Research On SOH Estimation And Prediction For Lithium-ion Battery Based On Energy Parameters

Lithium-ion batteries is widely used in electric drive and power storage.In order to ensure the reliability of battery,state of health（SOH）estimation is an important function of battery management system（BMS）.Accurate estimation of battery SOH can help users to achieve a better balance between system safety and economic benefits.At present,SOH usually takes capacity or resistance as characterization parameter,while energy cannot be reflected in SOH.Based on the actual operating conditions and degradation mechanism of the battery,the cyclic degradation performance of lithium-ion batteries under different regions is analyzed in this paper,and the method of SOH estimation and prediction is studied.The specific research contents are as follows:Taking 2.75Ah NCM material battery as the research object.Referring to domestic and international testing standards for lithium-ion battery,different partial SOC intervals were divided according to battery practical operating interval,phase transition interval and 20%discharge depth interval.The test protocol of cycle life test and performance test at different interval were designed.The cycle life test results were analyzed,and the three stages of battery energy decline were summarized.Based on the performance test results,the degradation performance of batteries in different intervals were studied.The variation trends of resistance,SOC-OCV curve and increment capacity curve with battery cycle times were summarized.The effects of constant voltage charging process,interval characteristics and phase transition process on battery aging were analyzed.Energy was taken as parameter to characterize SOH,the definition of SOH which characterized by available energy and SOH_ER which characterized by residual cumulative energy were proposed,and SOH_ERdescribe the battery’s remaining life because of SOH_ER has a linear relationship with battery cycle times.The variation trend of increment energy curve with aging of battery was analyzed,the peak position was used to estimate the available energy.SOHER was estimated based on the area equivalent model and other methods,and the error and applicability of the estimation method were analyzed.Based on the degradation curve of cell,a scaling mapping method for estimating battery pack SOH is proposed.The fading feature extraction program and the SOH estimation program were constructed based on MATLAB GUI.LSTM RNN network model is used to predict the inter-partition life of lithium-ion battery.The experimental data were divided into training data and verification data,and the Keras library is used to complete the construction,initialization and training of the cycling aging model of lithium-ion batteries.By inputting the upper and lower limits of the battery interval,the accurate prediction of SOH degradation under arbitrary interval stress is realized.The error is analyzed and the prediction model is improved,and the optimal hyperparametric combination of the model is obtained through searching the parameters such as the number of layers and the size of the model.And the Locally Linear Embedding algorithm is used to fuse the feature parameters and improve the prediction efficiency and accuracy of the model.