A method for calculating and predicting the deformation of an electric vehicle battery pack under its base.

Abstract

This invention discloses a method for calculating and predicting the bottom deformation of an electric vehicle battery pack, comprising the following steps: 1) establishing a finite element model of the battery pack system and a simulation model of the battery pack system impacting an obstacle at its bottom; 2) testing the vertical displacement of the lower shell of the finite element model of the battery pack system under different bottom impact conditions with relevant simulation parameter combinations; 3) constructing training data pairs using relevant simulation parameter combinations and corresponding lower shell vertical displacements; 4) training a machine learning model using multiple training data pairs to obtain a prediction model for the vertical displacement of the lower shell of the battery pack system; 5) predicting the vertical displacement of the lower shell of the battery pack system under relevant simulation parameter combinations using the prediction model for the vertical displacement of the lower shell of the battery pack system. This invention features high-precision prediction of the mechanical response of the battery pack system under bottom impact.

Description

Technical Field

[0001] This invention relates to the field of electric vehicles, specifically a method for calculating and predicting the deformation of an electric vehicle battery pack under its base. Background Technology

[0002] Electric vehicles are gaining an increasing share of the automotive market due to their environmental and technological advantages. As the power output device for electric vehicles, the safety of the battery pack system is paramount. Given the increasingly complex driving environment, various bottoming-out scenarios can cause different mechanical damage to the battery pack system, potentially leading to fires, explosions, and other safety incidents, significantly impacting the driving safety and vehicle stability of electric vehicles. Furthermore, to provide more interior space, the battery pack system is typically located at the bottom of the electric vehicle, sacrificing ground clearance and increasing the likelihood of bottoming out.

[0003] The battery pack system is the power source for pure electric vehicles and hybrid electric vehicles, and generally consists of components such as a lower shell, upper cover, battery modules, longitudinal beams / sides, cross beams / sides, connecting brackets, long / short brackets, and lifting lugs. For a battery pack system with a given structure, its bottom mechanical safety performance is mainly determined by the thickness of the lower shell and corresponding parameters under different bottoming conditions. Conducting extensive experimental analysis on different battery pack prototypes by changing component parameters to study their safety under a series of bottom impact conditions would be extremely resource-intensive and time-consuming. Therefore, using a combination of finite element simulation and machine learning to predict the vertical displacement of the lower shell deformation in the mechanical response of the battery pack system under bottoming conditions has significant engineering practical value.

[0004] In recent years, domestic and international experts and scholars have conducted extensive and systematic research on the mechanical safety of battery pack systems and individual battery cells. Collision safety is a crucial component, and research includes methods such as optimizing thickness parameters, using new materials, and adopting different optimized structures. To design a rational battery pack system, designers must perform tens of thousands of analyses based on finite element models to understand the mechanical characteristics of the entire system. For example... Figure 2 As shown, components of the battery pack system, such as the bottom shell, top cover, battery module, longitudinal beams / sides, cross beams / sides, connecting brackets, long / short brackets, and lifting lugs, must undergo power-order finite element analysis to select appropriate thickness and material parameters. This process is very laborious, expensive, and time-consuming, and therefore cannot be widely used in the fiercely competitive automotive industry. Summary of the Invention

[0005] The purpose of this invention is to provide a method for calculating and predicting the deformation of an electric vehicle battery pack under its load, comprising the following steps:

[0006] 1) Establish a finite element model of the battery pack system and a simulation model of the battery pack system impacting obstacles at its bottom;

[0007] 2) Test the vertical displacement of the lower shell of the battery pack system finite element model under relevant simulation parameter combinations;

[0008] 3) Construct training data pairs using relevant simulation parameter combinations and corresponding lower shell deformation and vertical displacement;

[0009] 4) A machine learning model was trained using multiple training data sets to obtain a prediction model for the vertical displacement of the lower shell of the battery pack system.

[0010] 5) The vertical displacement of the battery pack system's casing under the relevant simulation parameter combinations is predicted using the vertical displacement prediction model of the casing deformation under the battery pack system.

