关键词: 螺旋分级机, 数学模型, 径向基函数神经网络, 遗传算法

Abstract: The process of grinding circuit is multivariable and non-linear，the mathematic models based on the linear regression are hard to satisfy the requesting of accuracy.The fast developing neural networks technique can build a high precision model for the non-linear reflection of variables.Based on the elementary mathematic models about spiral classifier machine，this paper establishes the mathematic model of the spiral classifier by a Radial Base Function（RBF） neural network.Genetic algorithm is used to optimize the RBF.With the actual process data from a grinding circuit of a mineral factory，the result of simulation shows that the accuracy satisfies the craft request.

Key words: spiral classifier, mathematic model, Radial Base Function（RBF） neural network, genetic algorithm