Multi-Objective Optimization Design Of Automotive Power Battery Pack Structure Based On Firefly Algorithm

Battery packs are the power core of pure electric vehicles,requiring them to be lightweight,anti-resonance and have sufficient strength when processing and designing.From the perspective of vehicle weight distribution,the battery pack accounts for 26%-35% of the vehicle weight,and for every 5 kg of weight of the battery pack,the cruising range will be reduced by about 5 kilometers,which seriously affects the improvement of the cruising range of pure electric vehicles,and also greatly hinders the further development of pure electric vehicles,so the optimized design of the battery pack is extremely necessary.In this paper,the research focus is set on the optimal selection of the processing parameters of the lithium battery shell of a small pure electric vehicle and the enhanced optimization design of the structure of the vulnerable part of the lower shell,and the two sets of optimization processes are designed by finite element analysis,constrained modal experimental model verification,processing parameter neural network prediction,bionic structure optimization design and variable density topology optimization to solve the endurance problem,and the main research contents include:(1)The three-dimensional modeling and constrained mode experiments of the battery pack were carried out,and the model rationality was verified by comparison,which laid a foundation for the subsequent research.In this paper,the weight and threshold of the backpropagation neural network model are optimized by using the firefly optimization algorithm,and the optimal solution is output by simulating the firefly predation scheme and assigning it as a new parameter of the neural network.An N-FA-BP(Firefly-Backpropagation)network model is proposed,and the fitting and prediction performance of 185 sets of data sets of 13 neural network models such as random forest,support vector machine and wavelet base are compared with 7 evaluation indicators such as RMSE(root mean square error),MSE(mean square error)and 13 neural network models,which proves its better prediction effect.(2)In order to further improve the prediction accuracy,the different hidden layer topology of N-FA-BP was compared,7 design variables were obtained for the size parameters of the lithium battery shell by Monte Carlo method,and the parameter regression prediction was made for the structure after the variable density topology optimization of the anisotropic material interpolation model SIMP(isotropic material)using the moving asymptotic method using the Min GW-w64 compiler,and the 1-3-1 neural network model was selected to predict the size of the battery shell structure.The error of predicting each processing dimension of the overall imitation Π-shaped battery shell is within ±3%,which verifies its good prediction accuracy.The 3σ random vibration calculation of the optimized battery shell structure shows that the optimized structure still has sufficient safety factor,and the value of each design variable is [1,1.5,4.5,7.5,8,7,8](mm).The result is a corresponding 9.1% increase in frequency and a weight reduction rate of 23.6%.(3)With the rapid development of pure electric vehicles,research has focused on improving range.Due to the high mass of it,its structure needs to be designed and optimized while reducing the mass.In this thesis,the shell model of the new energy vehicle battery pack is reconstructed by improving the top shell material and adding reinforcements along the direction of the large leaf vein fibers in the leaf park.A quasi-Monte-Carlo method based on Sobol series and Latin hypercube design with variance sensitivity analysis were used to determine seven design variables.The122 datasets were trained and predicted using the basic gradient descent algorithm combined with the conjugate direction method,and the predictions were compared with static mechanical simulations under sharp bending conditions on a bumpy road.The results show that the battery pack case for new energy vehicles results in a weight reduction of 19.5%,a maximum stress reduction of27.49% and a displacement reduction of 29.29%,which meet the material requirements.First order modal frequency is increased by 6.3%,effectively preventing resonance with the road surface.