[0011] Furthermore, the steps for establishing the finite element model of the battery pack system include:

[0012] 1.1) Establish a finite element model of the battery pack system based on its housing dimensions, structure, and material;

[0013] 1.2) Based on the dimensions and materials of the battery modules in the battery pack system, establish a finite element model of the battery modules;

[0014] 1.3) Based on the connection relationship of each component of the battery pack system, the finite element model of the shell and the finite element model of the battery module are coupled to obtain the finite element model of the battery pack system.

[0015] Furthermore, the steps for establishing the finite element model of the battery module include:

[0016] 1.2.1) Establish a geometric model of the battery module based on its dimensional parameters;

[0017] 1.2.2) Homogenize the battery module materials;

[0018] 1.2.3) Define the material parameters of the battery module geometric model based on the battery module material information obtained from the homogenization process, and obtain the battery module finite element model.

[0019] Furthermore, the steps for establishing a simulation model of the impact obstacle at the bottom of the battery pack system include: performing statistical analysis on the impact obstacle at the bottom of the battery pack system, replacing the impact obstacle with a cone model, and simulating the sharpness of the impact obstacle by changing the size of the top radius of the cone model in order to extract obstacle features, thereby constructing a simulation model of the impact obstacle at the bottom of the battery pack system.

[0020] Furthermore, the relevant simulation parameter combination includes: impact velocity, top radius of the obstacle simulation model, thickness of the lower shell of the battery pack system, and impact angle.

[0021] Furthermore, the steps for training a machine learning model using multiple training datasets include:

[0022] 4.1) Divide multiple training data pairs into training and test sets;

[0023] 4.2) The traditional machine learning model is optimized using the Cuckoo Search optimization algorithm;

[0024] 4.3) The machine learning model is trained using the training set to obtain the trained prediction model of the vertical displacement of the lower shell of the battery pack system.

[0025] 4.4) The trained machine learning model is tested using a test set, and the model parameters are further adjusted to obtain the completed prediction model of the vertical displacement of the lower shell of the battery pack system.

[0026] Furthermore, when training the machine learning model or testing the prediction model of the vertical displacement of the lower shell of the battery pack system, the relevant simulation parameter combination in the training data pair is used as input, and the corresponding vertical displacement of the lower shell of the battery pack system is used as output.

[0027] Furthermore, the machine learning model includes an input layer, a hidden layer, and an output layer.

[0028] Furthermore, during training, the machine learning model uses an optimization algorithm to adjust the weights in the hidden layer;

[0029] Among them, the weight X after iterative update t+1 As shown below:

[0030] X t+1 =X t +γ*Heaviside(p-ε)*(X i -X j (1)

[0031] In the formula, t is the number of iterations, and X t Let be the weights for the t-th iteration; Heaviside() is the jump function; p is the probability of the optimization algorithm mechanism; γ and ε are uniformly distributed random numbers; X i X j To optimize the algorithm, the weights of other connected components on the path are optimized.

[0032] The jump function Heaviside(x) is shown below:

[0033]

[0034] Furthermore, the vertical displacement prediction model for the lower housing deformation of the battery pack system is shown below:

[0035] A0=Z0 (3)

[0036] Z i =W i T A i-1 +B i , i = 1…n (4)

[0037] A i =f i (Z i ), i = 1…n (5)

[0038] In the formula, Z0 is the input of the 0th layer, A0 is the output of the 0th layer, and Z i W is the input of the i-th layer. i Let B be the weight matrix of the i-th layer. i Let A be the bias matrix of the i-th layer. i f is the output of the i-th layer, n is the total number of hidden and output layers, and f i () is the activation function for the i-th layer. i-1 This is the output of the (i-1)th layer.

[0039] The technical effects of this invention are undeniable. This invention combines finite element modeling of the battery pack system with a machine learning model. By establishing a complete finite element model of the battery pack system, sufficient data samples can be obtained, thereby enabling the machine learning model to be fully trained. This allows the machine learning model to have the characteristic of high-precision prediction of the mechanical response of the battery pack system under impact, while avoiding the problem of complex prediction process when predicting the mechanical characteristics of the battery pack system through the finite element model. Attached Figure Description

[0040] Figure 1 A flowchart of a method for calculating and predicting the deformation of an electric vehicle battery pack under its base;

[0041] Figure 2 This is a structural diagram of the battery pack system;

[0042] Figure 3 A simulation model for the impact of obstacles at the bottom of the battery pack system;

[0043] Figure 4 Simulation analysis diagram for bottom support;

[0044] Figure 5 A cloud map showing the simulation results of the baseline working condition;

[0045] Figure 6 The machine learning model to be built;

[0046] Figure 7The figure shows the predicted vertical displacement of the lower shell in relation to the impact angle and impact velocity.

