: Based on the updated total cumulative damage Calculate the remaining lifespan of the power module RUL ; ； in, Rated lifespan designed for power modules; S5: Calculate the instantaneous accumulation rate of fatigue damage, thermal shock damage, and high-temperature creep damage of the power module using fatigue damage increment, thermal shock damage increment, and high-temperature creep damage increment. Predict the total cumulative damage of the power module in the future. Based on the predicted total cumulative damage and instantaneous accumulation rate, assess whether the operating condition of the power module is normal and perform thermal management of the power module.

2. The intelligent thermal management method for an electric vehicle power module according to claim 1, characterized in that, Step S1 includes: S11: Construct a simulation model of the electric vehicle drive system in the physical simulation platform. The simulation model includes a control model, a modulator and inverter model, and a motor model. The simulation model is used to reproduce the operation control logic of the electric vehicle and the conduction loss and switching loss of each power semiconductor device in the output power module. S12: Summate the conduction loss and switching loss of each power semiconductor device to obtain the power loss of each power semiconductor device, and output the power loss value sequence of each power semiconductor device in real time. , i This is the designation for power semiconductor devices. t For time.

3. The intelligent thermal management method for an electric vehicle power module according to claim 2, characterized in that, The method for outputting the conduction loss is as follows: The output characteristic curves provided in the datasheets of the power semiconductor devices in the power module are embedded into the simulation model in the form of a lookup table. The simulation model acquires the instantaneous current flowing through each power semiconductor device in real time. And duty cycle signal; The instantaneous on-state voltage drop corresponding to the calculated power semiconductor device can be queried. Based on instantaneous on-state pressure drop and instantaneous current Calculate instantaneous conduction power ; In a switching cycle T s Internal instantaneous conduction power Integrate and average to obtain the conduction loss of each power semiconductor device.

4. The intelligent thermal management method for an electric vehicle power module according to claim 3, characterized in that, The method for outputting the switching loss is as follows: The switching energy loss curves provided in the datasheets of the power semiconductor devices in the power module are embedded into the simulation model in the form of a lookup table; The simulation model queries and calculates the single-turn-on energy based on the real-time electrical boundary conditions at the time of each switching event. E on and shutting off energy E off ; Energy will be activated only once. E on and shutting off energy E off Multiply by switching frequency Thus, the switching losses are obtained.

5. The intelligent thermal management method for an electric vehicle power module according to claim 4, characterized in that, Step S5 includes: S51: Fatigue Damage Increment Incremental thermal shock damage and the increase in high-temperature creep damage The output power module is in the first k Real-time fatigue damage after a thermal cycling event Thermal shock damage and high temperature creep damage Calculate the instantaneous cumulative rate of fatigue damage, thermal shock damage, and high-temperature creep damage; S52: Constructing a fatigue damage-based system Thermal shock damage and high temperature creep damage Short-term lifespan consumption prediction function Predict a preset time window in the future Total cumulative damage to internal power modules ; in, , , These represent the instantaneous cumulative rates of fatigue damage, thermal shock damage, and high-temperature creep damage, respectively. S53: Set the threshold for damage rate ,like or If the power module is in the initial stage of a period of intense acceleration or deceleration or prolonged high load on the electric vehicle, it will send a command to the cooling system of the electric vehicle to improve the cooling effect; otherwise, the power module load will be considered normal. Set a moderate damage warning threshold ,like If the power module is under unacceptable rapid lifespan depletion conditions, a torque request limit command is sent to the vehicle controller of the electric vehicle to reduce power loss, and the cooling system operates at maximum power; otherwise, the power module load is normal. Set a high damage risk threshold ,like If the total accumulated damage to the power module is about to become excessive, a power limiting command is sent to the vehicle controller of the electric vehicle to reduce the output power, and a system protection warning is displayed to the driver through the instrument panel, while the cooling system operates at maximum power; otherwise, the power module load is normal.

Images