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

In the context of the global active development of renewable resources,lithium-ion batteries as a recyclable green new energy source have received great attention and rapid development.At the same time,the health management of lithium-ion batteries is becoming more and more important,among which the state of health estimation and remaining useful life prediction of lithium-ion batteries are the key issues.Based on the performance data of lithium-ion batteries,new health indicators representing the state of lithium-ion batteries are constructed through data analysis and models are established for prediction of remaining useful life.The specific work is as follows:Firstly,the working principle and degradation mechanism of lithium-ion batteries are analyzed.Combined with the performance data of lithium-ion batteries from the NASA PCo E Research Center,the health indicators that can characterize the state of lithium-ion batteries are described in detail,and a statistical description of the capacity degradation characteristics of lithium-ion batteries during charging and discharging is given.Secondly,the change law of voltage data during the discharging process of lithium-ion batteries is studied,and the exponential and logarithmic function models are established respectively according to the characteristics of the voltage change during the discharging phase and the self-recycling phase of lithium-ion batteries.Therefore,two new health indicators that can characterize the state of lithium-ion batteries online have been constructed.At the same time,linear regression models based on the new health indicators are established to estimate battery capacity,and the accuracy of models is corrected by Box-Cox transformation.Furthermore,the lithium-ion batteries degradation model based on new health indicators is established,and the remaining useful life prediction algorithm of lithium-ion batteries is designed based on relevance vector machine.The data results show that the constructed new health indicators can accurately estimate battery capacity and predict remaining useful life.Finally,the characteristics of current and voltage changes during the lithium-ion batteries charging process are analyzed.The charging current difference and charging voltage difference at different time intervals are extracted as two health indicators.The correlation verification between the proposed indicators and battery capacity is given,and a two-indicators linear regression model is established to estimate the capacity of lithium-ion batteries.On this basis,BP neural network and particle swarm optimization are combined to design the state of health estimation algorithm of lithium-ion batteries.Considering that there is a certain mapping relationship between the state of health and the remaining useful life of lithium-ion batteries,the remaining useful life prediction algorithm of lithium-ion batteries is designed.The data results show that the use of charge current difference and charge voltage difference together can achieve an accurate estimation of battery capacity and an effective prediction of remaining useful life.