[0047] Figure 8 The figure shows the predicted vertical displacement of the lower shell, which is the thickness of the lower shell and the radius of the top of the cone model.

[0048] Figure 9 Box plot showing the distribution of results from finite element analysis (FEM) and machine learning model (CS-BP).

[0049] In the diagram, 1 is the upper shell, 2 is the battery module, 3 is the lifting lug, 4 is the short bracket, 5 is the crossbeam, 6 is the connecting bracket, 7 is the long bracket, 8 is the bottom shell, and 9 is the upper bracket. Detailed Implementation

[0050] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0051] Example 1:

[0052] See Figures 1 to 9 A method for calculating and predicting the deformation of an electric vehicle battery pack under its load, comprising the following steps:

[0053] 1) Establish a finite element model of the battery pack system and a simulation model of the battery pack system impacting obstacles at its bottom;

[0054] 2) Test the vertical displacement of the lower shell of the battery pack system finite element model under relevant simulation parameter combinations;

[0055] 3) Construct training data pairs using relevant simulation parameter combinations and corresponding lower shell deformation and vertical displacement;

[0056] 4) A machine learning model was trained using multiple training data sets to obtain a prediction model for the vertical displacement of the lower shell of the battery pack system.

[0057] 5) The vertical displacement of the battery pack system's casing under the relevant simulation parameter combinations is predicted using the vertical displacement prediction model of the casing deformation under the battery pack system.

[0058] Example 2:

[0059] A method for calculating and predicting the deformation of an electric vehicle battery pack under its base, with the same technical content as in Example 1, further comprising the following steps for establishing a finite element model of the battery pack system:

[0060] 1.1) Establish a finite element model of the battery pack system based on its housing dimensions, structure, and material;

[0061] 1.2) Based on the dimensions and materials of the battery modules in the battery pack system, establish a finite element model of the battery modules;

[0062] 1.3) Based on the connection relationship of each component of the battery pack system, the finite element model of the shell and the finite element model of the battery module are coupled to obtain the finite element model of the battery pack system.

[0063] Example 3:

[0064] A method for calculating and predicting the deformation of an electric vehicle battery pack under its base, with the same technical content as any one of Embodiments 1-2, further comprising the following steps for establishing a finite element model of the battery module:

[0065] 1.2.1) Establish a geometric model of the battery module based on its dimensional parameters;

[0066] 1.2.2) Homogenize the battery module materials;

[0067] 1.2.3) Define the material parameters of the battery module geometric model based on the battery module material information obtained from the homogenization process, and obtain the battery module finite element model.

[0068] Example 4:

[0069] A method for calculating and predicting the bottom deformation of an electric vehicle battery pack, with technical content identical to any one of embodiments 1-3, further includes the following steps for establishing a simulation model of the battery pack system's bottom impact obstacle: Statistical analysis of the obstacles impacting the bottom of the battery pack system reveals that the obstacles are mostly irregular polyhedra, but typically only one part participates in the impact. A cone model is used, and the sharpness of the impact obstacle is simulated by changing the size of its apex radius. Obstacle features are extracted based on this rule, thereby constructing a simulation model of the battery pack system's bottom impact obstacle.

[0070] During the simulation, the cone model with a small top radius represents sharp impact objects such as parking space locks and sharp stones, while the cone model with a large top radius represents blunt impact objects such as speed bumps.

[0071] Example 5:

[0072] A method for calculating and predicting the bottom deformation of an electric vehicle battery pack, with the same technical content as any one of embodiments 1-4, further comprising the following related simulation parameter combinations: impact velocity, top radius of the obstacle simulation model, thickness of the lower shell of the battery pack system, and impact angle.

