An improved model based on artificial neural networks and Thevenin model for nickel metal hydride power battery

Based on artificial neural networks and Thenvenin model, this paper uses an improved model predicting state of charge. We combine artificial neural networks model with Thevenin model, and predict state of charge in real time at the same time. When the difference between the predictive value of artificial neural networks model and the predictive value of Thevenin model is more than 10%, we revised the predictive value of artificial neural networks model by weighted average value. The results show that it can reduce the error of artificial neural networks model obvious and the average error is 4.72%. It is lower independence on initial state of charge than artificial neural networks model.