Model Identification Of Lead-acid Batteries Based On Adaptive Neuro-fuzzy Inference System

The state-of-charge (SOC) of lead-acid batteries has a critica l influence on stably andhealthy operation of batteries. How to timely and accurately acquire the SOC of leadacidbatteries is one of key technologies in monitoring and ma nagement system of leadacidbatteries. At present, acquiring the SOC of batteries using nerve-fuzzy and softcomputing with combination of internal and external key parameters of batteries hasgradua lly become a hot spot in the study of the battery's monitoring and ma nagementsystem.With the analysis of electrochemica l properties of lead-acid batteries and commondetection method of the SOC, the thesis determines the key parameters with impact onthe SOC , and proposes an approach of performa nce parameter identification of batteries,which based on adaptive neuro-fuzzy inference system.With the study of some theories about lead-acid batteries , it is found that theelectrolyte density is the key parameter in the impact on the SOC. The Ag/AgSO4reference electrode was produced, and we use it to obtain positive and nega tive electrodepotentia l in charging and discharging experiments . Using BP networks, radia l basisfunction network (RBF) and adaptive neuro - fuzzy system (ANFIS) for identifica tion ofthe density, the training errors of the results meet the requirements。And the methods canobtain the SOC through the parameters of termina l voltage, current, positive and nega tiveelectric potentia l. By analyzing and comparing the three identifica tion methods, wechoose the method called ANFIS, which has a strong adaptive and genera lization ability.And the accuracy of simulation in MATLAB is idea l for the smart battery monitoringsystem. Finally the thesis has a further analysis of detection algorithm based on theANFIS of the lead-acid batteries .