[0073] Example 6:

[0074] A method for calculating and predicting the deformation of an electric vehicle battery pack under its load, with the same technical content as any one of embodiments 1-5, further comprising the step of training a machine learning model using multiple training data, including:

[0075] 4.1) Divide multiple training data pairs into training and test sets;

[0076] 4.2) Optimize traditional machine learning models using swarm optimization algorithms;

[0077] 4.3) The machine learning model is trained using the training set to obtain the trained prediction model of the vertical displacement of the lower shell of the battery pack system.

[0078] 4.4) The trained machine learning model is tested using a test set, and the model parameters are further adjusted to obtain the completed prediction model of the vertical displacement of the lower shell of the battery pack system.

[0079] Example 7:

[0080] A method for calculating and predicting the deformation of an electric vehicle battery pack under its bottom is provided. The technical content is the same as any one of embodiments 1-6. Further, when training the machine learning model or testing the prediction model of the vertical displacement of the lower shell of the battery pack system, the relevant simulation parameter combination in the training data pair is used as input, and the corresponding vertical displacement of the lower shell of the battery pack system is used as output.

[0081] Example 8:

[0082] A method for calculating and predicting the deformation of an electric vehicle battery pack under its bottom, with the same technical content as any one of embodiments 1-7, further comprising: an input layer, a hidden layer, and an output layer.

[0083] Example 9:

[0084] A method for calculating and predicting the deformation of an electric vehicle battery pack under its bottom, with the same technical content as any one of embodiments 1-8, further wherein the machine learning model uses an optimization algorithm to adjust the weights in the hidden layer during training;

[0085] Among them, the iteratively updated weight X t+1 As shown below:

[0086] X t+1 =X t +γ*Heaviside(p-ε)*(X i -X j (1)

[0087] In the formula, t is the number of iterations, and X tLet be the weights for the t-th iteration; Heaviside() is the jump function; p is the probability of the optimization algorithm mechanism; γ and ε are uniformly distributed random numbers; X i X j To optimize the algorithm, the weights of other connected components on the path are optimized.

[0088] The jump function Heaviside(x) is shown below:

[0089]

[0090] Example 10:

[0091] A method for calculating and predicting the deformation of an electric vehicle battery pack under its bottom, with the same technical content as any one of embodiments 1-9, further wherein the vertical displacement prediction model for the deformation of the lower shell of the battery pack system is as follows:

[0092] A0=Z0 (3)

[0093] Z i =W i T A i-1 +B i , i = 1…n (4)

[0094] A i =f i (Z i ), i = 1…n (5)

[0095] In the formula, Z0 is the input of the 0th layer, A0 is the output of the 0th layer, and Z i W is the input of the i-th layer. i Let B be the weight matrix of the i-th layer. i Let A be the bias matrix of the i-th layer. i f is the output of the i-th layer, n is the total number of hidden and output layers, and f i () is the activation function for the i-th layer. i-1 This is the output of the (i-1)th layer.

[0096] Example 11:

[0097] A method for calculating and predicting the deformation of an electric vehicle battery pack under its load, comprising the following steps:

[0098] S1. Establish a finite element model of the battery pack system and a simulation model of the impact obstacle at the bottom of the battery pack system;

[0099] In this embodiment, the finite element model can be implemented on different finite element software, such as LS-DYNA or HYPERMESH.

[0100] Step S1 includes the following sub-steps:

[0101] S11. Based on the housing dimensions, housing structure, and housing material of the battery pack system, establish a finite element model of the housing;

[0102] In this embodiment, the specific operation of step S11 is as follows: after obtaining the shell size, shell structure and shell material, define the shell model type, size, thickness and material parameters in the finite element software, and establish the shell finite element model.

[0103] S12. Based on the dimensions and materials of the battery modules in the battery pack system, establish a finite element model of the battery modules; such as... Figure 2 As shown, the battery pack system includes an upper shell 1, a battery module 2, a lifting lug 3, a short bracket 4, a crossbeam 5, a connecting bracket 6, a long bracket 7, a bottom shell 8, and an upper bracket 9.

