Abstract: In order to address the issue where the control system struggles to operate normally and track the setpoint of outlet NOx concentration during the denitration reaction process—due to changes in operating conditions or failures in the outlet NOx concentration detection device—resulting in prolonged excessive NOx emissions, this study proposes an optimized control strategy for SCR flue gas denitration. Based on the technical framework of Cyber-Physical System (CPS) and employing the Improved Generalized Predictive Control (IGPC) algorithm, the strategy integrates an outlet NOx concentration prediction model, an operating condition judgment/fault detection device diagnostic module, a self-healing compensator, a feedback compensator, and an online identification module. The designed SCR flue gas denitration CPS-IGPC control strategy adopts a two-layer structure consisting of an information layer (loop setting layer) and a physical layer (loop control layer). Comparative analyses were conducted against cascade PID control and conventional IGPC control methods..