基于BP及其优化神经网络的双电机多挡AMT挡位决策研究

Research of Gear Position Decision of Dual Motor Multiple Speed AMT based on BP and Its Optimization Neural Network

针对传统BP神经网络挡位决策存在的不足,利用三参数换挡规律进行算法和参数优化,得到优化后的BP神经网络。以纯电动汽车双电机多挡AMT为对象,采集试验数据,构建优化前后的神经网络,进行训练和仿真分析。对训练过程的分析说明,优化后的神经网络具有更快的学习速度;对训练后相应模型的仿真分析说明,优化后的神经网络挡位决策模型具有更高的精度。经过优化后的参数可为相应的理论研究提供参考。

By using the three-parameter shift law,the algorithm and parameter optimization are carriedout to overcome the shortcomings of traditional BP neural network in optimized BP neural network of gear posi-tion decision,and then the optimized BP neural network is obtained. Taking the dual motor multiple speedsAMT of pure electric vehicle as the object,the experimental data are collected,the neural network before andafter optimization is constructed,and then the training and simulation analysis are carried out. It can be con-cluded that the neural network has a faster learning speed by training the neural network. By analyzing the train-ing process,it is concluded that the optimized neural network has faster learning speed,and the optimized neu-ral network gear position decision model has higher accuracy through simulation analysis of the correspondingmodel after training,and the optimized parameters provide reference value for the corresponding theoretical re-search.