Driving Pattern Based Prognosis Of Lithium-ion Battery Capacity Degradation

As the core of an electric vehicle,the performance of on-board power battery has a decisive impact on the driving range as well as the total cost of an electric vehicle.Meanwhile,power battery is still a major bottleneck for the further market diffusion of electric vehicles,its lifecycle degradation of dischargable capacity has a tremendous impact on the driving range as well as the total cost of ownership for the users.This research focuses on the three major topics of battery degradation prognosis: the online capacity estimation,the battery degradation modeling and the battery working pattern prediction.In this way,a battery capacity degradation prognosis method based on the battery real working patterns is developed,which laid the foundation for the battery life monitoring and prognosis.To study the battery working pattern of electric vehicle in operation,the driving patterns of operating electric logistic vehicles are collected and the data-base of battery working pattern is established.Through a 4-month-lasting battery degradation experiment under dynamic working pattern,the working pattern and degradation pattern of 18650 batteries are examined,and the degradation mechanism is analyzed.To conduct the online battery capacity estimation,a method of resemble characteristic identification of charging curve based on the incremental capacity analysis is proposed after a through analysis of the battery degradation mechanism.Firstly,the resembleness of the charging curves are verified by using charging voltage consistency assumption.Then,characteristics are selected based on the incremental capacity analysis and correlation analysis.Finally,the estimation model is established and modified through with least square algorism estimation with forgetting factors.The results of the capacity estimation of two test batteries show that the overall estimation error is lower than 2.5%,indicating that the model has a good reliability.To establish the battery degradation model,a duo-temperature cycling experiment lasting 3 month is conducted.In this way,an Arrenius model based degradation model is established.By integrating the outcomes of online capacity estimation as well as applying PI closed-loop modification method,the historical degradation estimation error in steady state lowered to about 1%.To predict the future working patter of the battery,a prediction method based on markov chain is proposed.Through state space generation,transformation principle regulation,cycle generation and verification,a characteristic working cycle that represents future working patterns of the battery is generated.The estimation method achieved an estimation error lower than 10% while remaining high calculation efficiency.Through the systematic integration of of battery online capacity estimation,degredation modeling and future working pattern prediction,the battery capacity degradation prognosis method has proved to have an accurate progonosis,with a steady state error lower than 2% while having a good reliability.