Research On Intelligent Control Strategy For The Main Steam Temperature Of Supercritical Once-through Boiler

Supercritical once-through boiler is the development of large boiler in our country in the future. So, studying in-depth and mastering supercritical boiler control technology is very important. And the main steam temperature is the main performance parameters of the boiler, so the main steam temperature adjustment has important significance to the power plant security, economic and stable operation etc. The conventional PID control is difficult to remain good performance in nonlinear, large inertia, time-delay conditions. In this paper,600 MW supercritical boiler was studied, and two control strategies of the main steam temperature had been designed: fuzzy PID cascade control and neural network PID cascade control. Main works are as follows:1) This research target is the main steam temperature with nonlinear, large inertia, delay in time. In this paper, the mathematics model of the main steam temperature of the 600 MW supercritical boilers was achieved based on the analysis of dynamic characteristics.2) The main steam temperature conventional cascade PID control system was designed. Genetic algorithm was adopted for PID parameters optimization. The time error integral (ITAE) objective function was selected to get ideal PID parameters. The simulation results showed that the cascade PID controller in tracking and interference resistance performance has better practicability, but the robust performance was poor.3) The main steam temperature fuzzy PID cascade control system was designed, Variable domain was adopted. Variable domain fuzzy PID control algorithm changes the domain of fuzzy input and output variables based on a new scaling factor. And it adopts the simple function as scaling factor and it is simple, convenient and in low computation complexity and strictly testified. The simulation results showed that, compare with the conventional cascade PID controller, the fuzzy PID cascade control has small overshoot, high control precision, good control effect etc.4) The basic principle of single neurons and the BP neural network was researched. The single neuron PID controller with the neuron proportional coefficient to auto-tuning online and BP neural network controller that uses the Bi-Phases Weights Adjusting (BPWA) Algorithm to train single hidden-layer feedforward neural networks were designed to realize the main steam temperature control. The simulation results showed that, compare with the conventional cascade PID controller, the new control algorithms in tracking and interference resistance performance and robustness has better practicability.