[0104] Step S12 includes the following sub-steps:

[0105] S121. Establish a geometric model of the battery module based on its dimensional parameters;

[0106] S122. Homogenize the battery module materials;

[0107] S123. Define the material parameters of the battery module geometric model based on the battery module material information obtained from the homogenization process, and obtain the battery module finite element model.

[0108] S13. Based on the connection relationship of each component of the battery pack system, couple the finite element model of the shell and the finite element model of the battery module to obtain the finite element model of the battery pack system.

[0109] In step S13, coupling means establishing the connection relationship between the shell finite element model and the battery module finite element model. The connection relationship includes contact connection relationships such as welding and friction.

[0110] S14. Based on the statistical analysis of the impact obstacles at the bottom of various battery pack systems, extract the same features and establish an obstacle imitation model.

[0111] In step S14, the impact process between various obstacles and the battery pack system is analyzed. Typically, only one apex of the obstacle participates in the impact; therefore, a cone model with a bottom radius of 100mm is selected to represent the impacting obstacle. The sharpness of the impacting obstacle is simulated by changing the apex radius of the cone model. Figure 3 As shown.

[0112] S2. Test the vertical displacement of the lower shell of the battery pack system finite element model under the relevant simulation parameter combination;

[0113] In this embodiment, step S2 specifically involves: based on the requirements of the national standard GB18384-2020 and according to actual R&D needs, selecting the middle of the front end of the battery pack system as the impact point, conducting a bottom impact simulation analysis of the battery pack system, and obtaining the vertical displacement data of the deformation of the lower shell of the battery pack system. Figure 4 This is a simulation analysis diagram of the impact. Figure 5 The results of the bottom impact simulation are shown in the cloud map. Table 1 shows the influence factor levels involved in the bottom impact simulation of the battery pack.

[0114] Table 1. Influence factor levels of battery pack bottom impact simulation

[0115]

[0116] S3. Combine the relevant simulation parameter data with the corresponding lower shell deformation vertical displacement to form a training data pair;

[0117] The relevant simulation parameters in step S3 include: impact velocity, top radius of the obstacle simulation model, thickness of the lower casing of the battery pack system, and impact angle.

[0118] S4. Use multiple training data pairs to build a machine learning model;

[0119] like Figure 6 As shown, step S4 includes the following sub-steps:

[0120] S41. Divide multiple training data (datasets) into training set and test set;

[0121] Specifically, in the case of 600 training data pairs, 80% of the data can be used as the training set and 20% of the data as the test set.

[0122] S42. The traditional machine learning model is optimized by using the Cuckoo Search Optimization Algorithm, which significantly improves the predictive performance of the machine learning model and obtains a machine learning model for prediction.

[0123] Specifically, S42 involves optimizing the Bp neural network model using the optimization capability of the Cuckoo Search Optimization Algorithm, enhancing the weight search capability in the hidden layer, and improving prediction performance.

[0124] S43. Train the machine learning model using the training set to obtain the trained machine learning model;

[0125] S44. Test the trained machine learning model using a test set, and further adjust the model parameters to obtain the completed machine learning model.

[0126] In steps S43 and S44, when training or testing the machine learning model, the relevant simulation parameter combination data of the training data pair is used as the input of the machine learning model, and the corresponding vertical displacement of the lower shell deformation of the battery pack system is used as the output of the machine learning model.

[0127] The machine learning model in step S4 includes an input layer, a hidden layer, and an output layer. During training, the optimization algorithm adjusts the weights in the hidden layer to enable rapid allocation and optimization.

[0128] The neurons in the input layer are responsible for receiving data values ​​and propagating them forward to the neurons in the middle layers of the neural network, namely the hidden layers. The weighted sum of the hidden layers is then propagated forward to the output layer, which displays the output of the neural network.

[0129] The prediction model is as follows:

[0130] A0 = Z0

[0131] Z i =W i T A i-1 +B i , i = 1…n

[0132] A i =f i (Z i ), i = 1…n

[0133] Where Z0 is the input of layer 0, A0 is the output of layer 0, and Z i W is the input of the i-th layer. i Let B be the weight matrix of the i-th layer. i Let A be the bias matrix of the i-th layer. i f is the output of the i-th layer, n is the total number of hidden and output layers, and f i () represents the activation function of the i-th layer. The Tansig function is selected as the activation function for the hidden layer, and the Purelin function is used as the activation function between the hidden layer and the output layer.

