International Transactions on Electrical Energy Systems最新文献

The South African rail sector is a key contributor to the national economy, boosting gross domestic product (GDP) and creating jobs. However, serious malfunctions often jeopardize the reliability of locomotives such as the Class 8E locomotive, leading to lost output and longer lead times. Accurate forecasting of the catenary line voltage is essential to ensure timely activation of protective mechanisms and maintain the safe operation of electric traction systems under undervoltage conditions. To reduce unscheduled downtime in the 8E locomotives, this study proposes a framework that analyzes the impact of clustering methods and hyperparameter settings on artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) models. Real-time operational data, including line current, ambient temperature, oil temperature, and line voltage, were gathered on the 8E locomotive at Impala Platinum Mine in Rustenburg, South Africa (SA), between August and October 2024. Three distinct clustering methods, namely, subtractive clustering (SC), grid partitioning (GP), and fuzzy c-means (FCM), along with other key hyperparameters, resulting in a total of 24 developed submodels, were examined and analyzed. The performance of the developed models was analyzed using 7 renowned statistical metrics. With a clustering radius of 0.3, the ANFIS-SC model delivered improvements of 28.45% (MAPE), 28.64% (MAE), 20.80% (SD), 27.53% (CVRMSE), 28.11% (RMSE), and 27.50% (Theil’s U) compared to its ANN counterparts. In addition, better performance was obtained compared to the PSO-based ANFIS model. The study demonstrates the potential of the proposed model as a reliable tool for catenary line voltage in the Class 8E locomotive rail sector in SA.

Integrating electric vehicle (EV)-charging infrastructure presents environmental advantages, particularly in curbing carbon emissions within the transport sector and promoting sustainable energy solutions. However, the ascending adoption of EVs transforms the operational dynamics of low-voltage distribution networks by introducing bidirectional power flows that challenge conventional overcurrent protection schemes. Traditional protection systems cannot effectively manage the complexities of variable load conditions and bidirectional energy transfers, specifically Grid-to-Vehicle (G2V) and Vehicle-to-Grid (V2G) operational modes. These scenarios require the development of advanced, dynamic, and real-time protection mechanisms that are robust against challenging, faulty scenarios and cybersecurity threats. This study introduces an adaptive protection scheme that utilises digital overcurrent relays, LoRa-enabled sensors, a battery management system (BMS) and a central protection unit (CPU). This integrated framework dynamically recalibrates relay settings based on real-time grid conditions, ensuring optimal protection coordination during both G2V and V2G operations by employing a new optimisation algorithm called the transit search algorithm (TSA) and comparing the result to the water cycle algorithm (WCA). To assess the effectiveness of the proposed adaptive approach, simulations were performed on a 33-bus IEEE benchmark network, investigating a variety of fault scenarios and operation grid scenarios. The results indicate that the proposed system significantly mitigates relay miscoordination and reduces fault clearance durations, thus enhancing reliable protection in distribution networks with high EV penetration.

Virtual power plants (VPPs) can aggregate distributed resources across various nodes to participate and collaborate in electricity market trading. Unlike traditional standalone generators or load aggregators, this study leverages the dual role of VPPs as producers and consumers. It introduces a natural risk–hedging mechanism and proposes a price-acceptance bidding strategy for VPPs in the day-ahead spot market, which primarily relies on electricity price forecasts. This strategy is compared with scenarios where distributed resources or traditional generators/load aggregators bid independently. The analysis focuses on the success rate of market participation and the actual financial returns. The findings indicate that the proposed strategy based on the natural risk–hedging mechanism substantially enhances the resilience of VPPs in managing market risks and effectively mitigates the negative impacts of price volatility and forecasting errors on their economic benefits.

Multilevel inverters (MLIs) based on IGBT switches have gained prominence in AC power applications due to their capability to reduce harmonic distortion while offering cost-effective operation. They are widely adopted in power electronic systems; however, under high-stress conditions, power switches are prone to faults that can impair system performance. Hence, effective identification of faulty switches is crucial. This study focuses on detecting single and multiple switch open-circuit faults (OCFs) in reduced device count (RDC) MLI. A machine learning (ML)–based diagnostic framework is proposed, which utilizes only the output voltage signals for fault analysis. From these signals, three key features are extracted: standard deviation, half-cycle moving average, and total harmonic distortion for fault classification. Several ML classifiers are evaluated and benchmarked against recent approaches, with the decision tree (DT) model achieving the highest accuracy of 99.84% under a 70:30 training-to-testing split. The proposed method accurately identified both single and multiple switch OCFs in RDC-MLI within 10–30 ms. The complete diagnostic system is implemented and validated in the MATLAB/Simulink environment.

This article proposes a battery-based hybrid power flow controller (B-HPFC), offering enhanced flexibility for power flow regulation and energy storage. First, the structure of B-HPFC is given, where the cascaded H-Bridge (CHB) is integrated with the phase-shifting transformer (PST). The batteries are connected to the modules of CHB. Doing so, the power flow can be adjusted by tuning the PST and CHB while the batteries can be charged or discharged by tuning CHB. Then, the hierarchical control method is given. The power flow control strategy operates as the outer loop, and the battery control strategy operates as the inner loop. Finally, the proposed structure and control method are verified by the hardware in the loop prototype. The results show that the power flow can be controlled smoothly while the SOC of batteries can be balanced.

