Superheated Steam Temperature System SVM Model Predictive Control Based On Particle Swarm Optimization

Boiler superheated steam temperature is a typical object with large delay, inertia, time-varying and nonlinearity. Currently,the conventional Cascade PID control strategy is widely used in power plant superheated steam temperature control system. But it is difficult to achieve the desired control effect at large disturbance and variable conditions. In view of the good effects of model predictive control (MPC) algorithm in industrial processes applications and the outstanding nonlinear fitting ability of support vector machine (SVM), this paper create a prediction model by SVM method and design the superheated steam temperature control system based on this model.Particle swarm optimization (PSO) algorithm has good ability of global optimization. It is simple and easy to implement. In order to further improve the performance of superheated steam temperature control system, combine the PSO with SVM predictive control. An online rolling optimization controller based on PSO algorithm is constructed. Analyze the problem of system implementation, involved model identification, rolling optimization, feedback correction and parameter selection. Carry out the simulation test for a 600MW unit boiler superheated steam temperature object under Matlab. The results show that the method proposed in this paper has better performance than commonly used cascade PID control method in power plant.