Application Of LSTM In SOC Estimation Of Lithium Battery

With the development of the automobile industry,the penetration rate of automobiles has increased year by year,and the problems of environmental pollution and energy shortage have become more and more serious.The advent of new energy electric vehicles with clean energy as the power source has attracted widespread attention from all countries.In the process of continuous development of power batteries,lithium batteries have become the main energy storage device for current electric vehicles due to their high energy density and long cycle life.In order to ensure the safe and reliable operation of electric vehicles,it is important to establish a battery management system to monitor the battery status in real time.The state of charge(SOC)of the power battery is one of the key parameters in the battery management system,which quantifies The remaining power used in the current cycle of the battery is an important basis for the user’s travel planning,and its accurate estimation is conducive to improving battery utilization and prolonging battery life.The thesis verifies the feasibility of Long Short-Term Memory(LSTM)in power battery SOC estimation in a laboratory environment,and extends the LSTM construction model method to the market stock electric vehicle SOC estimation.Taking the ternary material lithium battery(model INR-21700)as the research object,simulating urban driving conditions(FUDS)in a laboratory environment,using the obtained voltage,current and other data as input parameters,and SOC as output parameters to establish the three The long short-term memory network(Long Short-Term Memory,LSTM)estimation model of the cell battery state of charge is compared with two input variables(voltage,current)and four input variables(voltage,current,voltage change rate,current change rate).The accuracy of the model built in this situation.Finally,the electric vehicle driving data with this type of battery as the power source is used as the test set to verify the generalization ability of the model,and it is proved that the LSTM neural network SOC estimation model constructed with laboratory simulation working condition data can be applied to actual vehicle SOC estimation.The paper extends the model method constructed above to the SOC estimation of the market stock electric vehicles.Taking a shared car as the research object,through the actual driving data of the car collected by the battery management system,a LSTM neural network SOC estimation model suitable for the car is constructed.And the actual data of the vehicle under the same operating conditions verify the accuracy of the constructed model,which proves that the LSTM neural network SOC estimation model based on the actual vehicle driving data can be applied to the actual vehicle SOC estimation.Aiming at the stock electric vehicles on the market,the SOC estimation method based on LSTM neural network does not require the establishment of an equivalent circuit model in a laboratory environment.It broke the limitations of the Kalman filter algorithm and provided a theoretical basis for updating and upgrading the SOC estimation method in the stock electric vehicle battery management system.