Against the backdrop of the current coal-fired power units’ participation in deep peak shaving and rapid frequency regulation, the traditional proportional-integral-derivative (PID) control is difficult to meet its requirements. To this end, a comprehensive and improved control strategy is proposed: based on the quantum genetic algorithm (QGA), the dynamic rotation Angle is adopted for improvement, and the genetic algorithm (GA), QGA, and the improved quantum genetic algorithm (IQGA) are tested and analyzed through nonlinear functions. The IQGA algorithm has a 100% success rate in searching for the optimal value and requires less than 20 iterations, demonstrating better convergence. Finally, the IQGA algorithm was applied to optimize the parameters of the internal model control (IMC) in the SCR denitration system. For the mathematical models of a certain 600 MW unit under three operating conditions, comparisons were made with the conventional cascade PID control in terms of tracking set values and anti-interference capabilities. The results show that the IMC-P control optimized by IQGA has a minimum reduction of 2% in overshoot and a minimum shortening of 43.1% in adjustment time when step disturbances or internal disturbances occur, and has better dynamic performance. Moreover, under the working condition of deep variable load of the unit, the fluctuation amplitude of the IMC-P internal model control during the first model switching process is 1.5%, showing better robustness.