Various studies on arc detection methods are described. Series AC arc is detected based on the characteristics extracted from arc voltage, frequency, and time domain of the current. Methods of arc detection using artificial intelligence have been studied previously. In the present study, the performance of multiple methods is analyzed by comparing different input parameters and artificial neural networks. In addition to the input parameters presented in the literature, the performance is compared and analyzed using the following parameters: zero-crossing period, frequency average, instantaneous frequency, entropy, combination of fast Fourier transform (FFT) and maximum slip difference, and combination of FFT and frequency average. These parameters and different neural networks are studied in the bounded and unbounded case, and the performance is compared. For different combinations of neural networks and input parameters, another research question is to identify the input parameters to be used if the number of training data is limited. Moreover, this study investigates the change in detection rate depending on the number of training samples. As a result, the minimum dataset size required to obtain the final detection rate is identified.

Deep learning-based series AC arc detection algorithms

Feature selection has been considered as an effective method to solve imbalanced classification problems. It can be formulated as a multiobjective optimization problem (MOP) aiming to find a small feature subset while achieving a high classification accuracy. With traditional MOP, the focus is on deriving an optimal solution (i.e., a feature subset), while ignoring the diversity in solution space (e.g., there could exist multiple feature subsets achieving the same accuracy). Providing more options for feature selection would be beneficial since some features can be more difficult to obtain than others. In this work, we treat feature selection as a multimodal MOP (MMOP) whose goals are to find an excellent Pareto front in objective space and as many equivalent Pareto optimal solutions (feature subsets) as possible in feature space. Note that though several multimodal multiobjective evolutionary algorithms (MMEAs) have been proposed, their use of a convergence-first selection criterion could cause the loss of solution diversity in an objective and feature space. To address the issue, a novel competition-driven mechanism is designed to assist the existing multimodal MMEAs in locating more equivalent feature subsets and a desired Pareto front. The effectiveness of the proposed mechanism is first verified on all 22 MMOPs from CEC2019. Then, the proposed method is applied to feature selection in imbalanced classification problems and a real-world application, i.e., credit card fraud detection. Experimental results show that the proposed mechanism can not only provide more equivalent feature subsets but also improve classification accuracy.

Competition-Driven Multimodal Multiobjective Optimization and Its Application to Feature Selection for Credit Card Fraud Detection

Arc faults are serious discharges, damaging insulation systems, and triggering electrical fires. This is a transversal topic, affecting from residential to aeronautic applications. Current commercial aircrafts are being progressively equipped with arc fault protections. With the development of MEAs, future airliners will require more electrical power to enhance fuel economy, save weight, and reduce emissions. The ultimate goal of MEAs is electrical propulsion, where fault management devices will have a leading role, because aircraft safety is of utmost importance. Therefore, current fault management devices must evolve to fulfill the safety requirements of electrical propelled aircrafts. To deal with increased electrical power generation, the distribution voltage must be raised, thus leading to new electrical fault types, in particular arc tracking and series arcing, which are further promoted by the harsh environments typical of aircraft systems, i.e., low pressure, extreme humidity, and a wide range of temperatures. Therefore, the development of specific electrical protections which are able to protect against these fault types is a must. This article reviews the state-of-the-art of electrical protections for aeronautic applications, identifying the current status and progress, their drawbacks and limitations, the future challenges, and research needs to fulfill the future requirements of MEAs, with a special emphasis on series arc faults due to arc tracking, because of the difficulty in detecting such low-energy faults in the early stage and the importance and harmful effects of tracking activity in cabling insulation systems. This technological and scientific review is based on a deep analysis of research and conference papers, official reports, white papers, and international regulations. 

https://upcommons.upc.edu/bitstream/2117/364283/3/Arc_Fault_Protections_Postprint.pdf

Arc Fault Protections for Aeronautic Applications: A Review Identifying the Effects, Detection Methods, Current Progress, Limitations, Future Challenges, and Research Needs

