The mill load modeling of combined grinding system based on RBF neural networks

In order to get the mill load modeling of combined grinding system in normal working condition, this paper proposes a method based on the RBF neural network. The neural network uses three kinds of kernel functions that are Gauss kernel function, multiquadric kernel function and inverse multinuclear kernel function. Using the gradient descent method trains the neural network. With the comparison of three neural network´s fitting error, I´ve come to the conclusion that the RBF neural network based on Gauss function is more accurate.