A 900MW nuclear power unit has been in commercial use for 20 years. The steam-driven feed water pump uses a regulating system constructed by an analog board to control the speed of the pump. Since the analog board has been discontinued, and its reliability is reduced. The power plant has carried out digital transformation on the control part of the speed control system. Shortly after the renovation, the speed of the steam feed pump continued to fluctuate. The phenomenon accompanying the rotational speed fluctuation is a periodic fluctuation similar to a sine wave. The range of fluctuations varies from time to time, and is interspersed with non-sinusoidal fluctuations. When one of the feed pumps is placed in the manual control state, the speed fluctuation of the other pump will increase. After analysis, with the aging of the mechanical parts of the feed pump, the follow-up ability of the pump continues to weaken. Under the combined action of the four loops of speed, W/S differ., SG level and feed water flow, continuous fluctuations occur in the four parameters.
Aiming at the fluctuation problem, this paper relies on the real-time monitoring data of nuclear power plants and Matlab/Simulink to establish a nonlinear model and “Accompanying system” that reflect the actual working conditions on site, and conduct in-depth research on the mechanism of the problem by using control theory, system identification and “Accompanying system”, and propose solutions to resolve fluctuations. This method provides theoretical guidance for reducing or eliminating rotational speed fluctuations, thereby avoiding potential safety hazards caused by multi-parameter fluctuations in the operation of nuclear power units.