Research On Sorting Method Of Zinc Silver Battery On FCM Least Squares Support Vector Machine

With the development of deep-sea equipment,national defense and aerospace,zinc and silver batteries are widely used because of their high specific energy and high specific power.It usually requires dozens or even hundreds of zinc and silver monomers in series and parallel to use in groups.However,due to the differences between the zinc and silver monomers,the continuous charge and discharge cycles magnify the differences between the zinc and silver batteries,resulting in the reduction of the capacity and lifetime of the zinc silver batteries,and even the safety problems caused by the continuous charge and discharge cycles.Therefore,it is very important to improve the consistency of initial performance of single cell by sorting before grouping zinc and silver batteries.This dissertation first introduces the research background and significance of the subject,analyzes the development status of Zn-Ag battery and the research status of battery sorting,studies the working principle and characteristics of ZnAg battery,and summarizes the reasons,forms and solutions of the inconsistency of Zn-Ag battery.Secondly,the basic theory of support vector machine(Support vector machine,SVM)is introduced,from linear standard SVM to nonlinear standard SVM,from two-classification problem to solving multi-classification problem.The least-squares support vector machine(Least square-Support vector machine,LS-SVM)battery sorting method is studied.In this method,equality constraints are used to replace the inequality constraints of standard SVM,and the quadratic programming problem is transformed into linear equations.The computational complexity is reduced,the solution speed is faster and the robustness is better.In order to solve the limitation of using prior knowledge classification to establish LS-SVM model and improve the accuracy of the model,a LS-SVM model building method based on Fuzzy C-means is proposed.The FCM clustering algorithm is studied.The battery sample classification is obtained by FCM clustering algorithm.The LS-SVM model is trained and tested by using the classification results.The performance of different methods is compared by simulation,and the superiority of LS-SVM is verified.Finally,the test platform of zinc-silver battery separation is introduced,and the capacity attenuation experiment is designed.The experimental results show that the prediction method of LS-SVM model based on FCM can quickly and effectively separate zinc and silver and improve the dynamic consistency of battery pack.