Artificial Neural Network Algorithm Based on Kalman Filter and It's Application in the Mechanical Activation of Minerals

In order to overcome the disadvantage of conventional BP algorithm, this paper has developed a improved algorithm, which based on the Kakman filter algorithm and lead two parameters: neural net gain g_i and state momentum factor γ_m into the iterative process for the purpose of smoothing the weigh matrix W. As a result, this algorithm can increase the learning and convergence rate and improve the stability of BP network performance. The practical application has shown that this algorithm is superior to any other algorighms in the field of Mechanical Activation of Minerals, and has the ability of solving practical problem.