Novel neural fuzzy network for product quality monitoring

摘要

A novel self-organizing neural fuzy network is proposed for nonlinear soft-sensing modeling of a cheical process. It is a feedforward multiplayer network and learns in two main phases. The proposed network was applied to a practical application of modeling the octane number of a continuous reforming plant. Results show that is psosesses better generalization ability and simple model structure, and can improve octane rating modeling.