Energy Informatics最新文献

To improve the adaptability of grid load collaborative scheduling, a multi-objective collaborative scheduling method based on a simulated annealing algorithm for the load storage of grid loads on the load side of a new power system is proposed. Local bus transmission technology is adopted to collect the dynamic parameters of energy network load energy storage on the load side of the new power system. The collected load dynamic parameters are fused with energy distribution state parameters to extract the state characteristics of energy network load storage. The simulated annealing algorithm is adopted to realize the load characteristics fusion and adaptive scheduling processing of energy network on the load side of the power system, and the spectral characteristics of the load dynamic parameters are extracted. The dynamic scheduling method of simulated annealing is used to realize the multi-objective optimization of dynamic load of energy network. Based on the co-optimization results of simulated annealing, the optimization application of the simulated annealing algorithm in the multi-objective co-scheduling of loads and energy storage in a new power system is realized. The experimental results show that after 400 iterations, the control convergence accuracy of the proposed method reaches 0.980, which is significantly better than that of the comparison method, and performs well in terms of scheduling efficiency improvement, load scheduling stability, scheduling time and energy waste ratio, proving that the method has good multi-objective integration and strong optimization ability in the scheduling process, and improves the load balanced scheduling and adaptive control ability of the power system.

This study aims to enhance the intelligence level of power grid fault analysis to address increasingly complex fault scenarios and ensure grid stability and security. To this end, an intelligent information system for power grid fault analysis, leveraging improved computer communication technology, is proposed and developed. The system incorporates a novel fault diagnosis model, combining advanced communication technologies such as distributed computing, real-time data transmission, cloud computing, and big data analytics, to establish a multi-layered information processing architecture for grid fault analysis. Specifically, this study introduces a fusion model integrating Transformer self-attention mechanisms with graph neural networks (GNNs) based on conventional fault diagnosis techniques. GNNs capture the complex relationships between different nodes within the grid topology, effectively identifying power transmission characteristics and fault propagation paths across grid nodes. The Transformer’s self-attention mechanism processes time-series operational data from the grid, enabling precise identification of temporal dependencies in fault characteristics. To improve system response speed, edge computing moves portions of fault data preprocessing and analysis to edge nodes near data sources, significantly reducing transmission latency and enhancing real-time diagnosis capability. Experimental results demonstrate that the proposed model achieves superior diagnostic performance across various fault types (e.g., short circuits, overloads, equipment failures) in simulation scenarios. The system achieves a fault identification and location accuracy of 99.2%, an improvement of over 10% compared to traditional methods, with an average response time of 85 milliseconds, approximately 43% faster than existing technologies. Moreover, the system exhibits strong robustness in complex scenarios, with an average fault prediction error rate of just 1.1% across multiple simulations. This study provides a novel solution for intelligent power grid fault diagnosis and management, establishing a technological foundation for smart grid operations.

In order to provide a reliable basis for the cost management of photovoltaic power generation, it is necessary to accurately predict the depreciation expense of photovoltaic power generation. Therefore, a hierarchical quantitative prediction method of photovoltaic power generation depreciation expense based on matrix task prioritization considering uncertain risks is proposed. Based on the conditional value-at-risk theory, a more comprehensive risk measure than VaR is provided, and the uncertainty risk value of photovoltaic power generation is calculated by considering the average loss exceeding this loss value. According to the calculated risk value, a double-layer photovoltaic power generation cost planning model is constructed, the upper and lower objective functions of the model are determined, and the constraint conditions are designed; Obtain a cost planning objective function solution base on a matrix task prioritization method, and generating a prioritization table; Prediction of photovoltaic power generation depreciation expense based on long-short memory neural network for each solution in the sorting table. In practical application, the test results show that this method can complete the risk quantitative analysis of uncertain factors, and the tracking ability and fitting degree of prediction are good; An ordered list of solutions of each objective function can be generated; The method in this paper is used to predict the depreciation expense of photovoltaic power generation in the first 10 solutions of priority ranking, and the maximum deviation of the prediction result is -0.65 million yuan.

