Research On Methods For Lithium Ion Battery SOC Estimation And Soh Prediction

The energy crisis and environmental pollution have already sounded the alarm bell to humanity,and vigorously developing new energy sources has become an irresistible trend.Since the iron phosphate lithium ion battery has a high energy,long life,high safety and environmental protection,etc.,its research has become an emerging technology today.The State-of-Charge(SOC)estimation,as the core of the battery management system,requires high real-time performance and is susceptible to various factors,which is a difficult point of the battery management system.The State-of-Heart(SOH)reflects the battery's service life and provides the necessary basis for the timely replacement of the battery.Therefore,the accurate estimation of SOC and SOH is the core and key to the management system.This paper combines the current background and research status of current SOC and SOH estimation of lithium-ion battery,analyzes the working principle,advantages,disadvantages and characteristics of lithium-ion battery.Based on the battery model,it focuses on the research of SOC and SOH of lithium-ion battery.It mainly completes the following two aspects of the study:1.Lithium-ion battery state-of-charge estimates(1)Aiming at the difficulty of real-time requirements for Lithium-ion battery SOC estimation,an SOC estimation strategy based on T-S fuzzy neural network algorithm is proposed.The experimental results under different test conditions show that the algorithm model can maintain a very high estimation accuracy,has a fast calculation efficiency and good self-adaptability,and can better solve the problem of Lithium-ion batteries susceptible to internal and external nonlinear factors.(2)Based on the advantages of ensemble learning methods,an SOC estimation strategy based on random forest regression algorithm is proposed.The experimental results under different test conditions show that the algorithm can effectively solve the problem of overfitting.At the same time,the learning process is fast and the errors can be balanced,which has higher estimation accuracy.The algorithm can evaluate the importance of input variables,and it is convenient for us to improve the prediction accuracy of the model byimproving the accuracy of parameter measurement.2.Lithium-ion battery state-of-health predictionThe capacity of the battery is non-linearly attenuated during use.Based on the advantages of ensemble learning methods,an SOH prediction strategy for lithium-ion batteries based on random forest regression is proposed.Experimental results show that the algorithm has good generalization ability in SOH prediction and can effectively solve the problem of low prediction accuracy.Moreover,it also gets good prediction accuracy in small sample size.Lithium-ion battery SOC and SOH can be well reflected by the model established in this paper.Experimental results show that the model can basically meet the actual simulation requirements,and has good self-adaptability and calculation efficiency.The experimental methods and the main conclusions involved in this paper also have certain commonality for other types of batteries,providing a certain theoretical reference value for future research.