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

Abstract

During AC series arc faults (SAFs), arcing current features can change significantly or vanish rapidly under different load-combination modes and fault inception points. The phenomena make it very challenging for feature-extracting algorithms to detect SAFs. To address the issues, this paper presents a detection model based on regular coupling features (RCFs). After the model is only trained by the samples in single-load circuits, it can detect SAFs under unknown multi-load circuits. To extract the RCFs, asymmetric magnetic flux is coupled by passing the live line and the neutral line through the current transformer. The coupling signals are not influenced by the multi-load circuits. According to the unique signals, two time-domain features and one frequency-domain feature are extracted to represent the RCFs, including impulsefactor analysis (IFA), covariance-matrix analysis (CMA) and multiple frequency-band analysis (MFA). Then, the impulse factor and its threshold are used to preprocess the signals and decrease analysis complexity for the classifier. Finally, the experimental results show that the proposed method has significantly improved generalization ability and detection accuracy in SAF detection.

Identifying DC Series and Parallel Arcs Based on Deep Learning Algorithms

“…Hence, the sample rate of 250 kHz was considered to be sufficiently high to ensure a balance between efficiency and execution time. Additionally, the recent arc fault research selected similar data durations (arrangements of 2, 3, or 4 ms periods) [21][22][23][24][25]. Therefore, this study chooses the window duration at a 2 ms period.…”

“…In Refs. [19][20][21][22], numerous AI models were employed to diagnose series arcing events using different characteristics as inputs. The adoption of AI algorithms for parallel arc diagnosis was proposed in [23].…”

Detection of AC Arc Faults of Aviation Cables Based on H-I-W Three-Dimensional Features and CNN-LSTM Neural Network

“…Therefore, for acquiring complete signal information, the sampling rate chosen in the test was 2.5 MHz. When the smallest observation window (OW) was selected, the main spectral information should be contained [14]. Different OWs (0.5 cycle, 1 cycle, and 2 cycles) were compared and the spectral information near the fundamental frequency was in the normal state (Fig.…”

“…Guanghai Bao et al [13] used a fourth-order cumulative volume algorithm to detect arc pulses at a sampling rate of 10 MHz, but the method could not distinguish normal pulses from arc pulses. Run Jiang et al [14] combined regular coupling features (RCFs) with SVM to determine the presence of arcs by impulse factor analysis (IFA) and then trained the SVM to further identify arc pulses and normal pulses by covariance matrix analysis (CMA) and multiple frequency-band analysis (MFA). This method successfully differentiated normal pulse signals from arc fault pulse signals and reduced the misidentification rate.…”

“…Metrics such as kurtosis, mean value, variance, and third, and fourth accumulation were also introduced into arc fault detection [12]. A method based on high-frequency pulse components of coupled signals was used in the identification of AC series arc faults in residential buildings [13]- [14]. On the basis of this method, Run et al proposed short observation window singular value decomposition and reconstruction algorithm [15] and time series reconstruction algorithm [16] for enhancing high-frequency pulse components and realized series arc fault identification.…”