Superheated steam temperature system of thermal power control engineering based on neural network local multi-model prediction

Abstract

The superheated steam temperature object of thermal power plant has the characteristics of time lag, inertia and time-varying parameters. The control quality of the conventional proportional integral derivate controller with fixed parameters will decrease after the object characteristics change. The generalized predictive control strategy of superheated steam temperature based on neural network local multi-model switching can achieve the goal of designing sub-controllers for fixed models under several typical operating conditions. When the system operating conditions change, the effective switching strategy is timely and accurate. Switch to the most suitable controller. The paper proposes a new smooth switching method, which can effectively suppress the large disturbance phenomenon of the object when switching. The simulation results verify the effectiveness of the control strategy.