Improving Elman neural network model via fusion of new feedback mechanism and Genetic Algorithm

In this paper, we propose an improved Elman neural network model which contains a new feedback mechanism composed of a special external feedback we proposed and inherent internal feedback. In order to guarantee the generalization ability of the established model, we adopt Genetic Algorithm to optimize initial connection weights and number of hidden layer nodes at the same time. This kind of improved Elman neural network model is applicable for time series prediction and we verify our model on the hourly air quality dataset. Comparing with two different neural network models, we reach the conclusion that the improved Elman neural network performs better than BPNN and traditional Elman neural network in terms of accuracy and convergence speed in experiment. And the improved Elman neural network shows a better stability in different time periods in the prediction process.