[0134] Meanwhile, the optimization algorithm operates in the hidden layer, with the specific model as follows:

[0135] X t+1 =X t +γ*Heaviside(p-ε)*(X i -X j )

[0136] Where t is the number of iterations. and These represent the current weight and the next weight change, respectively. Heaviside() is a jump function (x>0, =1; x<0, =0), p is the probability of the optimization algorithm mechanism, and γ and ε are uniformly distributed random numbers; X i X j To optimize the algorithm, the weights of other connected components on the path are optimized.

[0137] After one forward propagation, the error between the current machine learning model's output and the target output is calculated. If the error does not meet the set limit, the optimal weights are changed and forward propagation is performed again until the error meets the requirements. The loss function for calculating the error can be the mean squared error.

[0138] During training and testing, the parameters that are adjusted for the machine learning model include: the number of hidden layers, the number of neurons in each hidden layer, and the learning rate.

[0139] In this embodiment, by comparing the output results of the machine learning model training or testing process with the bottom impact simulation data of the finite element model of the battery pack system, parameters such as the number of hidden layers, the number of neurons in each hidden layer, and the learning rate can be further adjusted.

[0140] S5. Predict the vertical displacement of the lower shell of the battery pack system under relevant simulation parameter combinations using a machine learning model.

[0141] After the machine learning model is built, the relevant parameter combination data is input into the machine learning model, and the vertical displacement of the lower shell deformation output by the machine learning model is the predicted value of this invention.

[0142] Experimental results:

[0143] 1. Figures 7-8 This is a graph showing the vertical displacement prediction results from the machine learning model. Figure 9 Box plots comparing the distribution of results from finite element analysis (FEM) and machine learning models (CS-BP); elements in the box plots include the 25th-75th quantile distribution range, the minimum-maximum distribution range, the median, and the mean. From Figure 9 As can be seen, the established machine learning model can predict the vertical displacement of the lower casing of the battery pack system relatively well.

[0144] 2. To express the accuracy of the machine learning model in detail, the coefficient of determination (R²) is selected. 2 The mean absolute error (MAE), mean square error (MSE), mean absolute percentage error (MAPE), and root mean square error (RMSE) are used as accuracy evaluation indicators.

[0145] 3. MSE, MAE, MAPE, and RMSE are used to evaluate regression prediction models, and their values ​​represent the relevant errors. The smaller the error, the higher the accuracy of the model. R 2 R is used to measure the quality of a regression model. 2 The larger the value, the better the model's performance. Tables 2, 3, and 4 describe the experimental data for model building, the prediction results of the machine learning model for 30 samples, and the prediction accuracy of the machine learning model after 10 runs, respectively.

[0146] Table 2. Experimental data of machine learning models

[0147]

[0148] Table 3. Prediction results of machine learning models for 30 groups of samples.

[0149]

[0150] Table 4. Prediction accuracy of the machine learning model after 10 runs

[0151]

[0152] As can be seen from the table, MAE, MAPE, and RMSE are relatively small, and R... 2 The result is close to 1. The results show that the established machine learning model can predict the vertical displacement of the lower casing of the battery pack system when an electric vehicle bottoms out.

[0153] In summary, this embodiment comprehensively considers the calculation and prediction of electric vehicle battery pack deformation upon bottoming out. Results show that the established machine learning model can effectively predict the vertical displacement of the lower casing of the battery pack system when an electric vehicle bottoms out, thus enabling efficient and low-cost battery pack system design. Furthermore, this machine learning model method can be used in the design of battery safety warning systems to analyze the impact of different bottom impact conditions on battery module safety, such as the electrical and thermal signals and state of charge (SOC) of the battery module. This method can also be used to develop battery module damage prediction models to quantitatively assess damage and thermal runaway phenomena. Battery module damage information is fed back to the battery control unit in real time to ensure that the battery pack system is free of potential safety hazards during vehicle operation.

