Application of PSO-LSTM Algorithm to Predict the SOC State of Lithium Batteries

Due to the fact that the lifespan of lithium-ion batteries in new energy vehicles can lead to a gradual decline in their performance, accurate prediction of SOC status is of great significance for the safe use of batteries. By organizing and summarizing the battery status of lithium batteries under charging and discharging conditions in different temperature environments, a lithium battery SOC status database is established; Using the basic principles of standard particle swarm optimization algorithm and the prediction model established by long short-term memory neural network, this study investigates the fluctuation of SOC state error. PSO-LSTM is a high-precision prediction model. Compared with traditional neural network models, the improved neural network can improve the accuracy of fault prediction, and the actual error fluctuations tend to be more prone to prediction fluctuations. The PSO-LSTM algorithm model is better than traditional neural networks. It can be seen that the PSO-LSTM algorithm model has good performance as a tool for predicting the SOC state of lithium batteries.