Research On The Method Of Predicting The Remaining Life Of Lithium-ion Batteries

Nowadays,with the rapid development of society and improve of constantly scientific and technological level,energy problems begin to occupy the important position in the development of the society,so the lithium-ion battery is widely used as important energy storage device with production and life,and lithium-ion battery health management and useful life prediction problem began to become the focus of scholars attention.The prediction of the remaining useful life of lithium-ion battery is of major significance for evaluating the battery reliability and maintaining the safety and stability of the system.For the prediction of remaining useful life of lithium-ion batteries,this paper combines theoretical methods with practice,and mainly does the following work:First of all,the degradation status of lithium-ion batteries,which can identify by finding the right health factor characterization of battery state of degradation,decided to lithium-ion battery remaining life prediction effect is good or bad the primary factor is the selection of lithium-ion battery health factor,due to the capacity as a battery of health factors cannot be obtained by measuring directly,this article selects three characteristics to build health factor,and through the contrast with the result of capacity estimation filter out the appropriate features as lithium-ion battery of indirect health factor.Secondly,the thesis mainly aims at the specific of the lithium-ion battery remaining useful life prediction methods are discussed,the particle swarm is used to support vector regression machine to the optimal values of the three key parameters of selection,but as a result of its sometimes fall into local extremum,chaos optimization algorithm is introduced,through has the characteristics of ergodicity of chaos variable to improve the standard particle swarm optimization algorithm,so it can obtain the global optimal value,and the improved chaotic particle swarm optimization support vector regression machine prediction model is applied in the lithium-ion battery remaining life prediction.Then the support vector regression model of chaos particle swarm optimization is used as the weak regression of AdaBoost model for integrated learning,and the strong regression model is constructed by combining several weak regressors,so that the prediction accuracy of lithium-ion battery life can be further improved.Finally,the feasibility of the proposed method is proved by the simulation experiment with the open data set.Finally,to aging experiments of 18650 lithium-ion battery,by repeatedly charging and discharging experiments data collected by the battery state of degradation,which can identify and use the method of the paper was carried out on the battery life prediction,and the paper puts forward the model and other models for lithium-ion battery remaining life prediction results comparing,the feasibility of the method of the paper.