Research On Battery State Prediction Of Substation DC System

As an important part of substation,DC system is responsible for power supply of DC load such as protection system and emergency lighting.It is necessary to ensure the reliable and continuous power supply of DC system.In this paper,the optimal maintenance management of DC system battery in substation is studiedFirstly,taking the internal resistance of the battery as the degradation characteristic quantity,a data-driven battery degradation trend prediction method is proposed.Based on the historical data,the nonlinear mapping relationship between the relevant influencing factors and the internal resistance of the battery is established based on the LSTM algorithm.The internal resistance change trend of the battery is predicted according to the floating charge time,floating charge voltage and average charging time.Compared with similar research,this paper uses EEMD decomposition to preprocess the data,which can effectively increase the complexity of the prediction model and improve the accuracy of the prediction results.However,the historical data of battery operation is limited,and it is easy to appear the phenomenon of over fitting in the training process of the model,that is,it shows a high accuracy in the training set But the prediction accuracy is low in the test set.In view of this deficiency,dropout algorithm is introduced to enhance its generalization ability.In the process of training,the probability of single super neuron is reduced by changing the state of neuron,so that the prediction model has good generalization ability.Secondly,the battery health status identification model is constructed based on support vector regression method.The battery health state is identified by the internal resistance of the battery,the operating environment temperature and other factors,namely the real energy storage capacity of the battery.Combined with the characteristics of battery capacity identification,the kernel function and penalty function are optimized respectively to improve the accuracy of identification model for capacity identification.Combined with the results of internal resistance prediction before,the maximum energy storage capacity degradation of battery is predicted.Finally,the battery replacement model is constructed to minimize the battery replacement cost.The fault of battery is divided into accidental fault and inevitable fault.According to the historical operation data and data results of the battery,the accidental failure probability and inevitable failure probability of the battery and the corresponding fault loss are estimated respectively.Considering the operation life of substation primary equipment,fault tolerance of substation wiring and battery fault,the comprehensive cost of battery is estimated.Finally,the actual cases of two substations are used to compare and analyze the battery replacement scheme.It is pointed out that for the substation with high load and relatively long operation time,the battery replacement time should be appropriately advanced to reduce the economic loss caused by the fault.