In power energy distribution networks, switchgear is considered critical equipment. This is because the act of monitoring the operation and condition of switchgear, as well as performing the required corrective maintenance on any potentially problematic equipment, is critical. A single event may harm thousands of customers over time and pose a significant risk to operational staff. Many considerations must be put in place before using outages to switch down the system since they may raise maintenance costs and disrupt the power supply to users. As a result, proper interpretation of switchgear status evaluations is critical for the early identification of possible faults. Existing ultrasound condition-based monitoring (CBM) diagnostic testing techniques require the tester to manually interpret test data. This study aims to review the status of the recent development of CBM techniques with faults in switchgear. The current trend in electrification will be toward the safety and sustainability of power grid assets, which involves an evaluation of the electrical systems’ and components’ health and grids for medium-voltage distribution. This work provides a current state-of-the-art evaluation of deep learning (DL)-based smart diagnostics that were used to identify partial discharges and localize them. DL techniques are discussed and categorized, with special attention given to those sophisticated in the last five years. The key features of each method, such as fundamental approach and accuracy, are outlined and compared in depth. The benefits and drawbacks of various DL algorithms are also examined. The technological constraints that hinder sophisticated PD diagnostics from being implemented in companies are also recognized. Lastly, various remedies are suggested, as well as future research prospects.

/pdf/review-of-medium-voltage-switchgear-fault-detection-in-a-1uuozj8l.pdf

Review of Medium-Voltage Switchgear Fault Detection in a Condition-Based Monitoring System by Using Deep Learning

Feature selection has been considered as an effective method to solve imbalanced classification problems. It can be formulated as a multiobjective optimization problem (MOP) aiming to find a small feature subset while achieving a high classification accuracy. With traditional MOP, the focus is on deriving an optimal solution (i.e., a feature subset), while ignoring the diversity in solution space (e.g., there could exist multiple feature subsets achieving the same accuracy). Providing more options for feature selection would be beneficial since some features can be more difficult to obtain than others. In this work, we treat feature selection as a multimodal MOP (MMOP) whose goals are to find an excellent Pareto front in objective space and as many equivalent Pareto optimal solutions (feature subsets) as possible in feature space. Note that though several multimodal multiobjective evolutionary algorithms (MMEAs) have been proposed, their use of a convergence-first selection criterion could cause the loss of solution diversity in an objective and feature space. To address the issue, a novel competition-driven mechanism is designed to assist the existing multimodal MMEAs in locating more equivalent feature subsets and a desired Pareto front. The effectiveness of the proposed mechanism is first verified on all 22 MMOPs from CEC2019. Then, the proposed method is applied to feature selection in imbalanced classification problems and a real-world application, i.e., credit card fraud detection. Experimental results show that the proposed mechanism can not only provide more equivalent feature subsets but also improve classification accuracy. 

AC series arc is dangerous and can cause serious electric fire hazards and property damage. This article proposed a convolutional neural network -based arc detection model named ArcNet. The database of this research is collected from eight different types of loads according to IEC62606 standard. The two most common types of arcs, including arcs from a loose connection of cables and those caused by the failure of the insulation, are generated in testing and included in the database. Using the database of raw current, experimental results indicate ArcNet can achieve a maximum of 99.47% arc detection accuracy at 10 kHz sampling rate. The model is also implemented in Raspberry Pi 3B for classification accuracy. A tradeoff study between the arc detection accuracy and model runtime has been conducted. The proposed ArcNet obtained an average runtime of 31 ms/sample of 1 cycle at 10 kHz sampling rate, which proves the feasibility of practical hardware deployment for real-time processing.

ArcNet: Series AC Arc Fault Detection Based on Raw Current and Convolutional Neural Network

Although photovoltaic (PV) systems play an essential role in distributed generation systems, they also suffer from serious safety concerns due to DC series arc faults. This paper proposes a lightweight convolutional neural network-based method for detecting DC series arc fault in PV systems to solve this issue. An experimental platform according to UL1699B is built, and current data ranging from 3 A to 25 A is collected. Moreover, test conditions, including PV inverter startup and irradiance mutation, are also considered to evaluate the robustness of the proposed method. Before fault detection, the current data is preprocessed with power spectrum estimation. The lightweight convolutional neural network has a lower computational burden for its fewer parameters, which can be ready for embedded microprocessor-based edge applications. Compared to similar lightweight convolutional network models such as Efficientnet-B0, B2, and B3, the Efficientnet-B1 model shows the highest accuracy of 96.16% for arc fault detection. Furthermore, an attention mechanism is combined with the Efficientnet-B1 to make the algorithm more focused on arc features, which can help the algorithm reduce unnecessary computation. The test results show that the detection accuracy of the proposed method can be up to 98.81% under all test conditions, which is higher than that of general networks.

