AI, Год журнала: 2024, Номер 5(4), С. 2433 - 2460
Опубликована: Ноя. 15, 2024
Machine learning (ML) techniques have permeated various domains, offering intelligent solutions to complex problems. ML has been increasingly explored for applications in active power filters (APFs) due its potential enhance harmonic compensation, reference signal generation, filter control optimization, and fault detection diagnosis. This paper reviews the most recent of APFs, highlighting their abilities adapt nonlinear load conditions, improve classification accuracy, optimize system performance real time. However, this also highlights several limitations these methods, such as high computational complexity, need extensive training data, challenges with real-time deployment distributed systems. For example, marginal improvements total distortion (THD) achieved by ML-based methods often do not justify increased overhead compared traditional methods. review then suggests future research directions overcome limitations, including lightweight models faster more efficient control, federated decentralized digital twins monitoring. While remain effective, significantly APF
Язык: Английский