A RBF neural network soft sensing model for alumina density based on niche hierarchical genetic algorithm

In order to sense an alumina density of aluminum reduction cell in an on-line manner, a kind of soft sensing model based on a RBF neural network is proposed. The RBF neural network is used to establish a mapping from an error of cell resistance, a cumulative change of cell resistance, and a baiting quantity to an alumina density by taking advantage of approximating nonlinear functions with arbitrary precision. Moreover, a niche hierarchical genetic algorithm is used to describe the structure and parameters of the RBF network, which can solve the problem of determining the number of hidden neurons of RBF network. The practical result indicates that the proposed soft sensing model is effective.