Extreme weather events, particularly snow and ice storms, present significant threats to the stability and reliability of high-voltage transmission lines, leading to substantial disruptions in power supply. This study delves into the causes and consequences of transmission line trip faults that occur under severe winter conditions, with a focused case study in Inner Mongolia—an area frequently impacted by snow and ice hazards. By systematically analyzing field data collected during critical periods of ice accumulation, this research identifies and examines key factors contributing to faults, such as conductor galloping, insulator degradation, and structural fatigue. These issues are often exacerbated by prolonged exposure to low temperatures and high wind speeds, which further compromise the integrity of transmission infrastructure. In addition to field observations, comprehensive testing of the affected insulators and components reveals mechanical and electrical vulnerabilities that play a significant role in the occurrence of trip faults. To combat these challenges, the paper proposes a series of mitigation and prevention strategies. These include enhancing design specifications to ensure resilience against increased ice and wind loads, deploying real-time monitoring systems capable of detecting early indicators of conductor galloping and ice accumulation, and employing advanced de-icing technologies to reduce the risk of ice-related failures. Moreover, the integration of unmanned aerial vehicles (UAVs) and artificial intelligence (AI)-based fault detection tools presents promising opportunities for improving remote monitoring capabilities and enabling proactive maintenance interventions. By leveraging these innovative technologies, the resilience of transmission lines in harsh climates can be significantly enhanced. The findings of this study not only provide a comprehensive framework for minimizing the impact of extreme weather on transmission infrastructure but also contribute valuable insights toward fostering a more reliable and resilient power grid capable of withstanding the challenges posed by an increasingly volatile climate.

In response to the problem of low prediction accuracy in ultra short-term prediction of photovoltaic power, this study combines Hungarian clustering analysis and particle swarm optimization variational mode decomposition algorithm to construct a photovoltaic power ultra short-term forecasting model, to analyze data in depth and improve prediction accuracy. The experiment outcomes show that the Hungarian algorithm performs well in integrating single clustering results and effectively improves the problem of atypical classification. In addition, the clustering ensemble model shows significant improvement compared to other models on the Calinski-Harabasz index, and effectively reduces the overlap between clusters on the Davies-Bouldin index, improving the overall quality of clustering. Under different weather conditions, the convergence accuracy and speed of the multiverse optimization support vector machine, multiverse optimization support vector machine, and particle swarm optimization variational mode decomposition algorithms each have their own advantages, but the particle swarm optimization variational mode decomposition algorithm performs better. In addition, the Hungarian clustering model has high stability in predicting errors, with average absolute error and average relative error lower than BP and RBF models. The maximum absolute error and maximum relative error are reduced, indicating the effectiveness and predictive advantage of the proposed Hungarian clustering ensemble model in predicting photovoltaic power.

This study seeks to enhance the accuracy and economic efficiency of power system load forecasting (PSLF) by leveraging Artificial Neural Networks. A predictive model based on a Residual Connection Bidirectional Long Short Term Memory Attention mechanism (RBiLSTM-AM) is proposed. In this model, normalized power load time series data is used as input, with the Bidirectional Long and Short Term Memory network capturing the bidirectional dependencies of the time series and the residual connections preventing gradient vanishing. Subsequently, an attention mechanism is applied to capture the influence of significant time steps, thereby improving prediction accuracy. Based on the load forecasting, a Particle Swarm Optimization (PSO) algorithm is employed to quickly determine the optimal scheduling strategy, ensuring the economic efficiency and safety of the power system. Results show that the proposed RBiLSTM-AM achieves an accuracy of 96.68%, precision of 91.56%, recall of 90.51%, and an F1-score of 91.37%, significantly outperforming other models (e.g., the Recurrent Neural Network model, which has an accuracy of 69.94%). In terms of error metrics, the RBiLSTM-AM model reduces the root mean square error to 123.70 kW, mean absolute error to 104.44 kW, and mean absolute percentage error (MAPE) to 5.62%, all of which are lower than those of other models. Economic cost analysis further demonstrates that the PSO scheduling strategy achieves significantly lower costs at most time points compared to the Genetic Algorithm (GA) and Simulated Annealing (SA) strategies, with the cost being 689.17 USD in the first hour and 2214.03 USD in the fourth hour, both lower than those of GA and SA. Therefore, the proposed RBiLSTM-AM model and PSO scheduling strategy demonstrate significant accuracy and economic benefits in PSLF, providing effective technical support for optimizing power system scheduling.

This study aims to develop an efficient model to predict the arrears behavior of electricity users by integrating multi-scale feature learning with a backpropagation (BP) neural network. The goal is to provide accurate early warning systems and enhanced risk management tools for power companies. The BP neural network algorithm adjusts weights to minimize prediction errors, while multi-scale feature learning captures the diversity and regularity of user behavior by extracting data from various time dimensions, such as daily, weekly, and monthly intervals. First, electricity usage and weather data from the UMass Smart Dataset are preprocessed, including steps such as data cleaning, standardization, and normalization. Next, features are extracted across three time scales—daily, weekly, and monthly. These features are then input into the BP neural network model using the multi-scale feature learning method. A hierarchical neural network structure is designed to address the characteristics of different scales in distinct layers. Key model parameters are optimized, and a sensitivity analysis is conducted. The experimental results demonstrate that the BP neural network model incorporating multi-scale features outperforms traditional BP neural network models and other control models in several evaluation metrics. Specifically, the Gini coefficient is 0.55, the Kolmogorov-Smirnov statistic is 0.60, the Matthews correlation coefficient is 0.45, and specificity is 0.82. These results indicate that the proposed method offers significant improvements in capturing user behavior patterns and enhancing prediction accuracy. The study concludes that the effective fusion of multi-scale features not only enhances the model’s prediction performance but also strengthens its generalization ability. This method provides an advanced risk management tool for power companies, helping to increase the operational efficiency of smart grids and encouraging further research toward greater intelligence in the field.

