Study On Fading Characteristics And SOH Prediction For Lithium-ion Batteries

Facing the increasingly prominent energy problems and environmental issues,car development strategy is changing,and the electric vehicle industry and technology in recent years achieves rapid development.Battery is the power source of electric vehicles,and its performance determines the electric vehicle driving range.the state of the battery monitoring is one of the key technologies to electric car safety and reliable driving.SOH is the main indicator of battery decline,and its estimation method has become a research hotspot.The current research on battery decay focused on the effects of temperature,current,cutoff voltage,discharge or charge depth,and the ratio of Ni / Mn to battery degradation.Many researches on the capacity decay model of lithium ion battery are done in recent years,but the existing published research results have not fully revealed the correlation between capacity decay and working condition.In this paper,aiming at the regression characteristics of the lithium-ion battery,SOH prediction and other issues,the main work done as follows:(1)Research on the decline characteristics of lithium ion batteries.Aiming at little research on the relationship between SOC cycle and battery decay,especially in the case of quantitative analysis and mechanism studies,a number of cycle tests based on different SOC cycle intervals and current rates have been carried out.The effects of these two factors on the battery decline are analyzed from three aspects: capacity,internal resistance and decay mechanism.In order to more fully study the impact of battery decline factors,the NASA battery data was also analyzed,and the effects of temperature,discharge rate and discharge cutoff voltage on the battery degradation were studied.The results show that the higher the current magnification,the higher the cycle interval,the higher the temperature,the lower the discharge cutoff voltage,the faster the battery fading.A large number of battery decay data were collected.The experimental data includes the parameters such as capacity and internal resistance,which are directly related to the SOH of the battery.The indirect parameters such as the voltage drop discharge time are also included.All the direct and indirect parameters lay the foundation for the battery decay model clustering and SOH prediction.(2)Research on battery capacity and internal resistance prediction method based on battery decay feature clustering and SOH variation feature learning.Aiming at the shortcomings of current SOH prediction accuracy and the large amount of computation,after research on wavelet transform and neural network algorithm,the battery capacity and internal resistance decline prediction method based on battery decay feature clustering and SOH variation feature learning are proposed.Kohonen neural network is used to accumulate the battery decay pattern.Then,cluster results guide the training of the wavelet neural network,which is used to predict the decline of the battery in the next stage.Accordinig to the above algorithm,the prediction of battery capacity decline and the prediction of internal resistance decline are realized by programming respectively.Compared with other neural network algorithms,the accuracy of prediction is improved while the calculation amount is reduced.The maximum predicted error is 4%.Research on the indirect prediction method based on health indicator of battery SOH.The health indicator of battery SOH(time to equal discharge voltage difference)is researched.The relationship model between the discharge time and the SOH is established.The SOH is predicted by the predicted discharge time indirectly.With this method,the higher accuracy of prediction is obtained,and the difficulty of parameter acquisition is reduced and the practicability of the algorithm is improved.(3)Research on SOC estimation method of battery based on SOH correction.Aiming at the current situation of the SOH effect is rarely considered in SOC estimation,the SOC estimation method based on SOH correction and the BP neural network optimized by MEA is proposed.With the current,voltage and temperature as the input of the neural network,the output of the neural network is battery capacity,and the initial weights and thresholds of the BP neural network are optimized by the MEA algorithm.After the training of the neural network,the estimated error is less than 3%.On this basis,the SOC estimation method based on SOH correction is proposed,and the SOH is added to the input of the neural network,which further widens the application range of the algorithm.The algorithm is applicable to the SOC estimation of both the new and old battery.