Active Power Optimal Damping Control Strategy for Units Based on Deep Q-Network

• Published in: 2019 IEEE 3rd Conference on Energy Internet and Energy System Integration (EI2) pp. 2735 - 2739
• Main Authors: Wu, Xuhan, Lang, Yansheng, Chen, Feng, Yang, Xiaonan, Hu, Yunxu, Yu, Yiping
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2019
• Subjects: active power optimal; Damping; damping control; deep Q-network; Genetic algorithms; low-frequency oscillation; Neural networks; Optimization; Oscillators; Power grids; Power system stability
• DOI: 10.1109/EI247390.2019.9062073

With the gradual establishment of UHVAC/DC hybrid power grid in China, the structure of the power grid is becoming increasingly complex and the number of interval oscillation modes is increasing. Power oscillation has become a prominent problem which troubles the safe and stable operation of the modern power grid. In most papers, the oscillation frequency, damping ratio and sensitivity of electromechanical oscillation mode are selected as the analysis basis to suppress the low-frequency oscillation of power system and improve the system stability by installing PSS (power system stabilization). In this paper, the objective function of the weakest damping ratio of interval modes is used to establish the unit output optimization model, and a method of unit output optimization based on deep Q network is proposed. Simulations on 4-unit-2-zone system and 10-unit-39-bus system in New England show that changing the output of units can significantly improve the damping ratio of the weakest oscillation mode. The feasibility and effectiveness of this method are verified by the comparison with the genetic algorithm.