Breakage Prediction Of Electric Vehicle Battery Module In Rear-end Collision

In order to alleviate environmental pollution and energy crisis,major auto companies actively develop new energy vehicles,such as lithium battery electric vehicles,hybrid electric vehicles,fuel cell electric vehicles,etc.However,electric vehicle accidents frequently occur.for example,due to a collision of an electric vehicle,causing a short circuit of the battery pack and a fire accident or explosion.The accident of battery fire as a thermal runaway.These hinders the development of the electric vehicle.According to statistics,rear-end collisions have been estimated to account for 20–30% of all crashes,and about 10% of all fatal crashes.In order to enhance the battery safety of electric vehicles in rear-end collision accidents,this paper proposes a prediction algorithm model for predicting damage modules after battery collision.The model is based on the relationship between the stress change on the wall of the battery box in the collision and the battery module.The model is based on the stress curve of the sampling point of the battery box wall,and the damage of the battery module is the collision prediction algorithm established by the BP neural network and optimizes the model.The main research methods and contents are as follows:1)Taking a miniature electric vehicle as the prototype,using the finite element method,through meshing,establishing contact,connection and other processes to create a finite element simulation model of electric vehicles in rear-end collision.At different speeds,the model is calculated to extract the stress curve on the geometric center point of the rear face of the battery compartment wall and the damage of the battery module.2)In order to reduce the risk of the collision prediction model falling into local minimum during the training process,the genetic algorithm is used to calculate the optimal initial value of the collision prediction algorithm model.Then,the built-in collision prediction model is trained by serial training algorithm and batch training algorithm.After iterative calculation,the prediction algorithm model for predicting module damage in the rear-end collision is obtained.3)The variable learning ratio method and the additional momentum method are used to optimize the training algorithm of the collision prediction model,which improves the convergence speed and prediction accuracy of the model,and reduces the oscillation of the model during the training process and the risk of falling into thelocal minimum.