Neural network predictive control for superheated steam temperature based on modified particle swarm optimization
基于改进PSO算法的过热汽温神经网络预测控制

Combining modified particle swarm optimization (MPSO) with neural network predictive control (NNPC), we propose a model-prediction controller,based-on modified particle swarm optimization (MPSO) and radial basis function (RBF) hybrid optimization strategy (MPSO-RBF),and a nonlinear optimization controller,based-on MPSO.For the super- heated steam temperature control,we construct a cascade control system based on the neural network predictive control, and analyze all related problems,including the predictive model,the rolling optimizing algorithm,the feedback adjusting and the simulation-parameter setting.We also present the particle encoded format of MPSO,operating design method,and steps in hybrid optimization algorithm.Simulation experiments of the superheated steam temperature control were done in a super-critical-600 MW direct-current boiler,demonstrating the validity,the superior performance and the application prospects.