Diagnosis of Series DC Arc Faults—A Machine Learning Approach

Rory David Telford,Ian Elders,Stuart Galloway,Bruce Stephen

doi:10.1109/tii.2016.2633335

Rory David Telford, Ian Elders + Show 2 more

Open Access

PDF Available

https://doi.org/10.1109/tii.2016.2633335

Copy DOI

Export

Save

Cite

Abstract
Highlights/Summary
Full-Text PDF
Similar Papers

Abstract

Listen

Increasing prevalence of dc sources and loads has resulted in dc distribution being reconsidered at a microgrid level. However, in comparison to ac systems, the lack of a natural zero crossing has traditionally meant that protecting dc systems is inherently more difficult—this protection issue is compounded when attempting to diagnose and isolate fault conditions. One such condition is the series arc fault, which poses significant protection issues as their presence negates the logic of overcurrent protection philosophies. This paper proposes the IntelArc system to accurately diagnose series arc faults in dc systems. IntelArc combines time–frequency and time-domain extracted features with hidden Markov models (HMMs) to discriminate between nominal transient behavior and arc fault behavior across a variety of operating conditions. Preliminary testing of the system is outlined with results showing that the system has the potential for accurate, generalized diagnosis of series arc faults in dc systems.

Highlights

T HE prevalence of dc distribution is a consequence of an increasing reliance on distributed renewable energy sources, higher penetrations of electric vehicles and storage systems, and an overall rise in dc loads such as computers, solid-state lighting, and building networks [1]
hidden Markov models (HMMs) provide a log-likelihood (LL) metric that quantifies the probability of various fault hypotheses—this form of diagnostic explanation is not provided by ANNs, for example, which would only provide a binary classification or regression with no accompanying confidence metric
This paper has proposed IntelArc, a series arc fault diagnosis (AFD) system for application to dc networks, which is based on HMM and utilizes time–frequency and time-domain features extracted from network current data

Summary

INTRODUCTION

T HE prevalence of dc distribution is a consequence of an increasing reliance on distributed renewable energy sources, higher penetrations of electric vehicles and storage systems, and an overall rise in dc loads such as computers, solid-state lighting, and building networks [1]. These circuit imperfections often emerge as a contact separation or loose connection–in harsh operating environments vibration often results in series arcing exhibiting intermittent behavior These faults introduce additional impedance between source and load, and the resultant decrease in network current means they are difficult to detect using conventional overcurrent protection practices. The increased probability of a sustained arcing event means they present a significant fire hazard Their presence has been known to affect the secure and reliable distribution of power in photovoltaic [8], aircraft [9], and shipboard [10] systems. The section of the paper describes arc faults, including difficulties in detecting series conditions and previously proposed diagnostic systems.

SERIES ARC FAULTS AND EXISTING DIAGNOSTIC SYSTEMS

HMM-BASED ARC FAULT DIAGNOSIS

INTELARC—METHOD OVERVIEW

Generation of Arc Fault Data

Arc Fault Feature Extraction and Selection

HMM Training

AFD Algorithm

CASE STUDIES

Case Study 1

RESULTS

Comparison of IntelArc With Existing AFD Methods

Findings