Soft-sensor Modeling of Cement Raw Material Blending Process Based on Fuzzy Neural Networks with Particle Swarm Optimization

By combining particle swarm optimization algorithm (PSO) with fuzzy neural networks (FNN), a PSO fuzzy neural networks (PSO-FNN) was proposed, which takes full advantage of the global search ability of particle swarm optimization (PSO) algorithm and the local search ability of conjugate gradient algorithm with constraints. The new method assumed that FNN was used to construct the model of cement raw material blending process, while PSO was employed to optimize parameters of FNN. Experiment results show that the model based on PSO-FNN has higher precision and better performance than the model based on BPNN.