It is a great challenge to differentiate Partial Discharge (PD) induced by different types of insulation defects in high voltage cables. Some types of PD signals have very similar characteristics and are specifically difficult to be differentiate, even for the most experienced specialists. To overcome the challenge, a Convolutional Neural Network (CNN) based deep learning methodology for PD pattern recognition is presented in this paper. Firstly, PD testing for five types of artificial defects in Ethylene-Propylene-Rubber (EPR) cables was carried out in the High Voltage (HV) laboratory to generate signals containing PD data. Secondly, 3500 sets of PD transient pulses were extracted and then 33 kinds of PD features were established. The third stage applies a CNN to the data: typical CNN architecture and the key factors which affect the CNN based pattern recognition accuracy, are described. Factors discussed include the number of the network layers, convolutional kernel size, activation function, and pooling method. The paper presents a flowchart of the CNN based PD pattern recognition method and an evaluation with 3500 sets of PD samples. Finally, the CNN based pattern recognition results are shown and the proposed method is compared with two more traditional analysis methods, i.e. Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN). The results show that the proposed CNN method has higher pattern recognition accuracy than SVM and BPNN, and that the novel method is especially effective for PD type recognition in cases of signals of high similarity, which is applicable for industrial applications.

Optimal selection of features of Partial Discharge (PD) signals recorded from defects in High Voltage (HV) cables will contribute not only to the improvement of PD pattern recognition accuracy and efficiency but also to PD parameter visualization in HV cable condition monitoring and diagnostics. This paper presents a novel Random Forest (RF)-based feature selection algorithm for PD pattern recognition of HV cables. Firstly, the experimental data acquisition and feature extraction processes are introduced. PD signals were captured from faults created in a cable to obtain the raw PD data, then a set of 3500 transient PD pulses and a set of 3500 typical interference pulses were extracted, based on which 34 PD features were extracted for further processing. Furthermore, 119 two-dimensional features and 1082 three-dimensional features were generated. The paper then discusses the basic principle of the RF algorithm. Finally, RF-based feature selection was implemented to determine the optimal features for PD pattern recognition. The results were obtained and evaluated with the Back Propagation Neural Network (BPNN) and Support Vector Machine (SVM). Results show that the proposed RF-based method is effective for PD feature selection of HV cables with the potential for application to additional HV power apparatus.

This paper proposes the use of the Cox proportional hazard model (Cox PHM), a statistical model, for the analysis of early-failure data associated with power cables. The Cox PHM analyses simultaneously a set of covariates and identifies those which have significant effects on the cable failures. In order to demonstrate the appropriateness of the model, relevant historical failure data related to medium voltage (MV, rated at 10 kV) distribution cables and High Voltage (HV, 110 kV and 220 kV) transmission cables have been collected from a regional electricity company in China. Results prove that the model is more robust than the Weibull distribution, in that failure data does not have to be homogeneous. Results also demonstrate that the method can single out a case of poor manufacturing quality with a particular cable joint provider by using a statistical hypothesis test. The proposed approach can potentially help to resolve any legal dispute that may arise between a manufacturer and a network operator, in addition to providing guidance for improving future practice in cable procurement, design, installations and maintenance.

This two-part paper proposes a revised method for calculating HV (High Voltage) cable sheath currents, and for fault location of cross-bonded HV cable systems. The equivalent circuit and the cable impedance per unit length under short-circuit fault conditions are revised from those under non-fault conditions. Part 1 of this paper proposes a revised phasor domain mathematical model for the sheath current calculation under fault conditions, following demonstration that short-circuit faults can cause changes in the equivalent circuit of the cable metal sheath, and causing changes in the cable impedance per unit length. The calculation results, as given in a case study, show that there exists a unique relation between the fault position and the sheath fault currents, making it possible to locate a fault by analyzing the characteristics of the sheath currents during a fault. Part 2 of the paper focuses on the fault location method via sheath current monitoring.

In order to improve the practice in maintenance of power cables, this paper proposes a novel traveling-wave-based fault location method improved by unsupervised learning. The improvement mainly lies in the identification of the arrival time of the traveling wave. The proposed approach consists of four steps: (1) The traveling wave associated with the sheath currents of the cables are grouped in a matrix; (2) the use of dimensionality reduction by t-SNE (t-distributed Stochastic Neighbor Embedding) to reconstruct the matrix features in a low dimension; (3) application of the DBSCAN (density-based spatial clustering of applications with noise) clustering to cluster the sample points by the closeness of the sample distribution; (4) the arrival time of the traveling wave can be identified by searching for the maximum slope point of the non-noise cluster with the fewest samples. Simulations and calculations have been carried out for both HV (high voltage) and MV (medium voltage) cables. Results indicate that the arrival time of the traveling wave can be identified for both HV cables and MV cables with/without noise, and the method is suitable with few random time errors of the recorded data. A lab-based experiment was carried out to validate the proposed method and helped to prove the effectiveness of the clustering and the fault location.

In order to improve the practice in the operation and maintenance of high voltage (HV) cables, this paper proposes a fault location method based on the monitoring of cable sheath currents for use in cross-bonded HV cable systems. This method first analyzes the power–frequency component of the sheath current, which can be acquired at cable terminals and cable link boxes, using a Fast Fourier Transform (FFT). The cable segment where a fault occurs can be localized by the phase difference between the sheath currents at the two ends of the cable segment, because current would flow in the opposite direction towards the two ends of the cable segment with fault. Conversely, in other healthy cable segments of the same circuit, sheath currents would flow in the same direction. The exact fault position can then be located via electromagnetic time reversal (EMTR) analysis of the fault transients of the sheath current. The sheath currents have been simulated and analyzed by assuming a single-phase short-circuit fault to occur in every cable segment of a selected cross-bonded high voltage cable circuit. The sheath current monitoring system has been implemented in a 110 kV cable circuit in China. Results indicate that the proposed method is feasible and effective in location of HV cable short circuit faults.