Chengke Zhou

There is a growing concern about the high degree of non-technical losses (NTL) in developing countries especially sub-saharan Africa. Whereas several studies have employed artificial intelligence (AI) to analyze NTL, a major drawback in these studies is the focus on customer data only without considering the possible contribution of electricity distribution staff to NTL. This study introduces a novel approach to NTL reduction by analyzing a combined dataset of staff operational processes and customer consumption data. A deep-learning architecture called non-technical losses convolutional neural network (NTLCONVNET) was developed which consists of a series of three one-dimensional convolutional neural networks (1D-CNN) with different depths combined with several fully connected layers. Furthermore, limited or no research has studied the decision rationale influencing how AI models interpret the significance of features in predicting NTL. To achieve the explainability of the model, SHapley Additive exPlanations (SHAP) kernel and tree-based explainers were used for the deep and ensemble learning models respectively to determine the relative importance of the variables and how they contribute to the overall model prediction. A novel ranking framework was used to compute the holistic ranking of the variables across multiple models. The finding suggests that the staff-related variables omitted in the extant literature are significant predictors of NTL. The NTLCONVNET was compared with 5 ensemble learning algorithms and the results show that the NTLCONVNET significantly surpasses all other models, scoring 0.844, 0.838, 0.836 and 0.836 on weighted average Precision, Recall, f1 and accuracy respectively. This study suggests a policy outcome of introducing human resource metrics into NTL reduction strategies.

Sheath currents in high voltage cable circuits have proven to be one of the most convenient indicators for assessing cable health. The sheath currents are regularly distributed with fluctuation in a certain range under normal conditions, while the principle of distribution characteristics is still unclear when the cross-bonded grounding system of the cable circuit is abnormal. This paper classifies the common defects of grounding system into three types of circuit topologies and analyzes the correspondence between grounding system conditions and sheath currents. A fault diagnosis method is proposed to detect irregular distribution of sheath currents by quantifying the difference of sheath currents in each of the minor sections and the average currents of each sheath loop. Sensitive analyses are provided to evaluate the criteria, clarifying the influence of fault resistance in each defect. The results show that the method can be very effective when the ratio of the fault resistance to the earth resistance is less than 7.6 for a short-circuit defect in metallic sheath, or less than 1.2 for a short-circuit defect between metallic sheath and ground. The performance of the method is also shown to be very efficient by evaluation with the data collected on site.

Power cable condition diagnosis and deterioration location rely on signatures of aging characteristics which precede the final breakdown. The purpose of this study was to investigate how to diagnose and locate the aging and/or deterioration of power cables through the analysis of the impedance spectroscopy. The concepts of the reference frequency and characteristic frequency of cable impedance spectroscopy are defined for the first time. Based on the reference frequency, the optimal frequency range for analysis of impedance spectroscopy can be determined, whilst based on characteristic frequency, a set of criteria for assessing cable conditions are examined and established. The solution proposed in this paper has the advantage of being easier to implement than the previously reported “broadband” impedance spectroscopy methods, as it helps to reduce the frequency range for measurement instrumentations; the proposed method also does not need the measurements of the parameters of the cable being tested.

Urban power transmission and distribution rely on safe operation of vast network of power cables. Majority cable insulation defects can result in partial discharge (PD), which is therefore a degradation mechanism as well as a good indicator of cable insulation condition. Previously published PD models, however, failed to help quantitatively evaluate the insulation condition, or they fail to help differentiate those in between the conditions which are safe to operate from those which may need urgent testing or inspection. PD mechanisms in power cables are reviewed first in this paper. Then a novel model which is based on F distribution is proposed aiming to classify the PD activities which may be associated with different level of degradation. This is based on previous reports that PD activities manifests differently when the levels of degradation of the defect site(s) progress. After that, the features and effects of the model are presented with a case study, and the model is verified by analyzing the waveform similarities. Finally, the nature of the model is compared with the abc-model and the dipole model, demonstrating that proposed model can uniquely recognize the three stages of PD defect development.

Infrared thermography has been used as a key means for the identification of overheating defects in power cable accessories. At present, analysis of thermal imaging pictures relies on human visual inspections, which is time-consuming and laborious and requires engineering expertise. In order to realize intelligent, autonomous recognition of infrared images taken from electrical equipment, previous studies reported preliminary work in preprocessing of infrared images and in the extraction of key feature parameters, which were then used to train neural networks. However, the key features required manual selection, and previous reports showed no practical implementations. In this contribution, an autonomous diagnosis method, which is based on the Faster RCNN network and the Mean-Shift algorithm, is proposed. Firstly, the Faster RCNN network is trained to implement the autonomous identification and positioning of the objects to be diagnosed in the infrared images. Then, the Mean-Shift algorithm is used for image segmentation to extract the area of overheating. Next, the parameters determining the temperature of the overheating parts of cable accessories are calculated, based on which the diagnosis are then made by following the relevant cable condition assessment criteria. Case studies are carried out in the paper, and results show that the cable accessories and their overheating regions can be located and assessed at different camera angles and under various background conditions via the autonomous processing and diagnosis methods proposed in the paper.

The associated daily maintenance effort and cost has been increasing with the growing use of high‐voltage (HV) cable circuits. This study proposes a time‐delay concept‐based approach to maintenance scheduling of individual HV cable components, to allow periodicity of inspection to be optimised with consideration of time‐delay interval between the instant of defect inception and final failure. Here, the progress from a defect to its resultant final failure can be represented as a two‐stage process, one being a stochastic process of defect arrival and the other being a delay‐time interval from defect to failure. Then an optimisation model is established to schedule maintenance activities based on the two statistical processes, where the objective function is set as the minimal maintenance cost. Cable circuits are divided into five components in accordance with the objects of inspection and testing. The components within different service ages are considered individually in the statistical models, and their accumulate failure numbers during each service year are used to automatically formulate and update periodicity of inspection. To illustrate the feasibility of the approach proposed, theoretical analysis on thousands of cable circuits are carried out. Results show that the economic expenditure could be reduced by 22.2% after optimisation.

Among all failures in cable circuits, the failure rate of cable joints is much higher than that in cable bodies. In an effort to reduce cable failure rates and improve the reliability of power cable systems, this article studies the possibility of increasing the length of a cable section and thus reducing the number of cable joints via alternative sheath bonding methods. As the maximum length of an electrical cable section is governed by the permissible sheath standing voltage, this article first proposes a precise sheath voltage calculation model without approximations and simplifications. The sheath voltage per unit length has then been calculated for each bonding method using the proposed model. The maximum length of a cable section has then been calculated, based on the permissible sheath standing voltage, for the various sheath bonding methods. Results show that the traditional 500‐m minor section length could be increased to 2.21 km by using the single‐point bonding method. The maximum distance can be further lengthened by increasing the physical spacing among the three‐phase cables. The economic comparison for each bonding method has also been made. Finally, the proposed method has been applied to a HV cable circuit already in operation, and the operational data confirms the feasibility.

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 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.