Automobile cabin temperature field simulation analysis method and system based on multi-physics field coupling

Abstract

The invention relates to the technical field of automobile material temperature simulation analysis, in particular to an automobile cabin temperature field simulation analysis method and system based on multi-physics field coupling. According to the method, high-precision and high-efficiency simulation is realized through thermal-solid-fluid-electric coupling modeling in combination with the genetic algorithm, the neural network and reinforcement learning AI dynamic optimization, and the problems that the multi-physical field coupling effect cannot be comprehensively reflected and the parameter optimization efficiency is low in the prior art are solved. Time sequence analysis and dynamic boundary condition adjustment are introduced, transient working condition simulation is supported, and temperature field changes in actual driving can be better predicted. Through simultaneous solution of the Maxwell equation and the heat conduction equation, the temperature rise risk of the electronic component is accurately predicted, and the method is suitable for heat management optimization of key components such as a battery pack and a motor controller of a new energy automobile. Cloud high-performance calculation and edge end lightweight model cooperation are adopted, efficient simulation and real-time early warning are achieved, and rapid analysis and design optimization of extreme working conditions are supported.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile material temperature simulation analysis, and particularly relates to a vehicle cabin temperature field simulation analysis method and system based on multi-physical field coupling. BACKGROUND

[0002] The temperature field distribution in the vehicle cabin has an important influence on the performance of the engine, electronic components and cooling system. Traditional simulation methods mostly use single physical field of heat conduction or fluid dynamics for analysis, which is difficult to fully reflect the complex multi-physical field coupling effect in the cabin. In the prior art, the main concern is single component or single physical field, and the coupling effect of multiple components and multiple physical fields in the cabin is not fully considered, resulting in insufficient simulation accuracy and difficulty in fully predicting the temperature distribution and thermal failure risk in the cabin. In the prior art, multi-physical field coupling simulation has been partially applied, but lacks real-time response capability for transient conditions, and model calibration relies on manual experience. In addition, the existing method does not combine machine learning technology to optimize parameters, resulting in low calculation efficiency and difficulty in adapting to complex dynamic environments.

[0003] Chinese patent document CN112231826A discloses "a one-dimensional fuel vehicle whole vehicle thermal management simulation analysis method based on GT-SUIT", which calculates the change of engine water outlet temperature and flow with time through four steps of building a simulation model, inputting performance parameters, inputting boundary parameters, solving and post-processing, and evaluates according to relevant requirements. The disadvantages of this method are: first, it does not involve multi-physical field coupling modeling method, which cannot accurately simulate the thermal solid flow and electric interaction effect in the cabin; second, it cannot realize automatic parameter calibration and high-efficiency model convergence; third, it does not have a cloud-edge collaborative architecture and does not support large-scale simulation and real-time early warning; fourth, the simulation coverage is limited and cannot perform real-time and rapid simulation on transient extreme conditions. SUMMARY

[0004] Technical purpose: In order to overcome the deficiencies in the prior art, the present application provides a vehicle cabin temperature field simulation analysis method and system based on multi-physical field coupling, to solve the problems in the prior art that cannot accurately simulate the thermal solid flow and electric interaction effect in the cabin, cannot realize automatic parameter calibration and high-efficiency model convergence, and cannot perform real-time and rapid simulation on transient extreme conditions.

