A multivariable predictive control strategy for economical fossil power plant operation

To realize the most economical operation of the plant, requires the controller recognize the interaction between multiple inputs and the constraints imposed by the physical limits of the system. A model-based multivariable predictive optimal control strategy with real-time constrained optimization has been discussed to control steam temperature and pressure at their economic optimum during load-cycling operation of a 200 MW oil-fired drum-boiler fossil power plant, so that the plant could be operated at a higher efficiency and without impairing the life of the plant. Artificial neural networks (ANN) modeling technique has been used for identifying global dynamic models of the plant variables. Results are given to demonstrate the superiority of the strategy in a MIMO case to control the main steam temperature and reheat steam temperature and main steam pressure.