In the absence of explicit meteorological observations, the correlation among wind power, photovoltaic output, and load is difficult to characterize directly using conventional weather-driven models, which further increases the difficulty of net-load ramping assessment. To address this issue, this paper proposes a weak meteorological coupling method for wind–solar–load scenario generation and ramping demand assessment based on key meteorological-state labels. First, the load series is decomposed on multiple time scales to extract its trend, periodic components, and random residuals. Then, high-wind and low-wind states are identified on the wind side through continuous weather-process blocks, while PV scenarios are generated based on low-solar states and their overlap probabilities with wind states. Meanwhile, load scenarios are generated through seasonally conditioned residual sampling. In this way, wind–solar–load joint modeling is achieved without relying on explicit meteorological variables. Finally, net load is constructed from the generated joint scenarios to evaluate upward and downward ramping demand as well as extreme ramping risk. Case study results show that the proposed method can preserve the deterministic structure of load while reasonably characterizing the uncertainty of wind and PV output. At the system level, the prediction intervals of both upward and downward ramping demands achieve high coverage rates, indicating that the proposed method can stably reflect the range of bidirectional regulation requirements. Further analysis shows that extreme ramping events are mainly caused by the synchronous rapid drop of wind and PV output during critical periods rather than by abrupt load growth alone. The proposed method can provide useful support for ramping demand assessment of renewable power systems under limited meteorological information.

The energy transition increases wind/solar penetration, raising supply–demand uncertainties. Renewable Energy Sources (RES) intermittency and load variability heighten uncertainties. This study proposes Hierarchical Reinforcement Learning (HRL) with day-ahead scheduling and real-time regulation, enhanced by Generative Adversarial Network (GAN) for uncertainty modeling. A Deep Deterministic Policy Gradient (DDPG)-based reward balances efficiency and stability, while Autoregressive Integrated Moving Average (ARIMA) and Model Predictive Control (MPC) improve temporal matching. Simulation results demonstrate that under the mixed disturbance scenario with 70% renewable penetration, the proposed framework reduces renewable curtailment by 11.2% relative to conventional MPC (from 11.2% to 6.5% in absolute terms), and lowers total operating costs by 9.6% (from $1.487 M to $1.321 M). Under the load mutation scenario, the reductions are 37.1% and 10.8%, respectively. All reported improvements are based on 20 independent runs with averaged results. This offers an adaptive paradigm optimizing generation, grid, load, storage under “dual carbon” goals.

Driven by the global transition toward carbon neutrality, power systems with high penetration of variable renewable energy are confronted with dual challenges: hourly power imbalance induced by short-term renewable fluctuations, and seasonal energy mismatch caused by inter-month and inter-annual climatic variations. This paper systematically reviews the state-of-the-art multi-timescale risk-aware planning methods, and this review systematically identifies three critical research gaps in current studies. On this basis, it develops an integrated analytical framework embedded with Conditional Value at Risk (CVaR) for quantitative comparative verification. This work aims to elucidate the risk hedging effect of dual-timescale coordination, and provide methodological reference for flexible resource allocation in high-renewable power systems. Existing studies have made substantial advances in single-timescale risk measurement and stochastic planning, and the hybrid-resolution modeling scheme that couples hourly operational simulation with multi-scenario monthly energy balance has been widely acknowledged as an effective approach to strike a balance between simulation fidelity and computational tractability. As a coherent risk measure, CVaR has been extensively applied in short-term dispatch and long-term planning respectively, but systematic research on its nested deployment across dual timescales within a unified optimization framework remains absent. This paper embeds CVaR into both hourly operational constraints and monthly energy balance constraints, realizing the coordinated quantification of short-term load-shedding risk and long-term energy shortage risk. Comparative case studies on the Garver-6 and Modified 14-bus TEP benchmark system validate that the proposed dual-timescale risk planning scheme can effectively optimize the trade-off between system economy and tail risk, significantly improve renewable energy consumption rate, and enhance power supply reliability, with a slight and reasonable increase in levelized cost of electricity.

Accurate assessment of source-load complementarity and system regulation capacity is critical for secure dispatch and planning in high-penetration renewable power systems. Addressing limitations of existing methods—which rely heavily on static metrics, struggle to capture temporal and tail dependence characteristics, and provide insufficient support for dispatch decisions—this paper proposes a multi-level integrated evaluation framework. First, from a source—load matching perspective, we develop a novel complementarity metric, integrating real-time rate of change, temporal consistency, and tail dependency. An improved adaptive noise-complete set empirical mode decomposition combined with a hybrid Copula model is employed to isolate noise and to precisely quantify dynamic dependency structures. Second, we introduce the Minkowski measure and construct a net load fluctuation domain accounting for extreme fluctuations and coupling relationships. Subsequently, combining the Analytic Hierarchy Process (AHP) with probabilistic convolution enables multi-level comparative quantification of resource capacity and fluctuation domain requirements under varying confidence levels. Simulation results demonstrate that the proposed framework not only provides a more robust assessment of source-load complementarity but also quantitatively outputs the adequacy and risk level of system regulation capacity. This delivers hierarchical, actionable decision support for dispatch planning, significantly enhancing the engineering applicability of evaluation outcomes.

