Modified Predictive Optimal Control Using Neural Network-based Combined Model for Large-Scale Power Plants

With a neural network-based combined model (NNCM) for a power plant, a modified predictive optimal control (MPOC) system can be developed based on predictive control algorithms and intelligent techniques. During the NNCM simulation, an on-line identification (OLID) system is updated every few steps to provide information from the model to the MPOC. Moreover, the MPOC will use the OLID as a test process to optimize the control actions, minimizing tracking-error. To search for the best control action, the MPOC utilizes a heuristic optimization technique, particle swam optimization. With the proposed MPOC system the only input to the NNCM will be the unit load demand. Finally, major outputs of NNCM will be shown using the proposed approaches, validating the procedure as a means to design a control system for a new power plant.