Enhancing Power Quality Event Classification with AI Transformer Models

TL;DR

The study tackles PQE classification under realistic measurement imperfections by introducing an attention-enabled Transformer that processes raw voltage signals without feature extraction. The model demonstrates superior accuracy compared with CNN, DNN, SVM, and RF baselines, achieving up to $acc$ of 99.81% on ideal data and 91.43% under adverse noise, DC offset, and variation conditions. A synthetic PQE dataset with 10 classes and varied SNRs (5–60 dB) plus amplitude/frequency fluctuations underpins the evaluation, supplemented by a hyper-parameter sensitivity analysis that confirms robustness. The work suggests practical applicability for smart-grid PQE monitoring and outlines hardware-in-the-loop validation and resilience considerations for future deployment.

Abstract

Recently, there has been a growing interest in utilizing machine learning for accurate classification of power quality events (PQEs). However, most of these studies are performed assuming an ideal situation, while in reality, we can have measurement noise, DC offset, and variations in the voltage signal's amplitude and frequency. Building on the prior PQE classification works using deep learning, this paper proposes a deep-learning framework that leverages attention-enabled Transformers as a tool to accurately classify PQEs under the aforementioned considerations. The proposed framework can operate directly on the voltage signals with no need for a separate feature extraction or calculation phase. Our results show that the proposed framework outperforms recently proposed learning-based techniques. It can accurately classify PQEs under the aforementioned conditions with an accuracy varying between 99.81%$-$91.43% depending on the signal-to-noise ratio, DC offsets, and variations in the signal amplitude and frequency.

Figures (2)

• Figure 1: Transformer Model Architecture Transformer_Paper.
• Figure 2: PQEs Samples.