Qi-xiang, “Non-linear characteristics and prediction model of coal and gas outburst”, China university of mining and technology
- YU Wang Kai
Because there were a lot of facts that affect the intensity of coal and gas outburst, a BP neural network model for forecasting the intensity was constructed. Aimed at the shortcoming of the BP neural network, such as the slow training speed, easy to be trapped into the local optimums, and the premature convergence of Genetic Algorithm (GA) BP neural network, a method to design the BP neural network based on Immune Genetic Algorithm was proposed. The mechanisms of diversity maintaining and antibody density regulation exhibited in a biological immune system were introduced into IGA based on genetic algorithm. The proposed algorithm overcame the problems of GA on search efficiency¿individual diversity and premature¿and enhanced the convergent performance effectively. The results show that the IGA-BP neural network have better performance in convergent speed and global convergence, and the forecasting accuracy is improved.