The rapid development of renewable energy has made hydropower’s role as a flexible resource increasingly important in power systems. However, hydropower generation capability highly depends on water inflows, particularly during dry seasons, making it difficult to independently meet growing load demands. The application of hydro-wind-photovoltaic-storage systems offers a promising solution, yet faces challenges from the high-dimensional uncertainties in natural conditions. This paper proposes a capacity planning method that considers high-dimensional uncertainties characterized by spatiotemporal correlations of natural factors. Firstly, a scenario generation method based on the transition probability matrix and C-Vine Copula model is developed. The constructed scenario sets capture the temporal correlations of natural conditions and spatial correlations between different parameters. Secondly, a bi-level optimization model for capacity planning is established. The upper level minimizes the deviation of operational cost and grid supply revenue to determine optimal capacity allocation, while the lower level optimizes both economic and safe objectives for operational dispatch. The normal boundary intersection method is employed to obtain Pareto front solutions that balance economy and safety. Different case studies are conducted to validate the effectiveness of the proposed method. Compared with the fixed ratio and variable ratio capacity allocation strategies without uncertainty, the optimal total system cost is reduced by 2.90% and 3.88%, respectively.

Integrating power forecasting with wind turbine maintenance planning enables an innovative, data-driven approach that maximizes energy output by predicting periods low wind production and aligning them with maintenance schedules, improving operational efficiency. Recently, many countries have met renewable energy targets, primarily using wind and solar, to promote sustainable growth and reduce emissions. Forecasting wind turbine power production is crucial for maintaining a stable and reliable power grid. As renewable energy integration increases, precise electricity demand forecasting becomes essential at every power system level. This study presents and compares nine machine learning (ML) methods for forecasting, Interpretable ML, Explainable ML, and Blackbox model. The interpretable ML includes Linear Regression (LR), K-Nearest Neighbors (KNN), eXtreme Gradient Boosting (XGBoost), Random Forest (RF); the explainable ML consists of graphical Neural network (GNN); and the blackbox model includes Multi-layer Perceptron (MLP), Recurrent Neural Network (RNN), Gated Recurrent Unit (GRU), and Long Short-Term Memory (LSTM). These methods are applied to the EDP datasets using three causal variable types: including temporal information, metrological information, and power curtailment information. Computational results show that the GNN-based forecasting model outperforms other benchmark methods regarding power forecasting accuracy. However, when considering computational resources such as memory and processing time, the XGBoost model provides optimal results, offering faster processing and reduced memory usage. Furthermore, we present forecasting results for various time windows and horizons, ranging from 10 minutes to a day.

The reliability of power communication networks is vital to ensure uninterrupted operation in power electronics. Self-healing techniques address this need by automating fault identification and recovery. However, existing methods struggle with dynamic challenges like voltage fluctuations, thermal overloads, and multidimensional sensor data, often leading to delays in fault recovery and reduced safety. This study aims to develop the Self Heal Power Safe Predictor (SHPSP) model to overcome the limitations of prior self-healing techniques. The primary objectives include improving fault prediction accuracy, enhancing recovery speed, and ensuring resilience under diverse and high-stress operational conditions. The SHPSP model employs an ensemble-based classification strategy within a majority voting framework, focusing on multidimensional sensor data such as voltage, temperature, and safety indicators. Feature selection is optimized using ensembled filter and wrapper techniques to prioritize critical parameters. The model is validated against conventional methods using metrics like accuracy, precision, recall, F1-score, and MCC. Experimental results demonstrate that the SHPSP model significantly outperforms previous approaches, achieving higher fault detection accuracy and faster recovery, particularly during voltage drops, power surges, and thermal stress. The SHPSP classifier obtained 91.4% accuracy, 88.2% precision, 89.5% recall, 89.8% F1-score, 81.0% MCC, and a 92.0% ROC-AUC curve. The SHPSP model ensures enhanced safety, dependability, and robustness for power electronics systems, marking a significant advancement in self-healing technology.