Claims

1. A method for calculating and predicting the deformation of an electric vehicle battery pack under its base, characterized in that, Includes the following steps: Step 1) Establish the finite element model of the battery pack system and the simulation model of the impact obstacle at the bottom of the battery pack system; Step 2) Test the vertical displacement of the lower shell deformation of the finite element model of the battery pack system under the relevant simulation parameter combinations; Step 3) Construct training data pairs using relevant simulation parameter combinations and corresponding lower shell deformation vertical displacements; Step 4) Train the machine learning model using multiple training data sets to obtain a prediction model for the vertical displacement of the lower shell of the battery pack system. Step 5) Predict the vertical displacement of the battery pack system's lower casing under the relevant simulation parameter combinations using the lower casing deformation vertical displacement prediction model; The steps for establishing a simulation model of the bottom impact obstacle of the battery pack system include: statistical analysis of the bottom impact obstacle of the battery pack system, replacing the impact obstacle with a cone model, and simulating the sharpness of the impact obstacle by changing the size of the top radius of the cone model in order to extract the obstacle features, thereby constructing a simulation model of the bottom impact obstacle of the battery pack system. The relevant simulation parameter combination includes: impact velocity, top radius of the obstacle simulation model, thickness of the lower shell of the battery pack system, and impact angle. The steps for training a machine learning model using multiple training datasets include: Step 4.1) Divide the multiple training data pairs into training sets and test sets; Step 4.2) Optimize the traditional machine learning model using the Cuckoo Search optimization algorithm; Step 4.3) Train the machine learning model using the training set to obtain the trained prediction model of the vertical displacement of the lower shell of the battery pack system. Step 4.4) Test the trained machine learning model using a test set, further adjust the model parameters, and obtain the completed prediction model for the vertical displacement of the lower shell of the battery pack system.

2. The method for calculating and predicting the bottom deformation of an electric vehicle battery pack according to claim 1, characterized in that, The steps to establish a finite element model of a battery pack system include: Step 1.1) Establish a finite element model of the battery pack system based on its housing dimensions, structure, and material; Step 1.2) Establish a finite element model of the battery module based on the size and material of the battery module in the battery pack system; Step 1.3) Based on the connection relationship of each component of the battery pack system, couple the finite element model of the housing and the finite element model of the battery module to obtain the finite element model of the battery pack system.

3. The method for calculating and predicting the bottom deformation of an electric vehicle battery pack according to claim 2, characterized in that, The steps to establish a finite element model of a battery module include: Step 1.2.1) Establish the geometric model of the battery module based on its dimensional parameters; Step 1.2.2) Homogenize the battery module materials; Step 1.2.3) Define the material parameters of the battery module geometric model based on the battery module material information obtained from the homogenization process, and obtain the battery module finite element model.

4. The method for calculating and predicting the bottom deformation of an electric vehicle battery pack according to claim 1, characterized in that, When training a machine learning model or testing a prediction model for the vertical displacement of the lower casing of a battery pack system, the relevant simulation parameter combinations in the training data pair are used as input, and the corresponding vertical displacement of the lower casing of the battery pack system is used as output.

5. The method for calculating and predicting the bottom deformation of an electric vehicle battery pack according to claim 1, characterized in that, The machine learning model includes an input layer, a hidden layer, and an output layer.

6. The method for calculating and predicting the bottom deformation of an electric vehicle battery pack according to claim 1, characterized in that, During training, the machine learning model uses an optimization algorithm to adjust the weights in the hidden layer. Among them, the weights after iterative updates As shown below: （1） In the formula, t is the number of iterations. Let be the weight for the t-th iteration; Heaviside() is the jump function; p is the probability of the optimization algorithm mechanism; , It is a uniformly distributed random number; , To optimize the algorithm, the weights of other connected components on the path are optimized. The jump function Heaviside(x) is shown below: （2）。 7. The method for calculating and predicting the bottom deformation of an electric vehicle battery pack according to claim 1, characterized in that, The vertical displacement prediction model for the lower housing deformation of the battery pack system is shown below: （3） ， （4） ， （5） In the formula, This is the input for layer 0. This is the output of layer 0. For the first Layer input, For the first The weight matrix of the layer, For the first The deviation matrix of the layer, For the first The output of the layer, This represents the total number of hidden and output layers. For the first Activation function of the layer; For the first The output of the layer.

Images