Joint Estimation Of Li-ion Battery State Of Charge And State Of Health Based On Mogrifier LSTM-CNN

With the rapid development of new energy vehicles,lithium-ion battery technology has become the focus of research.State of charge and state of health are important parameters of lithium-ion batteries.Accurate estimation of SOC and SOH can avoid overcharge and discharge of batteries,which is of great significance to the healthy use of batteries.In this paper,the joint estimation method of SOC and SOH for lithium-ion battery is studied.The specific work is as follows:Firstly,the definition and research status of SOC and SOH are introduced,the composition and working principle of lithium-ion battery are analyzed,the charging and discharging characteristics of lithium-ion battery are studied,and the effects of temperature and discharge rate on SOC and SOH are explored.Finally,temperature and discharge rate were determined as the influencing factors of SOC and SOH estimation.Secondly,on the basis of long short-term memory recurrent neural network,a method for SOC estimation of lithium-ion battery based on mogrifier LSTM is proposed.In this method,two gating units are added to the original LSTM neural network to establish a more abundant interaction space between the input and output of the network.LSTM and mogrifier LSTM neural networks are built in Py Torch deep learning framework,and the network super parameters are optimized.The SOC estimation performance of the proposed method is tested on constant current,pulse and NASA random usage data sets at different temperatures and conditions.Experimental results show that,compared with the original LSTM algorithm,the improved algorithm can achieve higher SOC estimation accuracy under different temperatures and different discharge conditions,which verifies the robustness and applicability of the proposed algorithm.The experimental data of different time periods are selected from NASA data set to train mogrifier LSTM neural network,and the actual impact of battery aging on SOC estimation is explored.The experimental results show that the battery aging has a negative impact on the SOC estimation of lithium-ion battery in the whole life cycle.Finally,it is determined that SOH should be taken into account in the process of SOC estimation.Finally,a joint estimation method based on mogrifier LSTM and convolutional neural network is proposed.This method uses the voltage,current and temperature of lithium-ion battery to realize the joint estimation of SOC and SOH in the whole life cycle of the battery.Because the proposed joint estimation method takes SOH into account in the process of SOC estimation,the adverse effect of battery aging on SOC estimation is eliminated.The robustness and applicability of the proposed joint SOC and SOH estimation method are verified on NASA random usage data set and Oxford aging data set.The experimental results show that the proposed joint estimation method can achieve the joint estimation of SOC and SOH under different temperatures and conditions,and obtain high accuracy.