The increasing integration of renewable energy and the dynamic nature of modern power grids pose significant challenges to grid stability, reliability, and efficiency. To address these issues, this study proposes an Attention-Enhanced Deep Graph Learning-Based Optimization Model for power grid topology optimization. Leveraging graph neural networks (GNNs) enhanced with multi-head attention mechanisms, the model prioritizes critical nodes and edges, enabling efficient optimization while adhering to physical constraints. Key innovations include graph representation techniques for power grids, attention-based mechanisms for adaptive learning, and a custom loss function to balance optimization objectives. Furthermore, a digital twin framework integrates BIM-ROS for real-time data synchronization and Kalman filter-based sensor fusion to enhance monitoring accuracy and dynamic response. Case studies on the IEEE 118-bus system validate the model’s superiority, achieving 21.4% faster convergence and 14.9% energy loss reduction compared to Particle Swarm Optimization (PSO) and Genetic Algorithms (GA). These results demonstrate the model's potential to enhance grid efficiency and resilience, providing scalable solutions for modern power systems.

This study developed a corporate electricity load forecasting model that integrates the STIRPAT and ARMA models, assessing the impact of various economic, technological, and policy factors on load. The use of time series forecasting improved the accuracy of medium- and long-term predictions. Empirical results indicate that the model effectively captures dynamic changes in load, providing a scientific basis for corporate electricity planning. Future research may incorporate mor e advanced machine learning methods to further enhance the model's predictive power and applicability.

Accurate probabilistic forecasting of the multi-energy loads can provide essential uncertainty information about future loads for the management of integrated energy systems. The selection of appropriate features lays a critical foundation to achieve accurate forecasting, but such an issue is not thoroughly studied for probabilistic load forecasting, especially for multi-energy loads. In this article, we propose an adaptive feature selection framework for probabilistic multi-energy load forecasting by considering different operation patterns to select pattern-specific features. Specifically, we develop a ProbLassoNet model by integrating the multi-quantile regression model with the residual-connecting-Lasso operation to capture both linearity and nonlinearity for effective feature selection. We conduct experiments on an open dataset and validate that the proposed method can significantly improve probabilistic multi-energy load forecasting by distinguishing important features from redundant features. We also provide a comprehensive analysis of important features and multi-energy relationships in different periods, which can serve as a reference for further research on multi-energy load forecasting.

Distributed energy resources provide local power as a supplement on the customer side. Recent rapid development of the distributed energy source enhances the clear energy production at the terminal of the power system. Whereas the small capacity of a single distributed energy source and the scaling of numbers pose difficulties for market design and clearance. In addition, the stochastic and quickly varying output power of the amount of (distributed) renewable energy sources increases the necessity for flexible regulation capacities. In response to the above issues, this paper develops a modified Leiden algorithm to aggregate distributed energy sources with similar regulation properties and connectives, avoiding complex power allocation strategies within the intra-aggregator and ensuring ordered power flow among inter-aggregators. Then, a bi-level market mechanism is proposed to highlight the regulation contributions of both distributed aggregators and conventional energy sources. The upper-level model optimizes the price of combined frequency regulation and electricity markets. The lower-level model regulates the output power of the aggregators and conventional energy sources. Furthermore, the modification of the bi-level model is proposed via the Karush–Kuhn–Tucker condition to ensure its solvability. The proposed market mechanism and the aggregating method are verified using a modified IEEE 30-bus system with IEEE 123-node test feeders and terminal-side energy resources. The results reflect the incentive impacts of the designed market mechanism and the effectiveness of the aggregating algorithm.

To overcome the problems of low accuracy in capacity estimation, low balancing degree and low utilisation rate in traditional methods, a capacity configuration method for new energy storage system based on segmented peak shaving is proposed. The battery's internal resistance and terminal voltage signals of the new energy storage system are taken as inputs, and the capacity estimation is the output. A capacity estimation model based on an improved fuzzy neural network is established. The capacity configuration objective function is constructed by combining segmented peak shaving and economic cost. Hybrid frog-leaping algorithm is used to obtain the optimal parameters for segmented peak shaving and economic cost through population initialisation, position updates and frog swarm sorting to determine the optimal configuration scheme. Experimental results show that the average accuracy of capacity estimation using this method is 97.31%, the maximum balancing degree is 0.98 and the minimum utilisation rate is only 90.9%.

