Research On On-Line Diagnosis Method Of Valve Regulated Lead-Acid Battery Capability

Valve regulated lead acid(VRLA)battery is an important part of backup power system,and its reliability is related to the safety and stability of electrical equipment operation.Actually,most VRLA batteries cannot reach its lifetime for the reason that batteries are always in online float-charging state and be short of effective maintenance.With the decline of battery power,blackout accidents cause serious loss frequently occurred.At present,the only recognized and reliable detection method for battery state of charge(SOC)and state of health(SOH)is the complete load discharge test,but the method is time-consuming and off-line detection.However,the off-line detection method of VRLA battery is not practical for its application characteristics in backup power system determine.Because of the many factors affecting the performance of VRLA battery,and showing high nonlinear characteristics,it is a difficult problem to detect the SOC and SOH of the battery on line.The advantages and disadvantages of the existing methods of battery performance detection at home and abroad have been summarized and analyzed firstly.Then,through the in-depth study of the working principle,failure mechanism and factors affecting VRLA battery performance,three external characteristic parameters including the terminal voltage,current and temperature are determined,which are cheap and easy to be on-line measured on a large scale.In the end,two methods of on-line detecting battery SOC and SOH on line are proposed.1.The Coup de Fouet("voltage steep drop and the recovery")phenomenon occurs at the initial stage of VRLA battery discharge which is full charged.Based on the corresponding relationship between the trough voltage and the plateau voltage and the actual capacity of the battery under different discharge conditions in the Coup de Fouet phenomenon,the mathematical model for estimating SOH is established.Then,the training for the model is carried out by the BP neural network.The simulation results show that the model can accurately classify healthy and deteriorated batteries.Compared with traditional detection methods,this method is safe,convenient and efficient,and has broad application prospects.2.The mathematical model for estimating battery SOC is established based on the least squares support vector machine,and the ant colony algorithm is introduced to optimize the model parameters for more prediction accuracy.The battery SOC is estimated only with three external parameters(operating voltage,current,temperature).The simulation result shows that the precision of the SOC estimation model is above 98.4%.Through the cross test of sample data,it can be found that the model has strong generalization ability for the estimation of SOC in the whole range,which verifies the feasibility of the method and provides a powerful theoretical guidance and experimental basis for the SOC estimation of the VRLA battery in the backup power system.