Nonlinear predictive control on the load system of a thermal power unit based on EMRAN and SAPSO

Due to the strong coupling and nonlinear properties of large-scale boiler-turbine-generating unit load control systems, conventional linear control strategies don't yield satisfactory control performance. We hereby propose a novel nonlinear predictive control strategy based on extended minimal resource allocation network model and simulated annealing particle swarm optimization algorithm. A neural network model derived from online auto-tuning identification, is used for the prediction of future plant behavior. The receding horizon optimization of nonlinear predictive controller is achieved online by simulated annealing particle swarm optimization algorithm, in order to obtain the corresponding optimal control actions at each sampling instant. The simulation study results show the proposed control method has excellent control performance and enhanced self-adaptability, and thus is suitable to the boiler-turbine-generating unit load control systems.