https://www.mdpi.com/1996-1073/15/8/2877/pdf?version=1650001868

A DC Series Arc Fault Detection Method Based on a Lightweight Convolutional Neural Network Used in Photovoltaic System

Series AC arc fault is dangerous because it may lead to electrical fire hazards. It is very challenging to detect it accurately. This paper proposes a decision tree-based machine learning algorithm, random forest (RF), to detect series AC arc faults. The proposed model is simpler and lighter than traditional artificial neural network (ANN) or deep neural network (DNN) based algorithms. The model is evaluated with load currents of 8 different types of loads collected following the IEC62606 standard. The arcs in the database are generated by simulating loose cable connections and insulation faults, corresponding to the two most common types of arc faults. A grid search algorithm is used for hyperparameter tuning and a precision-recall trade-off analysis is performed for an optimal classification threshold of the proposed algorithm. The experimental evaluation of the model, using raw current as input, demonstrated an arc fault detection accuracy of 99.78% at 10 kHz sampling rate of the data. The proposed model is further optimized using the Gradient Boosting method to make it lightweight so that it can be implemented in a resource-limited edge computing device at ease with lower latency. The optimized model has an accuracy of 99.65% and an average runtime of 10.22 ms per sample (1 cycle) when implemented in Raspberry Pi 3B. This proves the feasibility of the model for practical deployment in real-world arc fault detection applications.

Series AC Arc Fault Detection Using Decision Tree-Based Machine Learning Algorithm and Raw Current

Series ac arc fault is dangerous and is difficult to detect because of load characteristics and external noise. Artificial Intelligence (AI)-based arc fault detection models even though can achieve a high arc fault detection accuracy, most of the reported researches are still not suitable for practical real-time and commercial application because of the higher computation burden. This research proposed a lightweight arc fault detection algorithm, Efficient-ArcNet, based on EffNet building blocks which can achieve an arc fault detection accuracy of 99.36%. Considering the practical application, the Efficient-ArcNet model is converted into the TF-Lite model to further optimize it for edge computing devices such as Raspberry PI 4 or Jetson Nano. The proposed model achieves an average runtime of 1.17 ms per sample using Raspberry PI 4B, which indicates its suitability for practical deployment in commercial microcontroller units (MCUs).

Efficient-ArcNet: Series AC Arc Fault Detection using Lightweight Convolutional Neural Network

Arc faults pose challenges to electric safety, which can cause serious fire hazards. However, the commonly used arc fault detection method is prone to nuisance tripping. This paper proposed a hybrid arc fault detection method based on the improved Mel-Frequency Ceptral Coefficients (MFCC) for preprocessing and a neural network model for arc identification called ARC_MFCC. As per the IEC 62606, twelve different loads/scenarios are considered for this research. An arc tangent-based core filter is employed to improve the MFCC to enhance the arc features within a bandwidth of 3 kHz to 7 kHz. A lightweight neural network model of fully connected cascaded with the MFCC-based preprocessing, which can distinguish the arc fault with normal operation under different test conditions. As a verification result, the ARC_ MFCC can achieve an accuracy of 99.34%. Moreover, the proposed method is implemented by Raspberry pie 4B. Test results show an average running time of about 4.2 ms per sample, which Ensures that the tripping time can meet IEC 62606.

Yao Wang

Papers

ArcNet: Series AC Arc Fault Detection Based on Raw Current and Convolutional Neural Network

A DC Series Arc Fault Detection Method Based on a Lightweight Convolutional Neural Network Used in Photovoltaic System

Series AC Arc Fault Detection Using Decision Tree-Based Machine Learning Algorithm and Raw Current

Efficient-ArcNet: Series AC Arc Fault Detection using Lightweight Convolutional Neural Network

A Novel Series Arc Fault Detection Method Based on Mel-Frequency Cepstral Coefficients and Fully Connected Neural Network