Prediction Of Remaining Useful Life Of Lithium-ion Battery Based On New Health Indicators

As a portable energy source,lithium-ion batteries are rapidly popularized with its advantages of lightweight,high energy density,high chemical reactivity and long life.It plays an important role in the fields of consumer electronics（mobile phones and laptops）,hybrid and electric vehicles in the automotive industry and space exploration.However,with the increase of charging and discharging times,lithium-ion batteries will inevitably degrade and even fail,if effective measures cannot be taken before the battery failure,the lithium-ion battery equipment will not be able to operate healthily,which may cause casualties in severe cases.The prediction of remaining useful life（RUL）of lithium-ion battery can effectively avoid the harm caused by battery failure or failure,which is conducive to solving the above safety problems.As a complex electrochemical system,the performance of lithium-ion battery will gradually degrade with use.However,in practical application,the capacity or internal resistance parameters characterizing its degradation state are difficult to monitor directly,and the health indicators（HI）extracted from the traditional charge discharge curve of lithium-ion battery,such as charge discharge voltage,charge discharge current and other parameters,although they can provide some information,they lack a more comprehensive understanding of the potential physical mechanism of lithium-ion battery degradation.In view of this,this paper extracts the health indicators that contain more comprehensive information and are easy to calculate than the traditional charge discharge curve to reflect the degradation state of lithium-ion battery,and uses the Gaussian process regression（GPR）method with uncertain expression to predict the capacity of lithium-ion battery,which solves the difficulty of internal parameter measurement in the process of RUL prediction of lithium-ion battery.The extraction process of health indicators is complex,and the extracted features contain incomplete information and lack of uncertain expression of prediction results.At the same time,it provides a new technical idea for the application of RUL prediction method of lithium-ion battery.The main research ideas of this paper mainly include the following parts:（1）Aiming at the problems of difficult measurement of internal parameters,complex extraction process of health indicators and incomplete information contained in the extracted features in the process of RUL prediction of lithium-ion battery,this paper uses incremental capacity analysis（ICA）to analyze the degradation state of lithium-ion battery,and uses the parameters such as voltage and current that are easy to be measured online,The slowly rising voltage plateau period is converted into a capacity increment curve with obvious characteristics,which directly reflects the violent chemical reaction changes in the battery.Then,the peak and sub peak area are extracted from the IC curve as HI to reflect the battery health state,and the proposed HI is analyzed qualitatively and quantitatively.Finally,based on the combination of ICA and GPR,a new RUL prediction method ICA-GPR for lithium-ion battery is proposed,and the uncertainty expression of the prediction results is given.（2）Aiming at the problem that the information contained in the features extracted in the degradation process of lithium-ion battery is not comprehensive enough,a differential thermal voltammetry method（DTV）is proposed to analyze the degradation state of lithium-ion battery.DTV method contains more entropy information,which can make a more comprehensive understanding of the potential physical mechanism of lithium-ion battery degradation,and the information similar to ICA method can be inferred only by using simple temperature measurement.Therefore,the peak height ratio and the area under the peak are extracted from the DTV curve as HI to reflect the battery health state,and the extracted hi is analyzed qualitatively and quantitatively.Finally,based on the combination of DTV and GPR,a new RUL prediction method DTV-GPR for lithium-ion battery is proposed,and the uncertainty expression of the prediction results is given.（3）According to the research content of this paper,a prediction system for the remaining useful life of lithium-ion battery based on Gaussian process regression is designed.The prediction based on ICA-GPR and DTV-GPR are integrated into one system,which meets the requirements of combining theory and practice.