A Coupling Method for Identifying Arc Faults Based on Short-Observation-Window SVDR

Abstract

This paper presents a new method for effective detection of AC series arc fault (SAF) and extraction of SAF characteristics in residential buildings, which addresses the challenges with conventional current detection methods in discriminating arcing and non-arcing current due to their similarity. Different from the traditional method, in the proposed method, the differential magnetic flux is coupled to obtain high-frequency signals by putting the live line and the neutral line through the current transformer, which can effectively solve the problem of SAF features disappearing in the trunk-line current. However, similar to the traditional method, the effectiveness of the proposed coupling method could also be compromised when being used in cases with dimmer load and load starting process. This is found to be caused by the presence of high-amplitude pulse phenomenon in the non-arcing signals in these scenarios, which are incorrectly detected as arcing signals in other loads. To address this issue, a short-observation-window singular value decomposition and reconstruction algorithm (SOW-SVDR) is used to enhance the capability to identify SAFs by the coupling method. The proposed method has been implemented and validated according to UL1699 standard with different types of loads connected to the system and also tested under their starting processes. The experimental results show that the proposed approach is more effective in detecting arc faults compared with existing methods.

Machine Learning Approach to Detect Arc Faults Based on Regular Coupling Features

“…Compared with other existing detection methods, the proposed method can not only detect SAFs under more kinds of loads with higher detection accuracy, but also shows significantly stronger generalization performance in detecting SAFs under unknown multi-load circuits. Reference [8] Reference [17] Reference [18] Reference [26] Main The reference states its SAF detection method based on load classification cannot work in multi-load circuits.…”

“…However, the detection results are not satisfactory in the multiload circuits (800 samples). Reference [8] has used shortobservation-window singular value decomposition and reconstruction to filter out the interference pulses. Then, variance energy is calculated based on the reconstructed signals.…”

“…In the former methods, discrete Fourier transformation (DFT) and discrete wavelet transformation (DWT) are often used to obtain distinctive indicators in a certain frequency band, where there is an obvious difference between arcing and nonarcing conditions [7][8][9][10]. The associated thresholds can then be determined to detect SAFs.…”

Parallel DC Arc Failure Detecting Methods Based on Artificial Intelligent Techniques

“…Studies illustrate comparisons of performance between various AI methods in DC systems in [19]. On the other hand, a short-observation-window singular value decomposition and reconstruction algorithm are proposed to identify AC series arc fault [20]. Although this method guarantees a high diagnosis rate with different load types, the complexity and the need for additional hardware are the limitations of this proposed method.…”

Identifying DC Series and Parallel Arcs Based on Deep Learning Algorithms

“…When the DC arc faults are initiated, several abnormal behaviors can be used to diagnose an arc event, such as current fluctuations and rapid changes in light and/or heat output. These abnormal phenomena can be used to detect DC arcs [6][7][8][9][10][11]. However, the investigation of parallel arc in DC systems is still at a primitive step [12][13][14][15].…”