Abstract: In view of the long time delay, large inertia and strong nonlinearity of the reheater outlet steam temperature in the process of deep peak shaving in thermal power plants, which makes the control less effective or hard to be put into service automatically, this paper proposes a humanoid intelligent control algorithm based on fuzzy switching. This method first optimizes the reheat steam temperature control system, and then uses the particle swarm optimization algorithm combined with the control experience to select the parameter settings. The simulation results show that the proposed method can enhance the robustness of the reheat steam temperature control system. In the actual operation of a 1 000 MW ultra-supercritical unit, satisfactory results have been achieved through this approach, which effectively improved the economic and security of the unit operation.