Battery life assessment based on discharge voltage platform research
To address the issue of certain batteries in traditional EMUs experiencing performance degradation prior to meeting repair conditions due to the repair method involving nickel-cadmium batteries,as well as the problem of a significant number of batteries meeting repair standards but not exhibiting excessive performance decline.The whole life accelerated aging experiment of single nickel cadmium battery was designed and the relevant experimental data were obtained.Firstly,the integrated empirical mode method was used to establish the classification model of the whole life health state of the single battery.Then,the discrete wavelet transform was used to eliminate the singular value of the discharge voltage platform data,and then the extreme learning machine algorithm was used to predict the life state of the battery.Finally,the accurate prediction and health state assessment function of the whole life cycle of the battery were realized.The experimental results show that compared with the traditional battery life threshold classification method,the health status classification model established by ensemble empirical mode can effectively avoid false alarms at the end of battery life.The discharge voltage platform data as the input of the fusion algorithm model is easy to obtain.The data preprocessing method based on discrete wavelet transform can improve the accuracy of the algorithm by nearly 3%,and finally reach 96%～98%.In addition,compared with the traditional neural network model,the fusion algorithm model does not involve iteration.It can take into account the prediction accuracy and computational efficiency of the algorithm.The F1 value for identifying the health status of the battery is 0.976 3,the F1 value for identifying the aging stage is 0.950 9,and the F1 value for identifying the fault stage is 0.939 394.Compared with the traditional method of determining whether the battery should be repaired based on the operating mileage and service life of the EMU,the fusion algorithm model provides a significant evaluation criterion,effectively identifies the health status of the battery,reduces the operating cost of the EMU,and ensures the safe operation of the EMU,providing a reference for battery life evaluation and maintenance strategy optimization.
life assessmentintegrated empirical mode decompositiondiscrete wavelet transformextreme learning machinedischarge voltage platformonline detection
万方数据
维普
