A neural network method for estimation of battery available capacity

Rechargeable batteries are widely used in many electrical systems to store and deliver energy. In order to use batteries more efficiently, their response to various operating conditions must be understood. There are many ways to define battery capacity include: rated capacity, SOC and BAC. Knowing the battery available capacity (BAC) in electric vehicles (EVs) is very important issue. BAC depends on discharge current and battery temperature for different kinds of battery but there is no exact relationship between BAC and discharge current in various temperatures. Recently neural networks have been successful used for power system applications. In the literature, there are many neural networks for power system applications. However, the multilayer perceptron (MLP) and the radial basis function (RBF) have demonstrated better capabilities. This paper presents two neural networks (RBF and MLP) for estimation of Lead-Acid BAC. The main contribution of this paper is consideration of temperature effect in BAC calculation. In addition, the results of RBF and MLP are compared.