Superheated Steam Temperature Control Based on Improved Recurrent Neural Network and Simplified PSO Algorithm

Coal-fired power plants are facing a rapid developing tide toward supercritical and ultra-supercritical boiler units with higher parameters and bigger capacity. Due to the large inertia, large time delay and nonlinear characteristics of a boiler’s superheater system, the widely-used conventional cascade PID control scheme is often difficult to obtain satisfactory steam temperature control effect under wide-range operating condition. In this paper, a predictive optimization control method based on improved mixed-structure recurrent neural network model and a simpler Particle Swarm Optimization (sPSO) algorithm is presented for superheated steam temperature control. Control simulation tests on the full-scope simulator of a 600 MW supercritical power unit showed that the proposed predictive optimization control scheme can greatly improve the superheated steam temperature control quality with good application prospect.