[0005] In view of the deficiencies of the prior art, one of the purposes of the present application is to provide a vehicle cabin temperature field simulation analysis method based on multi-physical field coupling, characterized in that it comprises a modeling process, a simulation process and a linking link of the modeling process and the simulation process; the modeling process is completed on a local client, comprising [S1] multi-source data acquisition and preprocessing, [S2] multi-physical field coupling modeling, [S3] AI parameter optimization and model calibration, after local modeling is completed, model data is uploaded to the cloud; the simulation process is completed on a cloud server, comprising [S4] cloud-edge collaborative simulation and result output, the simulation result is returned to the local or edge device; the linking link of the modeling process and the simulation process comprises [S5] dynamic feedback and continuous optimization, which is the parameter calibration of the modeling process and the iterative improvement of the simulation result; [S1] multi-source data acquisition and preprocessing, comprising: [S1.1] multi-source data acquisition, comprising static data acquisition, dynamic data acquisition and temperature data acquisition; the static data acquisition comprises CAD models, thermal conductivity, specific heat capacity and density of the engine, ECU and battery pack, which are obtained through a component material database; the dynamic data refers to dynamic working condition data collected in real time through a vehicle sensor network and a CAN bus, comprising vehicle speed, ambient temperature and motor load; the temperature data is collected through a thermocouple and an infrared thermometer, comprising surface temperatures of the cooling system of the automobile cabin internal components, the exhaust system and electronic components; [S1.2] AI preprocessing, comprising data cleaning, short-term thermal load prediction and feature extraction; [S1.2.1] data cleaning refers to automatically removing outliers from the multi-source data collected in step [S1.1] by using an isolation forest machine learning algorithm; [S1.2.2] short-term thermal load prediction refers to performing time series analysis on the dynamic working condition data in the result data of step [S1.2.1] by using a long short-term memory network (LSTM) to predict the thermal load change trend in the next 5-10 minutes; [S1.2.3] feature extraction refers to performing data dimensionality reduction on the result data of step [S1.2.2] by using principal component analysis (PCA) to extract key features and reduce computational complexity; [S2] multi-physical field coupling modeling, comprising thermal-solid-flow coupling modeling, thermal-electric coupling modeling and boundary condition dynamic binding; [S2.1] thermal-solid-flow coupling modeling, comprising: [S2.1.1] heat conduction model: based on Fourier's law, a heat conduction equation of the automobile cabin internal components is established, and the solution is a temperature distribution T 热固流 as shown in formula 1; Formula 1; in: p For density, c p is the specific heat capacity, T 热固流 is the temperature distribution, t For time, k is the thermal conductivity, Q source is the heat source power; [S2.1.2] Fluid dynamics model: Computational fluid dynamics (CFD) is used to simulate the air flow in the cabin and solve the Navier-Stokes equations. The solution is the flow velocity. u and pressure p , as shown in formula 2; Formula 2; in: u is the flow rate, t For time, p is the density, p For pressure, μ is the dynamic viscosity, g For external forces; [S2.1.3] Bidirectional fluid-structure interaction (FSI): By iteratively solving the coupled equations for fluid flow and solid heat conduction, the boundary conditions are updated to ensure accurate heat transfer between the fluid and the solid, as shown in Equation 3. Formula 3; in, F fluid is the force exerted by the fluid on the solid, which is calculated by the fluid dynamics equation Navier-Stokes equation; F solid is the reaction force of the solid on the fluid, which is calculated by the solid mechanics equation and the thermoelastic equation; Formula 3 transfers heat through the coupling interface , the temperature distribution calculated by formula 1 T The flow rate calculated by formula 2 is u and pressure As input, the boundary conditions are dynamically adjusted and updated according to the coupling calculation results to ensure the continuity of heat transfer; [S2.2] Thermal-electrical coupled modeling, including: [S2.2.1] Electrothermal coupling equations: For the electronic components described in [S1.1], solve Maxwell's equations and the transient heat conduction equation simultaneously to simulate the electrothermal coupling effect; [S2.2.2] Transient Thermal Analysis: Introducing Time Variables t Describe the changes of temperature and flow rate over time, and simulate the transient temperature rise process of electronic components under high load; [S2.3] Initial boundary condition settings: Inlet: ambient temperature 25℃, flow rate determined based on wind tunnel test data; Outlet: standard atmospheric pressure 101.325kPa; Heat source: engine and electronic component heat power calibrated based on bench test data; [S3] AI parameter optimization and model calibration, including grid optimization, heat source distribution optimization, and convective heat transfer coefficient optimization; [S3.1] Grid optimization, including: Initial grid generation: generate initial grid based on 3D CAD model of automotive engine compartment components, ensure higher grid density for key area high temperature components; AI optimization: use non-dominated sorting genetic algorithm for multi-objective optimization, automatically adjust grid density, preferentially increase grid density for key area high temperature components, and reduce waste of computing resources; [S3.2] Heat source distribution optimization: [S3.2.1] Initial heat source distribution: based on bench test data, set initial heat source power distribution according to component heat power proportion; [S3.2.2] AI calibration: use convolutional neural network (CNN) to extract features from the initial heat source power distribution described in step [S3.2.1]; adjust heat source power and position parameters according to simulation error through backpropagation algorithm; dynamically adjust based on real-time feedback to ensure consistency between heat source distribution and measured data; [S3.3] Convective heat transfer coefficient optimization: [S3.3.1] Initial convective heat transfer coefficient calculation: Use Dittus-Boelter formula to calculate forced convective heat transfer coefficient, as shown in formula 4; Formula 4; Where, Nu is the Nusselt number under forced convection conditions; Re is the Reynolds number, based on flow rate, pipe diameter, and fluid viscosity; Pr is the Prandtl number under forced convection conditions, based on fluid specific heat capacity, viscosity, and thermal conductivity; the flow rate is based on wind tunnel test data, with a value range of 0-20 m / s; the pipe diameter is based on the geometric dimensions of the cooling pipes in the engine compartment; the fluid viscosity is based on the physical properties of the cooling liquid; Use Churchill-Chu formula to calculate natural convective heat transfer coefficient, as shown in formula 5; Formula 5; Where, Nu is the Nusselt number under natural convection conditions; Ra is the Rayleigh number, based on temperature difference, gravitational acceleration, and fluid expansion coefficient; Pr = 0.71, the temperature difference is based on the temperature distribution in the automobile engine room, the gravitational acceleration = 9.81 m / s², the fluid expansion coefficient is based on the cooling liquid physical property parameters; [S3.3.2] AI optimization: according to the measured flow rate and the temperature data of the high-temperature components in the key area, the initial convective heat transfer coefficient calculated in step [S3.3.1] is optimized using the Q-learning reinforcement learning algorithm to improve the simulation accuracy; [S4] cloud edge collaborative simulation and result output, including cloud high-performance computing HPC and edge lightweight model; [S4.1] cloud HPC simulation: high-precision multi-physical field coupled simulation is completed in the cloud, supporting large-scale grid division and complex boundary conditions; GPU acceleration parallel computing technology is used to improve computing efficiency and shorten simulation time; the large-scale grid division refers to the number of grids in the order of millions to tens of millions, compared with the number of grids in the order of tens of thousands to hundreds of thousands in the usual grid division, the large-scale grid division requires higher computing resources and uses parallel computing technology; [S4.2] edge lightweight model: based on the cloud HPC simulation results in step [S4.1], a lightweight model is generated by simplifying the cloud simulation results through PCA dimension reduction and Surrogate Model, and is deployed on the vehicle-mounted edge computing device; the edge model monitors the temperature field changes in the engine room in real time, inputs the real-time working condition data of the vehicle speed, environmental temperature key parameters, and other parameters use default values or are ignored, and outputs the core indicators of high temperature early warning; [S4.3] result output: based on the calculation results of the edge lightweight model in step [S4.2], temperature cloud maps, high-temperature component lists, and optimization schemes are output; the temperature cloud map refers to outputting the temperature distribution cloud map in the automobile engine room, which directly shows the high-temperature area; the high-temperature component list refers to listing the components whose temperature exceeds the safety threshold and suggesting to replace the materials or adjust the layout; the optimization scheme refers to proposing fan speed control strategies and cooling liquid flow optimization schemes; [S5] dynamic feedback and continuous optimization: an error feedback mechanism is established to compare the simulation results with the test data and calculate the error rate; if the error is >1%, the grid density, heat source distribution, and convective heat transfer coefficient parameters are automatically adjusted; the continuous optimization scheme is executed, the AI model is trained using historical simulation data to improve the accuracy of parameter optimization; the simulation model is updated regularly to ensure that it adapts to the latest design changes and working condition requirements.

