Capacity failure prediction of lithium batteries for vehicles based on large data

With the vigorous development of electric vehicle, lithium-ion battery as the main core component of electric vehicle, accurate and effective prediction of its failure and timely replacement before lithium-ion battery failure can effectively guarantee the safety of vehicle and personnel and avoid major accidents. Based on the cycling test data of a lithium battery for a vehicle, a grey model algorithm is proposed to predict the capacity degradation of the battery in this paper. Considering the strong time-varying non-linearity of battery capacity decay and disturbance of external noise, a method of optimizing data improvement accuracy by wavelet threshold denoising is proposed. The obtained capacity decay data is taken as model training sample set to further improve the accuracy of grey model in predicting capacity degradation failure of lithium ion batteries. The feasibility and validity of the optimization model algorithm are verified by simulation experiments on two sets of lithium ion battery capacity data sets with different fading trends.