| Title |
Design of a Reinforcement Learning-Based Disturbance Observer for Line Fault Detection of a Single Machine Infinite Bus System |
| Authors |
장수영(Su Young Jang) ; 강상희(Young Ik Son) ; 손영익(Sang Hee Kang) |
| DOI |
https://doi.org/10.5370/KIEE.2019.68.9.1060 |
| Keywords |
Single-machine infinite-bus system; Reinforcement learning; Deep Q-Network; Disturbance observer; Fault detection; Out-of-step |
| Abstract |
According to the increase of electric power demand in the modern society the power system is gradually expanding. This results in a growing need for an intelligent method of fast determination and protection against various failures in the power system. As the computer platform is improved, the system fault detection and reliable protection devices have been trying to enhance their performances using artificial intelligence techniques. If a failure occurs in the single-machine infinite bus(SMIB) system. the electrical output of the generator changes, which can be regarded as a result of an external disturbance input. This paper presents a line fault detection method by using a reinforcement learning-based disturbance observer that estimates the magnitude of the equivalent disturbance. Reinforcement learning is an algorithm that models the relationship between the behavior of an agent and the reward from environment. This paper has adopted the Deep Q-Network for training of the proposed disturbance observer. The performance of the proposed reinforcement learning-based disturbance observer is verified by computer simulations. The results show that the disturbance can be estimated successfully and the estimate can be used to detect the line fault. |