[0006] Further, the key features in step [S1.2] include: temperature change rate reflecting heat load change trend, flow rate distribution reflecting cooling effect, heat source power, thermal conductivity, and specific heat capacity reflecting heat intensity.

[0007] Further, the key area in step [S3.1] refers to the engine, electronic control unit ECU, battery pack, and cooling system; and the high-temperature component refers to a component with a temperature exceeding a safety threshold.

[0008] Further, the complex boundary condition in step [S4.1] includes: non-uniform flow rate distribution, such as turbulent flow effect; dynamic heat source power, such as sudden engine heat generation power change during rapid acceleration; multiphase flow boundary condition, such as interaction between cooling liquid and air; and transient temperature change, such as fluctuation of ambient temperature over time.

[0009] The second object of the present application is to provide an automobile engine compartment temperature field simulation analysis system based on multi-physical field coupling, which applies the simulation analysis method based on multi-physical field coupling according to any one of claims 1-4, comprising: [M1] Multi-source data acquisition and preprocessing module, which functions as data acquisition, AI cleaning, and feature extraction; [M2] Multi-physical field coupling modeling module, which functions as thermal-solid-flow-electric coupling modeling and dynamic boundary condition binding; [M3] AI parameter optimization and model calibration module, which functions as grid division, heat source distribution, and AI optimization of convection coefficient; [M4] Cloud-edge collaborative simulation and result output module, which functions as cloud-based high-performance simulation and edge-based lightweight model real-time warning; [M5] Dynamic feedback and continuous optimization module, which functions as error feedback and model iterative update.

[0010] The present application has the following advantages: 1. The simulation analysis method based on multi-physical field coupling provided by the present application realizes high-precision and high-efficiency simulation through thermal-solid-flow-electric coupling modeling combined with genetic algorithm, neural network, and reinforcement learning AI dynamic optimization, solving the problem of low parameter optimization efficiency and inability to comprehensively reflect multi-physical field coupling effect in the prior art.

[0011] 2. The simulation analysis method based on multi-physical field coupling provided by the present application supports transient condition simulation through the introduction of time series analysis and dynamic boundary condition adjustment, significantly improving the authenticity and practicality of simulation results and better predicting temperature field changes in actual driving.

[0012] 3. The simulation analysis method based on multi-physical field coupling provided by the present application accurately predicts the temperature rise risk of electronic components through simultaneous solution of Maxwell's equation and heat conduction equation, and is particularly suitable for thermal management optimization of key components such as battery packs and motor controllers in new energy vehicles.

[0013] 4. The automobile cabin temperature field simulation analysis method based on multi-physical field coupling provided by the application cooperates cloud high-performance computing with edge lightweight models, realizes efficient simulation and real-time early warning, greatly improves simulation efficiency, and supports rapid analysis and design optimization under extreme conditions.

[0014] 5. The automobile cabin temperature field simulation analysis method based on multi-physical field coupling provided by the application automatically optimizes grid division, heat source distribution, convection coefficient and other parameters through error feedback and historical data learning, ensures consistency of simulation results and test data, and improves long-term applicability of the model.

