Grey BP neural network based on error grade iterative and its application

An improved BP neural network based on grey theory and error grade iterative method is proposed. The accuracy of BP neural network is improved by grey accumulating generation operator reducing the randomness of training samples. Using the grey correlation analysis method to analyze the number of the hidden node for BP neural network, this method realizes the optimization of BP neural network structure. The error grade iterative method improves the convergence speed of BP neural network. At last, it verifies its performance by applying to random drift model of flexible gyroscope and comparing with the normal BP neural network.