Publications

Abstract : Hydropower has become the main force of grid frequency regulation due to its regulation flexibility and rapid response characteristics. However, when the operating conditions are changed, the continued pursuit of a faster response speed deteriorates the damping characteristics of the system with wider water-head, causing low-frequency oscillations, threatening the power station's safety and stability and the power grid. In this study, an improved quantized damping method is applied. The settling time and damping coefficient variation under different PID parameters and operating conditions are recorded, revealing the contradiction between regulation performance and damping characteristics. Then the segmented optimal PID controller is proposed to balance this contradiction. The twin-delayed deep deterministic policy gradient learning algorithm enables the controller to find the optimal policy online. With the settling time-damping threshold control strategy, the controller optimizes parameters according to operating conditions, and changes parameters when reaching the threshold. The results show that, compared with using a set of PID parameters, the damping of the hydropower system is increased by 0.35 from negative to positive. In contrast, the settling time increases by 11.63s within limits. The proposed controller ensures the safe and coordinated operation of the power grid and the power station of the hydroelectric power system with a wide water-head.