[0015] 6. The automobile cabin temperature field simulation analysis system based on multi-physical field coupling provided by the application supports rapid simulation under extreme conditions through multi-condition switching and boundary condition dynamic adjustment, can effectively improve design robustness, and reduce thermal failure risk. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, a brief introduction will be given below to the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0017] Figure 1 It is a flow chart of the automobile cabin temperature field simulation analysis method based on multi-physical field coupling. Figure 2 It is a schematic diagram of the automobile cabin temperature field simulation analysis system based on multi-physical field coupling. Figure 3 It is an ESC temperature cloud chart of the automobile cabin temperature field simulation analysis method based on multi-physical field coupling. Figure 4 It is a steering system temperature distribution cloud chart of the automobile cabin temperature field simulation analysis method based on multi-physical field coupling. Figure 5 It is a warm air pipe and engine oil cooler inlet pipe temperature distribution cloud chart of the automobile cabin temperature field simulation analysis method based on multi-physical field coupling. Figure 6 It is a gearbox oil cooler inlet pipe temperature distribution cloud chart of the automobile cabin temperature field simulation analysis method based on multi-physical field coupling. DETAILED DESCRIPTION

[0018] The following will be described in combination with the drawings Figure 1 to Figure 6 The principles and characteristics of the application are described, and the examples are only used to explain the application, and not to limit the scope of the application.

[0019] Example 1, the simulation analysis is based on steady state operation, as Figure 1 shown, the automobile engine room temperature field simulation analysis method based on multi-physical field coupling includes: [S1] Multi-source data acquisition and pretreatment: [S1.1] Multi-source data acquisition, including static data acquisition, dynamic data acquisition, temperature data acquisition; Static data, including CAD model, thermal conductivity, specific heat capacity, density of engine, ECU, battery pack, obtained through component material database, the acquisition results are shown in Table 1; Table 1: Static data acquisition table; .

[0020] Dynamic data, dynamic operating condition data collected in real time through vehicle sensor network and CAN bus, including: vehicle speed, ambient temperature, motor load; The acquisition results are shown in Table 2; Table 2: Dynamic data acquisition table;

[0021] Temperature data, collected by thermocouple and infrared thermometer, including the surface temperature of the cooling system of the automobile engine room components, exhaust system and electronic components; Table 3: Temperature data acquisition table;

[0022] [S1.2] AI pretreatment, including data cleaning, short-term thermal load prediction, feature extraction; [S1.2.1] Data cleaning refers to using the isolation forest machine learning algorithm to automatically remove outliers from the multi-source data collected in Table 1, Table 2 and Table 3, and the data cleaning results are shown in Table 4; Table 4: Data cleaning result information table;

[0023] Among them, Table 1 and Table 3 have no abnormal data, which will not be repeated here; [S1.2.2] Short-term thermal load prediction refers to using long short-term memory network LSTM to analyze the time series of dynamic operating condition data in Table 4, and predict the thermal load change trend in the next 5-10 minutes. The thermal load prediction results are shown in Table 5; Table 5: Thermal load prediction result information table;

[0024] [S1.2.3] Feature extraction refers to reducing the dimensionality of the data in Table 5 through principal component analysis (PCA), extracting the top 5 key features with a contribution rate of >95%, and reducing the computational complexity. The feature extraction results are shown in Table 6. Table 6: Feature extraction results information table

[0025] [S2] Multi-physical field coupling modeling, including thermal-solid-fluid coupling modeling, thermal-electric coupling modeling, and dynamic boundary condition binding. [S2.1] Thermal-solid-fluid coupling modeling, including: [S2.1.1] Thermal conduction model: based on Fourier's law, establish the thermal conduction equation of the internal components of the vehicle cabin, the solution is the temperature distribution T 热固流 as shown in equation 1. Equation 1 The specific calculation process and results are shown in Table 7. Table 7: Thermal conduction model data information table

[0026] [S2.1.2] Fluid dynamics model: using computational fluid dynamics (CFD) method, simulating the air flow in the cabin, solving the Navier-Stokes equation, the solution is the flow velocity u and pressure p as shown in equation 2. Equation 2 The Navier-Stokes equation is the core equation of fluid dynamics, although it is formally an equation, but through the combination of the continuity equation mass conservation and boundary conditions, the flow velocity u and pressure p can be solved simultaneously; the specific method includes: SIMPLE, PISO pressure-velocity coupling algorithm through iterative solution. Supplementary equation: continuity equation $\nabla \bullet u = 0$ provides additional constraints, in actual calculation through discretization and finite volume method numerical method can be solved simultaneously flow velocity u and pressure p ; the specific calculation process and results are shown in Table 8. Table 8: Fluid dynamics model data information table

[0027] [S2.1.3] Two-way fluid-solid coupling (FSI): through iterative solution of the coupling equation of fluid flow and solid thermal conduction, update the boundary conditions, ensure the accurate heat transfer between fluid and solid, as shown in equation 3. Equation 3; where, F fluid is the force of fluid on solid, calculated by fluid dynamics equation Navier-Stokes equation; F solid is the force of solid on fluid, calculated by solid mechanics equation thermoelastic equation; Equation 3 transfers heat through the coupling interface , the temperature distribution calculated from Equation 1 T 热固流 and the flow velocity calculated from Equation 2 u and the pressure as input, dynamically adjust the boundary conditions according to the coupling calculation results, to ensure the continuity of heat transfer; a. Interface heat transfer: Fluid side: Calculate convective heat transfer through Navier Stokes equation (\( q_{\text{fluid}} = h(T_{\text{fluid}} T_{\text{wall}}) \)); Solid side: Calculate heat conduction through heat conduction equation (\( q_{\text{solid}} =k \nabla T_{\text{solid}} \)); b. Equilibrium condition: \( q_{\text{fluid}} = q_{\text{solid}} \) at the interface, that is: \[h(T_{\text{fluid}}T_{\text{wall}})=k\left.\frac{\partial T_{\text{solid}}}{\partial n}\right|_{\text{wall}} \] Physical meaning: Energy conservation of fluid convection and solid heat conduction; Relevance of ΔT: \( T_{\text{fluid}} \) comes from the solution of Navier Stokes equation; \( T_{\text{solid}} \) comes from the solution of heat conduction equation; Specific boundary conditions are shown in Table 9; Table 9: Boundary condition data information table;

