Nickel/Metal-Hydride Batteries Intelligent Control Reserch Based On Elman Network

MH-Ni battery as a rechargeable battery, because of their small resistance, large discharge current, voltage stability, long life, no memory effect, no heavy metal pollution, has been widely used. In the discharge process, the working situation of the battery will change with the different environment change, so it has great significance to set up the residual capacity evaluating model reflecting the dynamic characteristics of the batteries. In the charging process, the inner electrochemistrial reaction of MH-Ni battery is a complex nonlinear process, whose mathematic mode is very complex. So it has great significance to look for a better model of the network, to effectively set up model, to adopt the right control methods and to improve the rechargeable battery technology of the MH-Ni battery.At present we mostly adopt the static feed forward BP neural network and RBF neural network which is based on BP algorithm in industrial application. It eventually changes dynamic time construction model problem into static space construction model problem when using static feed forward network to distinguish dynamic system, so many questions will come out. Elman network is a kind of dynamic neural network, it has ability to mapping dynamic characters by memorizing inner statement on the basis of the basic structure of feed forward neural network, and it represents the orientation of the neural network construction model and control. So we adopt Elman network in this paper.Firstly, on the basis of analyzing the various of factors that impact the residual capacity of the MH-Ni battery and synthesizing several ways of predicting the MH-Ni battery, we research the problem of the MH-Ni battery capacity predicting. Taking into account the problems of the BP algorithm such as the slow convergence and getting into the local minimum point easily, we adopt genetic algorithm (GA) to optimize the Elman network model, the simulation results show that the method improved the accuracy of the predicting. Secondly, on the basis of analyzing the charging characteristics of the MH-Ni battery, utilizing these characters for Elman network of approaching any function in the limited time in any arbitrary precision, producing space mode and time mode after training, in this paper, we adopt neural network model reference adaptive control to design the intelligent charging system of the MH-Ni battery.The simulation results show that the charging time is shorten, the charging efficiency is improved, the battery life is extended by adopting the method, meeting the requirements of the industrial control.