Fuzzy, ANFIS and ICA Trained Neural Network Modeling of Ni-Cd Batteries Using Experimental Data

Nowadays, because of the many applications of batteries, modeling of this energy sources and accurate estimation of their behavior at different conditions is very important. In the traditional linear mathematical based models of batteries, nonlinear behavior of them and the variation of battery voltage during charge and discharge states are neglected. But various nonlinear physical, chemical and electrochemical factors should be considered to achieve accurate battery model. However, this is very difficult or even impossible to calculate that factors. Therefore, in this paper, in order to provide more accurate model and correct estimation of battery voltage during charge states, three nonlinear modeling methods named Fuzzy Logic, Adaptive Neuro Fuzzy Inference System (ANFIS) and Imperialist Competitive Algorithm (ICA) trained Neural Network are proposed for Ni-Cd battery modeling. The main advantage of the proposed models is that they can predict battery output voltage with no knowledge of numerous factors. In order to collect the required data for training of proposed models and to compare models by actual data, experimental data obtained from tests on a 7Ah, size F, Ni-Cd battery at different charge current. Simulation results are compared with the measured battery data at different charge current. The simulations show good agreements with measured data. Also the proposed models can use for other battery types.