Improving the connectivity, security, and throughput of near and far vehicles in cooperative non-orthogonal multiple access (C-NOMA) communication with energy-constrained relays remains challenging, especially in the presence of eavesdropping E attacks. Existing works consider a single energy harvesting (EH) without a direct link (DL) between the base station (BS) and users, thus limiting the performance for both near and far users. To address the above challenges, this study proposes a hybrid EH and secure cooperative NOMA (HSCNOMA) framework. The near user harvests energy from the received signal and operates as an energy-constrained relay to assist the far user. In contrast, the far user employs selection combining (SC) to select the strongest signal from direct and relayed links. Closed-form expressions for outage probability (OP), secrecy outage probability (SOP), and throughput (T) are derived under perfect successive interference cancellation (SIC). Numerical results illustrate that the proposed framework improves the performance of both near and far users. Outage probabilities of around 0.003 and 0.002 are achieved for the near and far vehicles, respectively, while secrecy outage probabilities are reduced to about 0.020 and 0.090. Furthermore, the proposed approach achieves improved throughput for both vehicles. It attains a higher system throughput of nearly 0.6 (normalized value), outperforming conventional time-switching-only, powersplitting-only, and cooperative NOMA schemes without a direct link.
In real-world contexts, most systems are not one hundred percent reliable. Breakdowns in the service area occur due to external shocks, which has given rise to the study of stochastic models with specific research into server breakdowns in queuing theory [1]. Queuing models with backup servers to the primary server have found many applications in computer systems and wireless communications. When the primary server breaks down, the backup server takes over and acts as the primary server working at a reduced rate during a working vacation. However, the sensitivity of the parameters on various performance indices is difficult to explore since the parameters of optimality are static. In order to address this issue, this article studies the performance model for an M/(M1, M2)/1 queue with working vacation and vacation interruptions, which is implemented and improved using a genetic algorithm (GA). Each of the parameters of the queuing model and its effects on the queue length were investigated and it was found that the GA produced improved results and allowed for more flexibility in the design of the queuing system. The proposed GA model offered an average of 10.79% lower expected queue length. These investigations conducted through simulations indicate that the proposed GA can be used with the queuing model presented to provide an easier way to design M/(M1, M2)/1 queuing systems with working vacations and vacation interruptions with lower queue lengths.
Power transformers are critical assets that require reliable monitoring due to their high economic and operational importance. Dissolved Gas Analysis (DGA) is extensively used as a condition monitoring technique because of its ability to detect internal faults without interrupting operation. However, its effectiveness is often constrained by reliance on expert knowledge and inconsistencies among traditional interpretation techniques. To address these challenges, this study explores the use of intelligent computational methods to enhance diagnostic accuracy. An Artificial Neural Network (ANN) model and Fuzzy Inference Systems (FIS) are evaluated using transformer DGA data. The ANN utilises a feedforward backpropagation algorithm with Bayesian regularization, while data preprocessing includes dimensionality reduction using Principal Component Analysis (PCA) and dataset balancing through Adaptive Synthetic Sampling (ADASYN). In parallel, fuzzy logic is incorporated into four conventional DGA approaches: IEC, Rogers, Doernenburg, and Duval Triangle to assess whether uncertainty handling improves classification performance. The proposed ANN model demonstrated improved classification performance reaching an accuracy of approximately 77% on unseen data, exceeding the performance of all conventional techniques. Among the fuzzy-enhanced methods, only the IEC approach demonstrated noticeable improvement. The findings indicate that ANN-based models offer a more reliable and consistent alternative for transformer fault diagnosis compared to both classical and fuzzy-modified techniques.
Electricity theft represents a significant source of non-technical losses (NTLs) in power distribution networks, leading to substantial financial deficits, grid instability, and safety concerns. This study introduces a machine learning-based methodology to detect and predict electricity theft by examining residential consumption patterns in Pietermaritzburg Municipality, South Africa. Utilising meter data from approximately 42,000 customers, of whom 21,935 made at least one electricity purchase over a five-year period (2019-2024), the research evaluates five machine learning techniques: decision trees, logistic regression, K-nearest neighbours, XGBoost, and random forests. The approach integrates survey data, audit outcomes, and audit evaluations for a holistic analysis. Results indicate that Random Forest outperforms other models, achieving the highest scores across accuracy (0.8), precision (0.75), recall (0.9), and F1-score (0.82) for detecting tampering. Survey findings reveal that 85.0% of respondents link theft to financial hardship, underscoring socio-economic influences. This study provides utilities with practical insights to reduce NTLs, enhance grid reliability, and foster sustainable energy management.
When a miniature circuit breaker (MCB) interrupts the flow of current due to a short circuit, it is a big event compared to the onload opening of contacts that causes internal damage to the MCB. Internal damage is not visible as a result of the construction of the MCB and is not easily measured while used. Thermal degradation is mainly due to the temperature of the arc and the time it takes to dissipate the arc when the contacts are opened. The erosion of the contact material is due to the heat flux of the arc during the opening, closing, or bouncing of contacts. This paper investigates the heat stress exerted on contacts during a short circuit and how that relates to the degradation of the contacts. Although the arc is short in duration, the energy is high at high temperatures. An Ansys model was developed to simulate how an MCB would radiate Joule heat based on default Ansys material property library values, and then after degradation has set in, by using experimentally measured values in the Ansys model. The Ansys simulations were verified through thermal images of an MCB internal conduction path that were subjected to a number of short circuits. The default library values had an over 26% discrepancy between the infrared (IR) and Ansys values. After adding the degradation calculation values to the Ansys simulation, the discrepancy decreased to 0.98%. The radiated heat measured with the thermal imager.
