http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
측정 잡음을 고려한 기 무한모선 시스템의 고장 판별을 위한 기반 외란 관측기 설계
양선직,장수영,손영익 대한전기학회 2020 전기학회논문지 Vol.69 No.7
Increasing demand on electric power supply results in a need of an intelligent method for fast detection of various failures in the power system. This paper presents a reinforcement learning-based disturbance observer (DOB) design for the determination and protection against a line fault occurred in the single-machine infinite bus (SMIB) power system. Whilst a high gain disturbance observer could estimate the system states and the external disturbance successfully, the high gain of the observer can cause problems in the presence of the measurement noise. When measurement noise exists in the output, fault detection methods based on the estimated states may often result in false alarms. To solve the problem, this paper designs an adaptive DOB using Deep Q-Network (DQN) which is one of reinforcement learning algorithms. For the proposed observer design, this paper explains the definitions of the state, the action, and the reward for the reinforcement learning. Matlab simulations have been conducted based on the observer gains trained using the power angle data from the swing equation. The results show that the estimation performance of the proposed DQN-based observer can be satisfactory against both an external disturbance and the measurement noise.
측정잡음에 강인한 강화학습 기반의 적응형 외란 관측기 설계
양선직(Sun Jick Yang),아마레 네비옐레울 다니엘(Nebiyeleul Daniel Amare),손영익(Young Ik Son) 대한전기학회 2021 대한전기학회 학술대회 논문집 Vol.2021 No.7
본 논문은 1기 무한모선 계통의 운전 중에 발생할 수 있는 3상 단락과 같은 고장으로 인한 외란을 추정하기 위한 관측기를 제안한다. 내부 모델 원리를 기반으로 하여 설계한 제안하는 강화학습(RL; Reinforcement Learning) 기반의 외란 관측기(DOB; Disturbance Observer)가 측정잡음 하에서도 외란을 강인하게 추정할 수 있음을 모의실험을 통해서 확인한다.
Off-Policy 학습 기반 LQR을 이용한 파라미터 추정 및 전력 시스템 동기 탈조 판별을 위한 외란 관측기 설계
양선직(Sun Jick Yang),장수영(Su Young Jang),손영익(Young Ik Son) 대한전기학회 2021 전기학회논문지 Vol.70 No.1
In this paper, a reinforcement learning-based Linear Quadratic Regulator(LQR) design method has been adopted to identify unknown system parameters. The off-policy learning-based LQR can obtain the optimal control gain through an iteration technique known as policy iteration, without using the system model parameters. Augmented states, using the system output integration, can help to alleviate the rank condition on the proposed parameter estimation method. Increasing the system model information accuracy allows the disturbance observers to detect various system faults with the least amount of estimation error. The line fault detection ability of a power system for out-of-step prediction has been studied by applying the proposed parameter estimation scheme to a single machine infinite bus system. Simulation results show that both the system parameters and the external disturbance can be successfully estimated through the proposed method.
측정 잡음을 고려한 1기 무한모선 시스템의 고장 판별을 위한 DQN 기반 외란 관측기 설계
양선직(Sun Jick Yang),장수영(Su Young Jang),손영익(Young Ik Son) 대한전기학회 2020 전기학회논문지 Vol.69 No.7
Increasing demand on electric power supply results in a need of an intelligent method for fast detection of various failures in the power system. This paper presents a reinforcement learning-based disturbance observer (DOB) design for the determination and protection against a line fault occurred in the single-machine infinite bus (SMIB) power system. Whilst a high gain disturbance observer could estimate the system states and the external disturbance successfully, the high gain of the observer can cause problems in the presence of the measurement noise. When measurement noise exists in the output, fault detection methods based on the estimated states may often result in false alarms. To solve the problem, this paper designs an adaptive DOB using Deep Q-Network (DQN) which is one of reinforcement learning algorithms. For the proposed observer design, this paper explains the definitions of the state, the action, and the reward for the reinforcement learning. Matlab simulations have been conducted based on the observer gains trained using the power angle data from the swing equation. The results show that the estimation performance of the proposed DQN-based observer can be satisfactory against both an external disturbance and the measurement noise.
모델 불확실성이 있는 1기 무한모선 계통의 선로 고장 감지를 위한 PI 관측기의 성능 개선
김준우(Jun Woo Kim),양선직(Sun Jick Yang),손영익(Young Ik Son) 대한전기학회 2021 전기학회논문지 Vol.70 No.2
This paper proposes a method for improving disturbance estimation performance of a PI observer for line fault detection of the single–machine infinite bus (SMIB) power system under the model uncertainty. The SMIB model of the paper is described by an improved swing equation that reflects more realistic dynamic characteristics than the conventional one. The proposed method begins with the conventional swing equation and the PI observer for the exact disturbance estimation. Model uncertainty arising from the model difference is first estimated using the estimated equivalent disturbance by the PI observer during equilibrium operation. Next, the uncertain parameters are estimated by applying the Recursive Least Square algorithm to the modified swing equation by using the estimated term at the first step. The angular velocity and acceleration information for the RLS algorithm can be also estimated by the PI observer based on the modified swing equation. Finally, the estimated parameters are adopted to the proposed PI observer and its disturbance estimation performance can be improved effectively for the line fault detection. The performance of the proposed algorithm is verified by computer simulations of the improved SMIB power system model.