[0028] [S2.2] Thermal-electric coupling modeling, including: [S2.2.1] Joule heat calculation: Maxwell equations are solved simultaneously with transient heat conduction equations for the electronic components described in [S1.1] to simulate the electro-thermal coupling effect; Joule heat calculation: Maxwell equations describe the basic laws of electromagnetic fields, in which Joule heat is the heat generated by the current passing through the resistance, and the formula for calculating Joule heat is shown in Equation 6: Equation 6; Solve Maxwell equations and transient heat conduction equations simultaneously, and embed Joule heat as a heat source term in the transient heat conduction equation, and solve the temperature distribution T 热电 as shown in Equation 7: Equation 7; The calculation process and results are shown in Table 10; Table 10: Joule heat calculation data information table;

[0029] [S2.2.2] Transient thermal analysis: Introduce time variable t to describe the change of temperature and flow rate with time, and calculate the results by step [S2.2.1] T 热电 to calculate the temperature rise with time as the initial temperature field, simulate the transient temperature rise process of electronic components under high load; the calculation results of this step are a series T 热电 of temperature rise curves, as shown in Table 11; Table 11: Transient thermal analysis data information table;

[0030] [S2.3] Initial boundary condition setting: Inlet: ambient temperature 25℃; Outlet: standard atmospheric pressure 101.325kPa; Heat source: typical value based on bench test, as shown in Table 12; Table 12: Engine and electronic component heat power data table calibrated based on bench test data;

[0031] [S3] AI parameter optimization and model calibration, including grid optimization, heat source distribution optimization, and convection heat transfer coefficient optimization; [S3.1] Grid optimization, including: Initial mesh generation: Generate initial mesh based on 3D CAD model of automotive cabin components, ensure higher mesh density for components with temperature exceeding safety threshold 80°C, such as engine, ECU, battery pack, and cooling system areas. Mesh partitioning method: Generate mesh for each automotive cabin component separately, ensure interface mesh matching, such as node alignment on fluid-solid contact surface; correlation mechanism, transfer data through shared nodes, such as solid surface temperature → fluid boundary conditions; example: Engine mesh encryption area: Combustion chamber wall (mesh size ≤0.1 mm); Battery pack mesh sparse area: Shell (mesh size ≤1 mm); High and low temperature determination rule: High temperature, temperature > safety threshold × 80%; Low temperature, temperature < ambient temperature + 10°C; specific examples are shown in Table 13. Table 13: High and low temperature example information table;

[0032] AI optimization: Use non-dominated sorting genetic algorithm for multi-objective optimization, automatically adjust mesh density, preferentially encrypt key area high temperature component mesh, reduce waste of computing resources; its optimization results are shown in Table 14: Table 14: Mesh partitioning AI optimization comparison example table;

[0033] [S3.2] Heat source distribution optimization: [S3.2.1] Initial heat source distribution: Based on bench test data, set initial heat source power distribution according to component heat power proportion; as shown in Table 15: Table 15: Initial heat source distribution proportion allocation;

