国产欧美精品一区二区,中文字幕专区在线亚洲,国产精品美女网站在线观看,艾秋果冻传媒2021精品,在线免费一区二区,久久久久久青草大香综合精品,日韩美aaa特级毛片,欧美成人精品午夜免费影视

In view of the high nonlinearity and time-varying of lithium batteries, it is difficult to accurately estimate the remaining power, which affects the management and control of the battery. Based on the BP neural network model, under the random noise interference, the computational time and generalization performance of the battery's remaining power estimation are analyzed and compared with the deep learning model of different architectures, and based on particle swarm optimization (PSO), Nesterov momentum-based RMSProp. The learning rate algorithm optimization model, combined with mathematical programming, designs the optimal framework at different depths and compared with a variety of neural network models. The comparison between the experimental data and the model estimation results shows that the optimization algorithm can effectively reduce the computation time of the model. Under the optimal framework of the double hidden layer, the SOC average estimation error is around 0.1.