Model identification and predictive control of steam temperature in coal-fired power plant

Steam temperature is a critical parameter to the efficient operational of steam turbines in coal-fired power plant. The conventional Proportional-Integral-Differential (PID) control method is difficult to achieve a satisfied performance in this application because of the challenges of significant time delay and dynamic disturbances from combustion fluctuation in boiler. The technologies of particle swarm optimization (PSO) based model identification and dynamic matrix control (DMC) based predictive control are proposed to overcome these challenges in this paper. These methods are validated in both a commercial power plant simulation platform and a field operating 300 MW coal-fired power plant. Steam temperature deviations are reduced by more than 50% comparing to the existing PID based control strategy.