Cooperative Change Detection for Online Power Quality Monitoring

TL;DR

This paper introduces a cooperative, real-time power quality disturbance detection method using a generalized likelihood ratio detector and a low-bandwidth, multi-meter communication scheme for efficient online monitoring.

Contribution

It proposes a novel GLLR detector for quick disturbance detection and a level-triggered sampling scheme for low-overhead, cooperative power quality monitoring.

Findings

The GLLR detector achieves lower detection delay than traditional methods.

The multi-meter scheme maintains high detection performance with minimal communication.

Performance approaches ideal centralized detection despite limited bandwidth.

Abstract

This paper considers the real-time power quality monitoring in power grid systems. The goal is to detect the occurrence of disturbances in the nominal sinusoidal voltage/current signal as quickly as possible such that protection measures can be taken in time. Based on an autoregressive (AR) model for the disturbance, we propose a generalized local likelihood ratio (GLLR) detector which processes meter readings sequentially and alarms as soon as the test statistic exceeds a prescribed threshold. The proposed detector not only reacts to a wide range of disturbances, but also achieves lower detection delay compared to the conventional block processing method. Then we further propose to deploy multiple meters to monitor the power signal cooperatively. The distributed meters communicate wirelessly to a central meter, where the data fusion and detection are performed. In light of the limited bandwidth of wireless channels, we develop a level-triggered sampling scheme, where each meter transmits only one-bit each time asynchronously. The proposed multi-meter scheme features substantially low communication overhead, while its performance is close to that of the ideal case where distributed meter readings are perfectly available at the central meter.