The widespread utilization of renewable energy sources, such as wind and solar energy, plays a crucial role in achieving the dual-carbon goal. However, the uncertainty associated with these energy can impact the stable operation of the power system and the integration of renewable energy. Therefore, this paper focuses on investigating the potential applications of wind-photovoltaic-pumped storage system. A multi-objective optimization model is developed to consider both the maximization of economic benefits and the minimization of system power fluctuations. The model considers the actual output of wind and solar energy as random variables and uses the scenario method to describe their uncertainty. To solve this model, the paper employs a normalization process to transform the multi-objective problem into a single-objective problem. Furthermore, Conditional Value-at-Risk (CVaR) is utilized to convert complex chance constraints into a set of inequalities for solution. Case studies demonstrate that the proposed model effectively enhances the economic benefits of the system and significantly reduces the volatility of power grid connection.

The integration of electric vehicles (EVs) into power systems is crucial for sustainability but poses prediction challenges for dispatchable EV capacity, essential for grid stability. This paper presents a novel capacity prediction model for power systems using an improved long short-term memory (LSTM) algorithm. Our enhanced LSTM model addresses the limitations of traditional models by optimizing architecture and integrating power system-specific features. Through extensive data analysis and model adjustments, our approach outshines existing methods in accuracy and reliability, showcasing significant improvements in EV capacity forecasting. The simulation results verifies the effectiveness of the proposed method.

With the widespread adoption of electric vehicles (EVs) and the enhancement of charging infrastructure, EVs have become a significant demand-side resource with notable regulation potential. However, current studies primarily focus on the capability of EVs to participate in demand response (DR), ignoring the fact that the configuration of charging stations (CS) and distributed energy resources (DER) can considerably enhance the engagement rate of EV users in DR. This work proposes an integrated configuration strategy for distributed generations (DG) and EVCSs, taking into account the spatial scheduling capacity of the EV charging load. A mixed liner integer model is established to achieve the optimized configuration. Based on the IEEE 33-bus distribution network, a simulation analysis is carried out. The integrated configuration of DG and EVCSs under various conditions is analyzed. Simulation results demonstrate the effectiveness of the proposed model.

The power system has experienced a gradual increase in renewable energy, such as wind farms (WF). To address their variability and unpredictability, a large number of energy storage units have been installed accompanying with them. Previous research primarily focused on individual energy storage system (ESS) or combined ESS and WF. However, in practices, the growing number of energy storage units pose significant challenges for power system both in control and operation simulation. To simplify the simulation, this paper proposes a aggregation model of multiple energy storage units at the WF side based on effect approximation method. The parameters of the aggregation model is optimized with Genetic Algorithms (GA), and the main idea of the method is to reduce their quantity while maintaining the overall regulation effect. Simulation is conducted and results indicate that the proposed method outperforms the conventional aggregation method based on accumulation.

To improve the stability requirements of power systems with high renewable energy penetration, more and more converters adopt Virtual synchronous generator (VSG) control strategies. Especially, the independent islanded system with distributed power sources and large line impedance is prone to voltage quality and stability problems, which put higher requirements on the dynamic performance of converters. Therefore, this paper constructs a dynamic mathematical model of VSG in an isolated system considering the control dynamic of VSG and the interaction with the line impedance. The proposed model provides an analytical model basis for the dynamic performance improvement of VSG. The correctness of the proposed dynamic model of VSG is verified by dynamic simulation in MATLAB/Simulink.

The rapid growth of the number of electric vehicles (EV) has promoted the gradual improvement of EV charging facilities in the urban distribution network. However, the diverse charging characteristics and travel preferences of EVs will bring challenges to the optimal configuration of EV charging stations. Efficiently coordinating the charging speed of the charging station with the specific charging requirements of various EV users will emerge as a pressing issue that has to be addressed. This paper proposes an optimal configuration method for EV charging stations considering multi-type EV travel and charging characteristics. An optimal configuration model of EV charging station considering various EV charging characteristics is established. Based on the IEEE-33 bus system, the simulation analysis is carried out. Simulation results show the effectiveness of the proposed model.

Hydrogen direct reduction iron coupled with electronic arc furnace (H2DRI-EAF) technology, as an important technology for decarbonisation in the iron and steel industry, has the advantages of high electrification and low carbon emissions. However, the large demand for hydrogen in this technology relies significantly on the production of electrolytic hydrogen, leading to a substantial increase in power consumption in the steel production process. Moreover, the use of an unclean power source in electrolytic hydrogen production leads to increases in indirect carbon emissions, reducing the low-carbon attributes of the technology. This study investigates the integrated flexible operation mode of a steel plant. An illustrating method is utilised for modelling the entire steel production process and power to hydrogen (PtH2) process in detail for the H2DRI-EAF steel plant, which includes natural gas, photovoltaic, wind power self-provided power plants, and carbon capture and storage (CCS) systems. A mixed integer linear programming (MILP) model is developed for the comprehensive scheduling of the steel mill. The results of the case studies indicate that by reliably integrating the production of renewable energy and natural gas power plants, the PtH2 system can fully consume the renewable energy output while ensuring the smooth progress of steel production and maximising the reduction of carbon emissions from hydrogen production and the total cost of steel production.