Data-Driven RBFNN-ARX Modeling of Steam Boilers in Thermal Power Plants with Nonlinear Model Predictive Control

The study has started to test and analyze some of the most advanced control strategies for the boiler-Turbinegenerator train, concentrating on the steam boiler of a thermal power plant. Nonlinearities, system coupling, and the inability to foresee future behavior are the main difficulties that conventional logic-based PLC control systems face. A data-driven model of the boiler's drum level, steam temperature, and pressure was employed to create an RBFNN-ARX structure based on measured input-output data as a solution to these difficulties. Besides, control techniques, Linear Quadratic Regulator (LQR) and Non-Linear Model Predictive Control (NMPC), were designed and simulated on the model in MATLAB/Simulink before being tested. Outcomes from the simulation indicate that the LQR controller while being able to do satisfactory tracking, is outperformed by the NMPC in regard to presenting smooth responses, quick settling times, and more effective disturbance handling for all the parameters. The suggested MPC plan put forward a feasible way to modernize the conventional PLC-based systems that are already in place in thermal power plants by incorporating predictive, constraint-Aware, and more efficient control capabilities.