[0034] [S3.2.2] AI calibration: Use convolutional neural network (CNN) to extract features from the initial heat source power distribution described in step [S3.2.1]; adjust heat source power and position parameters according to simulation error through back propagation algorithm; dynamically adjust based on real-time feedback to ensure consistency between heat source distribution and measured data; Optimization process: Simulation output temperature field → CNN feature extraction → generate heat source power allocation scheme; weight matrix: used for CNN back propagation, such as convolution kernel weight update, used in optimization layer; CNN input: temperature field distribution map, resolution: 512 × 512 pixels; output: optimized heat source power allocation, such as engine power proportion 60%, ECU proportion 15%; Dynamic adjustment: Update weight matrix according to simulation error back propagation, example: Initial heat source: engine 60%, ECU 15%; After optimization: engine 55%, ECU 20%, error reduced by 5%; [S3.3] Optimization of convection heat transfer coefficient: [S3.3.1] Calculation of initial convective heat transfer coefficient: The forced convection heat transfer coefficient is calculated using the Dittus-Boelter formula, as shown in Formula 4; Formula 4; in, is the Nusselt number for forced convection conditions; is the Reynolds number, which is based on flow velocity, pipe diameter, and fluid viscosity; The Prandtl number for forced convection conditions is based on the specific heat capacity, viscosity, and thermal conductivity of the fluid. The flow rate is in the range of 0 to 20 m / s based on wind tunnel test data. The pipe diameter is based on the geometric dimensions of the cooling pipe in the cabin. The fluid viscosity is based on the physical properties of the coolant. The Churchill-Chu formula is used to calculate the natural convection heat transfer coefficient, as shown in Formula 5; Formula 5; in, is the Nusselt number for natural convection conditions; is the Rayleigh number, which is based on the temperature difference, gravitational acceleration, and fluid expansion coefficient; is the Prandtl number for natural convection conditions; the temperature difference is based on the temperature distribution in the vehicle cabin; the acceleration due to gravity = 9.81 m / s²; the fluid expansion coefficient is based on the physical properties of the coolant; the typical value ranges for this step are as follows: Forced convection Dittus-Boelter: $h = 50–150\ \text{W / m}²\cdot\text{K}$ (at a flow rate of 10 m / s); Natural convection Churchill-Chu: $h = 5–20\ \text{W / m}²\cdot\text{K}$ ($\Delta T = 20°C$); [S3.3.2] AI Optimization: Based on the measured flow rate and temperature data of high-temperature components in key areas, the Q-learning reinforcement learning algorithm is used to optimize the initial convective heat transfer coefficient calculated in step [S3.3.1] to improve simulation accuracy. Reward: +10 for every 1% decrease in error; penalty of 5 for every increase in error; State space: flow rate, temperature difference, pipe diameter; Optimization output: adjusted convection coefficient h, range 50-200 W / m²·K; [S4] Cloud-edge collaborative simulation and result output, including cloud high-performance computing HPC and edge lightweight model; [S4.1] Cloud high-performance computing HPC: high-precision multi-physical field coupling simulation is completed in the cloud, supporting large-scale grid division and turbulence effect, sudden engine heating power mutation during rapid acceleration, interaction between cooling liquid and air, complex boundary conditions of environmental temperature fluctuation over time; GPU acceleration parallel computing technology is used to improve computing efficiency and shorten simulation time; the large-scale grid division refers to the number of grids in the order of millions to tens of millions, compared with the number of grids in the order of tens of thousands to hundreds of thousands in the usual grid division, large-scale grid division requires higher computing resources and uses parallel computing technology; parallel computing uses MPI protocol to distribute computing nodes, each node has 32-core CPU + 4 GPU; [S4.2] Edge lightweight model: based on the high-precision simulation results in the cloud, extract temperature field distribution and heat source position key features, use PCA dimension reduction technology to simplify and generate a lightweight model, and deploy it on a vehicle-mounted edge computing device; the lightweight model is in the form of a neural network-based surrogate model, which inputs real-time operating condition data key parameters such as vehicle speed and environmental temperature, calculates the temperature field distribution temperature map and high-temperature risk warning core indicators through the model, and uses default values or ignores other parameters; the temperature field distribution refers to real-time monitoring of the cabin temperature field changes through the edge model to provide high-temperature risk warning; the high-temperature risk warning refers to listing components with temperatures exceeding the safety threshold, proposing fan speed control strategies and cooling liquid flow optimization schemes; Model construction: generate a simplified temperature field based on radial basis function RBF interpolation, and the specific temperature map is as shown in Figures 3 to 6 The picture has a temperature bar, which can be used to determine the temperature of the area.

[0035] Warning rules: if the component temperature > safety threshold, trigger sound and light alarm; specific examples are shown in Table 16; Table 16: High-temperature component list;

[0036] [S5] Dynamic feedback and continuous optimization: establish an error feedback mechanism, compare simulation results with test data, and calculate error rate; if the error > 1%, automatically adjust the grid density, heat source distribution, and convective heat transfer coefficient parameters; implement a continuous optimization scheme, train an AI model using historical simulation data to improve the accuracy of parameter optimization; update the simulation model regularly to ensure that it adapts to the latest design changes and operating condition requirements; error optimization examples are shown in Table 17; Table 17: Error optimization examples:

[0037] The above is the steady state working condition input data and simulation analysis result, the extreme high temperature working condition initial input parameter and simulation analysis result are shown in table 18. Table 18: input data and simulation analysis result example under extreme high temperature working condition:

[0038] High-speed running, transient condition, rapid acceleration, rapid deceleration, idle state working condition initial data input and simulation analysis process are similar to steady state working condition and extreme high temperature working condition, which will not be specifically described one by one.

[0039] Embodiment 2, as shown in Figure 2 The automobile engine compartment temperature field simulation analysis system based on multi-physical field coupling, the application of any one of claims 1-4 based on multi-physical field coupling automobile engine compartment temperature field simulation analysis method, comprising: [M1] multi-source data acquisition and pretreatment module, its function is data acquisition, AI cleaning and feature extraction; [M2] multi-physical field coupling modeling module, its function is thermal-solid-flow-electric coupling modeling, dynamic boundary condition binding; [M3] AI parameter optimization and model calibration module, its function is mesh division, heat source distribution, AI optimization of convective heat transfer coefficient; [M4] cloud edge collaborative simulation and result output module, its function is cloud high performance simulation, edge light weight model real-time early warning; [M5] dynamic feedback and continuous optimization module, its function is error feedback, model iteration update.

[0040] The above only describes the preferred embodiments of the present application, and does not limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. The automobile cabin temperature field simulation analysis method based on multi-physics field coupling is characterized by: It includes the modeling process, the simulation process and the connection between the modeling process and the simulation process; the modeling process is completed on the local client, including [S1] multi-source data acquisition and preprocessing, [S2] multi-physics field coupling modeling, [S3] AI parameter optimization and model calibration, and after the local modeling is completed, the model data is uploaded to the cloud; the simulation process is completed on the cloud server, including [S4] cloud-edge collaborative simulation and result output, and the simulation results are returned to the local or edge device; the connection between the modeling process and the simulation process includes [S5] dynamic feedback and continuous optimization, whose function is parameter calibration of the modeling process and iterative improvement of the simulation results.

