Model predictive optimal control for the coordinated system of supercritical power unit based on firefly algorithm and neural network modeling

Abstract

With the widespread implementation of Automatic Generation Control (AGC) in regional power grids, large-capacity supercritical and ultra-supercritical (SC/USC) power units are required to participate in peak load regulation frequently and often operate under wide-scope variable load conditions. Since a SC boiler unit is a MIMO strong coupling system with nonlinearity and large time delay characteristics, the traditional coordinated control strategy based on PID controllers often cannot meet the requirements with slow load response and large steam pressure fluctuations. Therefore, a model predictive optimal control (MPOC) scheme is proposed for the coordinated system control of a supercritical power unit on the basis of an improved firefly algorithm (FA) and neural network modeling. The MPOC scheme is programmed with MATLAB software and implemented in the full-scope simulator of a 600MW supercritical power unit. The test results show that the method can greatly improve the load response speed and keep the main steam pressure within safety limits.