Series arc fault (SAF) has severe impacts on the safety of dc power supply systems. Timely and accurate SAF detection under different operating conditions is an open and challenging problem. To address this problem, this article proposes an integrated SAF detection method for different operating conditions. In the proposed method, dual-tree complex wavelet transform (DT-CWT) is employed to obtain an accurate current signal decomposition. The singular values of each wavelet component are then extracted by using an improved matrix construction method, which can effectively reduce the computational cost of constructing the high-dimension features. Finally, the kernel extreme learning machine (KELM) is applied to fuse the feature information for SAF detection. A series of experiments are presented to demonstrate the effectiveness of the proposed method. The results of offline experiment show that the accuracy of the proposed method has higher detection accuracy than that of the six state-of-the-art methods under different operating conditions. In this article, the proposed method is then embedded into the hardware of the experimental platform for online in-service implementation. The online experimental results show that the proposed method achieves fast and accurate SAF detection and, at the same time, offers outstanding reliability and stability in system dynamic transients.

Series arc faults (SAFs) are able to cause electrical fires, which threaten the security of DC distribution systems critically. The noise of SAFs can interfere in the adjacent lines and the SAF detectors based on the information in a single line are prone to trip. Therefore, in order to correctly diagnose SAF, it is of significant value to consider the centralized SAF diagnosis methods which analyze the information in entire system. This article analyzes the mechanism of arc noise interfering in normal lines in detail and proposes an SAF diagnosis method which comprehensively considers the information in entire system. In the proposed method, the time domain features, frequency domain features and singular values obtained from different scales of current are utilized to construct high-dimensional feature vectors. Then, in order to reduce the redundant information in feature vectors, feature importance analysis and feature selection are realized using random forests (RF). Finally, RF is employed to fuse the information of feature vectors for SAF diagnosis. The offline experiment results indicate that the diagnosis accuracy of proposed method can reach up to 98.93%, the effectiveness and advancement of proposed method is verified by comparing with different state-of-the art methods. Besides, the proposed method is implemented in the microcontroller for the real-time operations. The results of online experiment show that the proposed method is of outstanding reliability and robustness under transient conditions, at the same time, the diagnostic speed is less than 250ms and can satisfy the requirements of standard.

Series arc faults (SAFs) are a primary cause of fire incidents in photovoltaic systems. Accurately and rapidly detecting SAF under the interference of power electronic devices remains a significant challenge. This article proposes an SAF detection method based on the time-frequency Markov permutation transition field (TFMPTF). First, variational mode decomposition is used to decompose the current signal into modes containing different frequency components to prevent interference between the information of different frequency bands. Then, the modes are transformed into two-dimensional matrices using TFMPTF. Innovatively, the concept of time-frequency permutation patterns state transition analysis in TFMPTF is proposed, from which the distinct structural information of the current signal can be effectively depicted. Afterward, singular value decomposition is employed to extract fault features from matrices. Finally, fault features are processed using a kernel extreme learning machine to obtain detection results. Offline experimental results show that the average detection accuracy of the proposed method is 98.97%, and the advancedness and adaptability of the proposed method are verified by comparing it with different methods. The proposed method and comparison methods are implemented in a microprogrammed control unit (MCU) for online experiments, further confirming that the detection speed and detection accuracy of the proposed method are reliable.

The insulation performance of cables plays an important role in the safe and stable operation of aircraft. Time–frequency domain reflectometry (TFDR) is an effective method to detect the local insulation faults of cables. Usually, Wigner–Ville distribution (WVD) is used to analyze the energy distribution of reflected signals in TFDR. However, WVD can easily introduce cross-term interference to the detection results. In addition, most of the current studies have used TFDR to perform offline detection of cable insulation fault, and the performance of TFDR for online operation is unclear. In order to solve the above problems, first, this paper eliminates the interference of cross-term based on smoothed pseudo-Wigner–Ville distribution. Then, an online cable insulation fault detection platform based on non-contact signal injection is built. According to the requirements of the standard GJB181A, the interference degree of the incident signal of TFDR to the original signal of cable is analyzed, and the feasibility of using TFDR method to perform online detection is proved. Finally, the correlation between different features of reflected signals under different fault degrees of cable is analyzed. It is proved that time-domain correlation has a better correlation with the fault degree of cable compared to other four features.

Series arc faults (SAFs) pose a significant threat to the safety of photovoltaic (PV) systems. However, the complex operating conditions of PV systems make accurate SAF detection challenging. To tackle this issue, this article proposes a SAF detection method based on time–frequency composite recurrence plots (TFCRPs). Initially, variational mode decomposition (VMD) is employed to decompose the current into distinct modes. Subsequently, the proposed TFCRP transforms these modes into two-dimensional matrices, enabling the measurement of composite similarity between different phase states. Lastly, extra tree (ET) is utilized to fuse the fractional recurrence entropy (FRE) and the singular values extracted from the matrices, thereby achieving SAF detection. Experimental results indicate that the proposed method achieves a detection accuracy of 98.75% and can accurately detect SAFs under various operating conditions. Comparisons with different methods further highlight the advancement of the proposed method. Furthermore, the detection time of the proposed method (209 ms) meets the requirements of standard UL1699B.