Home // ENERGY 2016, The Sixth International Conference on Smart Grids, Green Communications and IT Energy-aware Technologies // View article

Finding Needles in a Haystack: Line Event Detection on Smart Grid PMU Data Streams

Authors:
Duc Nguyen
Scott Wallace
Xinghui Zhao

Keywords: Smart Grid; Phasor Measurement Unit (PMU); Machine Learning; Event Detection

Abstract:
Smart Grid technology, in particular Phasor Measurement Units (PMUs) provide a mechanism for monitoring the state of a power system across a wide geographical area at high resolution and with high fidelity. These measurements form a large corpus of state information that power systems engineers and researchers can use to find and analyze interesting phenomena either post-hoc, or in real-time. In this paper, we present our work with machine learning to develop an event detector for use with the Bonneville Power Administration's (BPA's) current PMU installation. Our system can be used post-hoc or in real-time and focuses on identifying line faults in the data stream since these events can be easily verified by records BPA maintains. One challenge for machine learning algorithms is that the modern transmission systems are very often well-behaved. Since PMUs record measurements at 60 samples/sec and each PMU typically records up to 16 phasor signals, each PMU records upwards of 80 million measurements per day. Line events, in contrast, happen very rarely (on the order of 100 per month), at least on a transmission system such as BPA's. In this paper, we examine the performance of multiple classifiers within this power system domain. In addition to examining classifier performance on a set of validated line events, we perform a detailed analysis of false alarms and explore multiple methods for reducing false alarms in a real system.

Pages: 42 to 47

Copyright: Copyright (c) IARIA, 2016

Publication date: June 26, 2016

Published in: conference

ISSN: 2308-412X

ISBN: 978-1-61208-484-8

Location: Lisbon, Portugal

Dates: from June 26, 2016 to June 30, 2016