Skip to content
Open access

Complex Power Quality Disturbance Detection Across Edge and High-Performance Platforms via Embedded Feature Coding and Feature Fusion

2026 · IEEE Transactions on Instrumentation and Measurement · Vol 75, pp. 3524714-3524714 · 0 citations · 45 references

Abstract

Rapid and accurate detection of power quality disturbances (PQDs) is essential for stable operation of modern grids. This study proposes an efficient PQD detection method, signal recurrence coding feature enhancement (SRE)–multiscale feature fusion network (MSFFNet). The proposed framework consists of an SRE layer and an MSFFNet classifier. First, the SRE layer is designed to characterize the recurrence relationships and intrinsic similarity patterns of PQD signals. As a preceding layer of SRE-MSFFNet, the SRE layer optimizes the soft threshold parameter in recurrence coding through backpropagation, thereby enabling adaptive recurrence feature enhancement. Second, an MSFFNet classification structure is proposed to efficiently extract and fuse multiscale disturbance features for accurate PQD identification. Experimental results indicate that the proposed method achieves better performance than other popular methods, with a PQD validation accuracy of 99.38 % $\pm ~0.06$ % and a single-sample runtime of 5.03 ms on the workstation. Finally, the edge-computing hardware experiments show that the proposed method can accurately identify different PQD categories, with single-sample inference latencies ranging from 81.17 to 760.60 ms across different embedded platforms, demonstrating its effectiveness and execution capability for power quality monitoring and disturbance identification on resource-constrained instrumentation devices.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.