2. The automobile cabin temperature field simulation analysis method based on multi-physics field coupling according to claim 1 is characterized by: [S1] Multi-source data acquisition and preprocessing, including: [S1.1] Multi-source data acquisition, including static data acquisition, dynamic data acquisition, and temperature data acquisition; the static data acquisition includes the CAD model, thermal conductivity, specific heat capacity, and density of the engine, ECU, and battery pack, obtained through the component material database; the dynamic data refers to the dynamic operating condition data collected in real time through the on-board sensor network and CAN bus, including: vehicle speed, ambient temperature, and motor load; the temperature data, collected through thermocouples and infrared thermometers, includes the temperature of the cooling system of components in the vehicle cabin, the surface of the exhaust system, and the surface of electronic components; [S1.2] AI preprocessing, including data cleaning, short-term heat load forecasting, and feature extraction; [S1.2.1] Data cleaning refers to using the isolation forest machine learning algorithm to automatically remove outliers from the multi-source data collected in step [S1.1]; [S1.2.2] Short-term heat load forecasting refers to using a long short-term memory (LSTM) network to perform time series analysis on the dynamic operating condition data in the result data of step [S1.2.1] to predict the heat load change trend in the next 5 to 10 minutes; [S1.2.3] Feature extraction refers to performing data dimensionality reduction on the data obtained in step [S1.2.2] through principal component analysis (PCA), extracting key features, and reducing computational complexity. [S2] Multi-physics coupling modeling, including thermal-solid-fluid coupling modeling, thermal-electric coupling modeling, and dynamic binding of boundary conditions; [S2.1] Thermal-solid-fluid coupled modeling, including: [S2.1.1] Heat conduction model: Based on Fourier's law, the heat conduction equation of the components in the vehicle cabin is established, and the solution is the temperature distribution T 热固流 , as shown in formula 1; Formula 1: in: ρ For density, c p is the specific heat capacity, T 热固流 is the temperature distribution, t For time, k is the thermal conductivity, Q source is the heat source power; [S2.1.2] Fluid dynamics model: Computational fluid dynamics (CFD) is used to simulate the air flow in the cabin and solve the Navier-Stokes equations. The solution is the flow velocity. u and pressure p , as shown in formula 2; Formula 2: in: u is the flow rate, t For time, ρ is the density, p For pressure, μ is the dynamic viscosity, g For external forces; [S2.1.3] Bidirectional fluid-solid interaction (FSI): By iteratively solving the coupled equations for fluid flow and solid heat conduction, the boundary conditions are updated to ensure accurate heat transfer between the fluid and the solid, as shown in Equation 3. Formula 3: in, F fluid is the force exerted by the fluid on the solid, which is calculated by the fluid dynamics equation Navier-Stokes equation; F solid is the reaction force of the solid on the fluid, which is calculated by the solid mechanics equation and the thermoelastic equation; Formula 3 transfers heat through the coupling interface , the temperature distribution calculated by formula 1 T The flow rate calculated by formula 2 is u and pressure As input, the boundary conditions are dynamically adjusted and updated according to the coupling calculation results to ensure the continuity of heat transfer; [S2.2] Thermal-electrical coupled modeling, including: [S2.2.1] Electrothermal coupling equations: For the electronic components described in [S1.1], solve Maxwell's equations and the transient heat conduction equation simultaneously to simulate the electrothermal coupling effect; [S2.2.2] Transient Thermal Analysis: Introducing Time Variables t Describe the changes of temperature and flow rate over time, and simulate the transient temperature rise process of electronic components under high load; [S2.3] Initial boundary condition settings: Air inlet: The ambient temperature is 25°C and the flow rate is determined based on wind tunnel test data; Air outlet: standard atmospheric pressure 101.325kPa; Heat source: Calibrate the heating power of the engine and electronic components based on bench test data; [S3] AI parameter optimization and model calibration, including grid optimization, heat source distribution optimization, and convection heat transfer coefficient optimization; [S3.1] Mesh optimization, including: Initial mesh generation: Generates the initial mesh based on the 3D CAD model of the vehicle's interior components, ensuring a high mesh density for high-temperature components in key areas. AI optimization: Utilizes non-dominated sorting genetic algorithm for multi-objective optimization, automatically adjusts mesh density, prioritizes mesh densification of high-temperature components in key areas, and reduces computing resource waste. [S3.2] Heat source distribution optimization: [S3.2.1] Initial heat source distribution: Based on bench test data, set the initial heat source power distribution according to the component heating power ratio; [S3.2.2] AI Calibration: Use a convolutional neural network (CNN) to extract features from the initial heat source power distribution described in step [S3.2.1]. Adjust the heat source power and position parameters based on the simulation error using a backpropagation algorithm. Dynamic adjustments are made based on real-time feedback to ensure that the heat source distribution is consistent with the measured data. [S3.3] Optimization of convection heat transfer coefficient: [S3.3.1] Calculation of initial convective heat transfer coefficient: The forced convection heat transfer coefficient is calculated using the Dittus-Boelter formula, as shown in Formula 4; Formula 4: in, is the Nusselt number for forced convection conditions; is the Reynolds number, which is based on flow velocity, pipe diameter, and fluid viscosity; The Prandtl number for forced convection conditions is based on the specific heat capacity, viscosity, and thermal conductivity of the fluid. The flow rate is in the range of 0 to 20 m / s based on wind tunnel test data. The pipe diameter is based on the geometric dimensions of the cooling pipe in the cabin. The fluid viscosity is based on the physical properties of the coolant. The Churchill-Chu formula is used to calculate the natural convection heat transfer coefficient, as shown in Formula 5; Formula 5: in, is the Nusselt number for natural convection conditions; is the Rayleigh number, which is based on the temperature difference, gravitational acceleration, and fluid expansion coefficient; is the Prandtl number for natural convection conditions; the temperature difference is based on the temperature distribution in the vehicle cabin; the acceleration due to gravity = 9.81 m / s²; the fluid expansion coefficient is based on the physical properties of the coolant; [S3.3.2] AI Optimization: Based on the measured flow rate and temperature data of high-temperature components in key areas, the Q-learning reinforcement learning algorithm is used to optimize the initial convective heat transfer coefficient calculated in step [S3.3.1] to improve simulation accuracy. [S4] Cloud-edge collaborative simulation and result output, including cloud-based high-performance computing (HPC) and edge-side lightweight models; [S4.1] Cloud-based HPC simulation: High-precision multi-physics coupled simulations are performed in the cloud, supporting large-scale meshing and complex boundary conditions. GPU-accelerated parallel computing technology is used to improve computational efficiency and shorten simulation time. Large-scale meshing refers to mesh sizes ranging from millions to tens of millions. Compared to conventional meshing with mesh sizes ranging from tens to hundreds of thousands, large-scale meshing requires higher computing resources and employs parallel computing technology. [S4.2] Edge lightweight model: Based on the cloud-based HPC simulation results described in step [S4.1], simplify the cloud-based simulation results through PCA dimensionality reduction and the Surrogate Model to generate a lightweight model, which is deployed on the vehicle-mounted edge computing device. The edge model monitors changes in the cabin temperature field in real time, inputting key parameters such as vehicle speed and ambient temperature real-time operating condition data. Other parameters are used as default values ​​or ignored, and the core indicators for high-temperature warning are output. [S4.3] Result output: Based on the edge lightweight model calculation results described in step [S4.2], output a temperature cloud map, a list of high-temperature components, and an optimization plan. The temperature cloud map is a cloud map of the temperature distribution in the vehicle cabin, visually displaying high-temperature areas. The high-temperature component list is a list of components whose temperatures exceed the safety threshold, along with suggestions for material replacement or layout adjustment. The optimization plan is a proposed fan speed control strategy and a coolant flow optimization plan. [S5] Dynamic feedback and continuous optimization: Establish an error feedback mechanism, compare simulation results with test data, and calculate the error rate; if the error is >1%, automatically adjust the grid density, heat source distribution, and convective heat transfer coefficient parameters; implement a continuous optimization plan, use historical simulation data to train the AI ​​model, and improve the accuracy of parameter optimization; regularly update the simulation model to ensure it adapts to the latest design changes and working conditions.

