Power grid engineering is necessary for the distribution of power, but problem detection and categorization are made extremely difficult by their growing complexity, particularly with the incorporation of renewable energy sources. Decision trees and support vector machines are two examples of fault detection techniques that frequently fail to handle the dynamic and non-stationary character of power grid data. These techniques’ efficacy in real-time defect identification is limited because they are unable to capture the complex relationships and temporal dependencies present in time-series data. Furthermore, a lot of conventional models are unable to generalize to different kinds of problems, which results in errors and delays in fault identification. For improved fault detection and classification in power grids, this research suggests a hybrid approach that combines Temporal Fusion Transformers (TFT) with Long Short-Term Memory (LSTM) Neural networks. While the LSTM neural network is used to represent sequential data, the TFT model is particularly good at capturing complicated linkages in time-series data. By combining these two models, the suggested approach can improve grid efficiency and dependability by offering precise fault forecasts in real-time. The findings show that the suggested hybrid model works better than conventional techniques, with a fault classification accuracy of 98.66% as opposed to decision trees’ 97.42% and SE-CDAE’s 97.98% accuracy. The performance of the model demonstrates its potential for real-time implementation in contemporary power grids, guaranteeing improved fault identification and prompt reactions to avert system breakdowns.
Underground mine cable distribution systems are critical industrial local energy infrastructures, where ground faults can lead to production interruption, difficult fault searching, and safety risks. This paper proposes a ground-wire active-current based fault-location method for smart monitoring of underground mine cable distribution systems. The method uses zero-sequence voltage as the fault trigger and phase reference, and measures ground-wire currents at both ends of each monitored cable section. By extracting the active components of the head-end and tail-end ground-wire currents, a fault judgment quantity is constructed through their algebraic summation. Theoretical analysis shows that, for non-faulted cable sections, the active ground-wire current mainly behaves as through-current and is largely cancelled by the double-ended summation. In contrast, for the faulted cable section, the fault current flows from the fault point toward both ends, producing a dominant active-current summation. Simulation results under transition resistances from 0 Ω to 1000 Ω verify that the proposed method can correctly identify internal cable-section faults and distinguish terminal busbar/switchgear faults without falsely locating a healthy cable section. A distributed smart monitoring architecture integrating ground-current sensors, zero-sequence voltage measurement, RS485/LoRa communication, and master-station processing is also presented. The proposed method provides an interpretable and practical solution for online fault location, faster fault isolation, and improved reliability of underground mine local energy distribution systems.
Industrial electricity load forecasting is crucial for the stable operation of power systems and energy management. However, the complex temporal patterns and dynamic interdependencies between loads from different industries make it challenging for traditional forecasting methods to model effectively. To address this, this paper proposes a forecasting model based on an inter-industry association dynamic graph neural network that integrates periodic features. The proposed method uses the Time2Vector block to adaptively capture multiple periodic fluctuations in the load sequence, and combines this with the cointegration relationship and error correction mechanism from the Vector Error Correction Model (VECM) to quantify the association strength between industries. Each industry is represented as a node, and the association strengths define the edges and their weights. Thus, this forms a Dynamic Inter-industry Association Graph (DIAG). This graph is then integrated into a Dynamic Spatial-Temporal Aware Graph Neural Network framework. As a result, the Inter-industry Association Dynamic Graph Neural Network (IADGNN) is formed. This model captures the complex dynamic characteristics of electricity loads across different industries. Test cases based on industrial load data from one province in China show that this method significantly outperforms traditional models in terms of forecasting accuracy, providing a novel solution for addressing the complex industrial load forecasting problem.
With the high penetration of distributed energy resources (DERs), the strong uncertainty of their output and the complex interaction of network topology pose severe challenges to the real-time performance, scalability, and intelligence level of dispatching methods. This paper proposes a hierarchical distributed collaborative intelligent dispatching architecture based on deep reinforcement learning (DRL) and graph neural networks (GNNs). This architecture adopts a two-layer design of “regional agent-edge controller.” The upper-layer regional agent uses GNN to process the global power flow state and generate node marginal price signals; the lower-layer edge controllers act as independent DRL agents, making autonomous decisions based on local observations and price signals. Through a hybrid mechanism of centralized training and distributed execution, collaborative learning and plug-and-play expansion of the system are achieved. This method significantly reduces dispatch costs when dealing with uncertainty, achieving a high photovoltaic grid integration rate of up to 100%, and controlling the maximum net power deviation between day-ahead planning and real-time operation to within 0.35 MW. Even when the system is expanded to 500 DER units, the online decision-making time remains at 7.5 milliseconds, and communication overhead only slowly increases to 35.0 KB, demonstrating its significant effectiveness in improving system resilience, operational economy, and scalability.
The current trends in electric vehicle adoption in distribution networks have brought some positives into microgrid operations. EVs can reduce greenhouse gas emissions down and create a more flexible source of energy by using V2G technology; however, uncontrolled charging of these vehicles is likely to incur losses in energy, fluctuations in voltage, and reduced efficiency to the system in total. This research proposes an optimal scheduling framework for the charging/discharging of EVs understandably connected to distribution systems embedded in microgrids. Analysis and classification of forecasted EV charging will be done based on charging data obtained from Kaggle; the charging behaviors will be classified into peak/off-peak utilization and flexible/inflexible groups, wherein energy losses are minimized, and distribution system efficiency is maximized through a multi-objective optimization model with realistic operating conditions. Simulation results on the IEEE 33-bus and 69-bus test systems demonstrate that with suitable coordination of scheduling, energy loss is reduced up to 25%, voltage stability is ensured, green energy is utilized maximally, and load relief during the peak period takes place. The results from this study highlight how smart scheduling of EVs could enhance smart grid performance in the future concerning the technology, economic, and environmental aspects.
