Adaptive inter-area power oscillation damping from offshore wind farm and MMC-HVDC using deep reinforcement learning

Abstract

The coordination of the offshore wind farm (OWF) and high-voltage direct current (HVDC) links to provide ancillary power oscillation damping (POD) services may become a mandatory requirement from the transmission system operators in the near future. However, the performances of the POD controllers (PODCs) are vulnerable to the power system uncertainties and random communication delays of the wide-area signals. To address these issues, this paper proposes the design of the coordinated PODCs by using the deep reinforcement learning (DRL) method. The DRL-based PODCs employ one of the state-of-the-art DRL algorithms, proximal policy optimization, which can learn to adapt to the system uncertainties and time-varying communication delays by interacting with the power system continuously. A detailed simulation model of an OWF connected to the IEEE-39 bus AC grid via the modular multilevel converter based HVDC link has been built as a simulation platform, and the efficacy of the proposed DRL-based PODCs has been validated across a broad spectrum of operating conditions, disturbances, and communication latency.