3. According to the automobile cabin temperature field simulation analysis method based on multi-physics field coupling according to claim 1, the key features of step [S1.2] include: The temperature change rate that reflects the trend of heat load changes, the flow velocity distribution that reflects the cooling effect, the heat source power, thermal conductivity and specific heat capacity that reflect the heating intensity.

4. According to the automobile cabin temperature field simulation and analysis method based on multi-physics field coupling according to claim 1, the key areas in step [S3.1] refer to the engine, electronic control unit ECU, battery pack, and cooling system; the high-temperature components refer to components whose temperatures exceed the safety threshold.

5. According to the automobile cabin temperature field simulation analysis method based on multi-physics field coupling according to claim 1, the complex boundary conditions in step [S4.1] include: Non-uniform velocity distribution, such as turbulence effects; Dynamic heat source power, such as sudden changes in engine heat power during rapid acceleration; Multiphase flow boundary conditions, such as coolant-air interaction; Transient temperature changes, such as ambient temperature fluctuations over time.

6. A vehicle cabin temperature field simulation and analysis system based on multi-physics field coupling, applying the vehicle cabin temperature field simulation and analysis method based on multi-physics field coupling according to any one of claims 1 to 5, comprising: [M1] Multi-source data acquisition and preprocessing module, which is responsible for data acquisition, AI cleaning and feature extraction; [M2] Multi-physics coupling modeling module, which provides thermal-solid-fluid-electric coupling modeling and dynamic boundary condition binding; [M3] AI parameter optimization and model calibration module, which is responsible for AI optimization of meshing, heat source distribution, and convection coefficient; [M4] Cloud-edge collaborative simulation and result output module, which provides high-performance simulation on the cloud and real-time warning of lightweight models on the edge; [M5] Dynamic feedback and continuous optimization module, whose functions are error feedback and model iterative update.

Images