Prediction Of Remaining Useful Life Of Lithium-ion Battery Based On Data-driven

Lithium-ion batteries have been widely used because of their high energy density,long cycle life,light weight and low self-discharge rate,especially in the field of electric vehicles.As one of the three core components of electric vehicles,the performance of lithium battery used online for a long time will inevitably degrade,and its life and stability will decrease.In order to improve the battery online service cycle and ensure the reliable,safe and efficient operation of the battery system,it is of great significance to accurately predict the Remaining Useful life(RUL)of lithium-ion batteries.In this paper,the RUL prediction of lithium-ion batteries is studied as follows:Firstly,the working principle of lithium battery is introduced from the working characteristics of each component of lithium battery.On this basis,the degradation of lithium battery caused by the degradation of each component is analyzed.Taking B5 battery data in NASA PCo E lithium battery dataset as an example,the changes of charging current and charging voltage parameters in the degradation process of lithium battery are introduced,and then the degradation characteristics of lithium battery in the charging stage are constructed,and the correlation analysis is used to evaluate the degradation characteristics.After comparison,the most practical degradation characteristics are selected as the health indicators.Secondly,in view of the nonlinearity and non-stationarity of the local regeneration component and noise fluctuation component of the degradation characteristics caused by the use of lithium battery,the singular spectrum analysis method is used to identify and separate the main degradation characteristics and noise signals contained in the health indicators to obtain the reconstructed health indicators.The validity and accuracy of the proposed reconstruction health index for lithium battery capacity estimation are verified by using the data of CS2-35 and CS2-36 batteries in the CALCE lithium battery data set of the University of Maryland and the data of B5 and B6 batteries in the NASA PCo E lithium battery data set.Finally,the RUL prediction model of lithium-ion battery based on1DCNN-LSTM and polynomial regression was established.Using B5 and B6 battery data in NASA PCo E lithium battery data set and CS2-35 and CS2-36 battery data in CALCE lithium battery data set of the University of Maryland to verify the accuracy of RUL prediction of the combined model,the experimental results show that the combined model shows good accuracy and generalization while ensuring practicability.