Deep Reinforcement Learning-Based Robust Protection in Electric Distribution Grids.
Year: 2020
Type: preprint
Abstract: This paper introduces a Deep Reinforcement Learning based control architecture for protective relay control in power distribution systems. The key challenge in protective relay control is to quickly and accurately detect faults from other disturbances in the system. The performance of widely-used traditional overcurrent protection scheme is limited by factors including distributed generations, power electronic interfaced devices and fault impedance. We propose a deep reinforcement learning approach that is highly accurate, communication-free and easy to implement. The proposed relay design is tested in OpenDSS simulation on the IEEE 34-node and 123-node test feeders and demonstrated excellent performance from the aspect of failure rate, robustness and response speed.
Source: arXiv (Cornell University)
Authors: Dongqi Wu, Dileep Kalathil, Le Xie
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3010064516