In order to accurately evaluate and tap the mutual-aid capacity potential of interconnected power stations under the scenario of peak-to-peak compensation between new energy output and load demand, this paper uses Copula function to describe the correlation structure of wind, light, and load from the perspective of source–load matching, quantify the complementary degree of residual power and new energy output after source–load matching, and determine the feasible interval of mutual aid. On this basis, a space–time mutual-aid capacity optimization model with the goal of minimizing the total operating cost of the interconnected area is constructed. The model takes the principle of priority mutual aid in the station area and comprehensively considers the internal and external energy interaction constraints such as the transaction cost of purchasing and selling electricity with the superior power grid, the two-way power constraints of the tie line, the transformer capacity, and the renewable energy output constraints. Finally, the model is solved efficiently by the CPLEX solver of MATLAB. The simulation results of the example show that the proposed method can automatically establish cross-regional mutual power channels in the period of significant complementarity and significantly improve the renewable energy consumption level and overall operation economy of the station area while ensuring load power supply.

Wind energy conversion systems (WECSs) require robust and efficient control strategies to ensure optimal energy conversion. This study proposes a nonlinear and resilient control approach using a fractional-order proportional integral- and fractional-order proportional derivative (FOPI–FOPD) controller for direct power regulation of a doubly fed induction generator (DFIG)–based WECS. To meet the control objectives, two cascaded FOPI–FOPD controllers were designed, resulting in 12 parameters requiring precise tuning. To optimize these parameters, the Gazelle optimization algorithm (GOA) was employed, targeting the minimization of key performance-based cost functions: mean error (ME), mean absolute error (MAE), mean-square error (MSE), and integral time absolute error (ITAE). These functions integrate dynamic response criteria such as overshoot, rise time, and settling time. Simulation results highlight the effectiveness of the GOA-tuned FOPI–FOPD controller, particularly when using ITAE as the optimization criterion. The controller significantly reduces power ripples by 86.13% in active power and 75.66% in reactive power. It also improves transient response by reducing rise time by 0.035 ms, settling time by 0.3 ms, and completely eliminating overshoot. Moreover, the proposed strategies lower the current total harmonic distortion (THD) by approximately 21.43% compared to the basic strategy. The proposed ITAE–GOA–FOPI–FOPD technique ensures system stability and enhances performance across various operating conditions.

DC series arc faults (DC SAF) in photovoltaic (PV) systems can lead to electrical fires and electric shock hazards. Therefore, DC SAF modeling and detection is a significant process for ensuring the safety of PV panels and is necessary for producing PV systems in actual applications. Using real data, for the first time, this study presents a DC SAF modeling technique based on machine learning (ML) algorithms. Considering the unpredictable and nonlinear nature of such arcs and the application of ML in solving nonlinear and complex problems, multilayer perceptron, radial basis function, and support vector machine algorithms are used to model DC SAF in PV systems. The performance of proposed ML-based approaches is compared with well-known traditional models by using error indices, which are computed using a test data set. Finally, comprehensive evaluations and results of modeling demonstrate that proposed models based on ML methods remarkably improved modeling accuracy and generalization capability in DC SAF modeling.

One renewable energy (RE) source that shows promise for producing electrical energy is wind energy (WE). The coordination between the grid and WE conversion systems has become necessary due to high wind power penetration into the grid and varying wind speeds (VWSs). When incorporated into the grid, wind systems encounter challenging scenarios, including voltage fluctuations, power loss, and the troublesome dynamics of RE sources. Conventional PI control systems and fuzzy logic controllers (FLCs) face difficulties in resolving these problems. Applying hybrid artificial neural networks (ANN) will enhance the efficiency of the VWS system. The suggested controller can facilitate uninterrupted power transmission between generators and the grid, enabling a seamless connection with the grid. Here, it can all be facilitated by the constant voltage and power source supplied by a suggested controller. The training of hybridized ANNs with model predictive control (MPC) can minimize computing demands and device version errors. For ANN-MPC, the WE systems for DC microgrids are optimal. Simulink simulations in MATLAB/Simulink are conducted using the suggested hybrid ANN controller. The proposed ANN can consistently achieve better voltage balance and accuracy across various loading cases compared to conventional FLC and PID controllers. The outcomes demonstrate this. The outcomes of these simulations verify the efficiency of the ANN-based strategy. With an accuracy rate of 92.6% and a performance rate of 95.8%, the proposed hybrid ANN-MPC model outperforms similar current methods, as demonstrated by the experimental results.

With the continued growth in energy consumption, the installed capacity of clean energy, represented by wind power, is steadily increasing. However, the precise modeling of newly built wind farms is challenging due to a lack of data. Additionally, the dynamic updates of data associated with the wind farm’s operating conditions and the difficulty in capturing time-varying features further complicate accurate wind power forecasting. In response to these challenges, this paper proposes a wind power prediction method tailored for the data-scarce scenario of newly constructed wind farms. To prevent over-reliance on single-source domain data, a similarity measurement method combining Mahalanobis distance and dynamic time warping (DTW) is used to establish a multisource transfer learning-based pretrained model using a dilated convolutional neural network–bidirectional long short-term memory (DCNN–BiLSTM) network. Furthermore, to better capture the influence of time-varying scenario data on prediction accuracy, an online adaptive module-based prediction method is introduced to enhance the model’s generalization ability. Additionally, the elastic online deep learning (EODL) method is applied to address the issue of concept drift in dynamic streaming data, enabling quick adaptation to changes in data distribution. The proposed method is validated using data from a wind farm cluster in Northwestern China, demonstrating its superior ability to filter source domain data and provide more accurate power predictions.