[Objective] To address the issues of high carbon emission intensity in thermal power units, insufficient seasonal constraints of the current annual carbon trading mechanism, and the lack of synergistic carbon reduction mechanisms for oxy-combustion and biomass co-firing carbon capture power plants (BCCPP), this paper proposes an optimal scheduling method for park-level integrated energy systems (PIES) incorporating oxy-combustion and seasonal carbon trading. [Method] Firstly, the oxygen-enriched combustion technology is introduced into the biomass co-combustion carbon capture power plant, and the BCCPP model of oxygen-enriched combustion is constructed to realize the technical coordination of oxygen-enriched combustion and biomass co-combustion at the carbon capture level, and a PIES operation framework model including “energy layer-carbon management layer-scheduling layer” is established. Secondly, in order to break through the limitation of single index allocation of traditional baseline methods and solve the fairness problem of quota allocation among multi-carbon emission entities in the park, a multi-index carbon quota allocation model considering grey relational analysis (GRA) and technique for order preference by similarity to ideal solution (TOPSIS) is proposed to stimulate the low-carbon economic operation of multi-carbon emission entities in PIES. Finally, in order to adapt to the seasonal characteristics of PIES, the step-by-step carbon trading is reclassified according to carbon quotas and actual carbon emissions, and the reward and punishment factor is introduced on the basis of the step-by-step carbon price to form a seasonal carbon trading mechanism considering the reward and punishment mechanism suitable for the actual operation of the park. [Results] The combined utilization of the oxy-combustion BCCPP reduces the energy penalty for carbon capture by 36.31% and cuts the carbon emissions of the generating units by 56.15%. With the integration of the GRA-TOPSIS-based multi-index carbon quota allocation and the seasonal carbon trading mechanism, the annual carbon emissions of the PIES are lowered by 26,577.68 tons, while the operational cost decreases by approximately 5.09×10⁶ CNY, representing a reduction of 0.75%. [Conclusion] Case verification confirms that the seasonal carbon trading mechanism, via differentiated seasonal quota allocation, realizes cross-seasonal quota optimization and enhanced PIES economic-low-carbon coordination.

[Objective] Affected by meteorological variations and the non-stationary characteristics of wind power generation,wind power sequences exhibit strong stochastic fluctuations,making it difficult for traditional prediction models to achieve effective modeling. To improve the accuracy of short-term wind power forecasting,this paper proposes a forecasting method based on a dual-branch KAN-enhanced model and cross-modal deep reinforcement learning. [Methods] First,meteorological features are selected using the Pearson correlation coefficient. Subsequently,an improved beaver behavior optimization (IBBO) algorithm is employed to optimize the parameters of variational mode decomposition (VMD),thereby achieving multi-scale decomposition of wind power sequences. Furthermore,a dual-branch prediction framework is constructed,where the KAN-enhanced Transformer model is utilized to extract global temporal dependencies,while the TCN-KAN model is designed to capture local dynamic patterns. On this basis,a cross-modal attention-driven proximal policy optimization (PPO)-based dynamic fusion module is introduced to adaptively integrate the prediction results from multiple models,ultimately obtaining accurate wind power forecasting results. [Results] Experiments are conducted using the operational data collected from a 99 MW wind farm throughout the year 2020,and comparisons are made with several baseline forecasting models. The results demonstrate that the proposed method achieves superior forecasting accuracy on the independent test dataset covering the entire year. [Conclusions] The proposed method effectively improves short-term wind power forecasting performance and exhibits good predictive stability under year-round operational conditions.

[Objective] In response to the Stackelberg game framework involving multiple virtual power plants (VPPs) and a distribution system operator (DSO), where existing dynamic pricing models inadequately capture the impact of renewable energy generation uncertainty on game strategies, and conventional solution methods suffer from poor solution quality in high-dimensional decision spaces, this paper proposes a dynamic pricing model and solution approach based on distributionally robust optimization (DRO) within a Stackelberg game framework. [Methods] A Stackelberg game dynamic pricing model is established, with the DSO as the leader and multiple VPPs as followers. In the VPP optimization model, a distributionally robust chance constraint is introduced to characterize wind power output uncertainty. By constructing an ambiguity set using only the first and second moments of wind power, the chance constraint is transformed into a deterministic equivalent form via the Cantelli inequality. Furthermore, a neural network-enhanced regional particle swarm optimization (NN-RPSO) algorithm is proposed, which builds a predictive model for VPP behavior to replace frequent calls to the lower-level optimization and integrates a regional partitioning strategy to achieve efficient search. [Results] Decision-makers can balance operational risk and cost by adjusting the risk budget parameters. Compared with the regional particle swarm optimization (RPSO) algorithm, the proposed algorithm increases the DSO's revenue by 47.2% while reducing the number of lower-level model calls by approximately 67.4%. [Conclusions] The proposed DRO model provides decision-makers with a flexible approach to handle uncertainty, and the proposed solution algorithm significantly improves solution quality, offering an effective solution for the multi-agent Stackelberg game problem.

[Objective] This paper aims to fully harness the regulation potential of electric vehicles (EVs) within multi-market environments. The study proposes an optimal scheduling model that balances the interests of multiple stakeholders to resolve the critical issues of information asymmetry and scheduling deviations between aggregators and users. [Methods] A three-layer "electricity market-aggregator-user" Stackelberg game model was established. In this framework, the aggregator acted as the leader to coordinate revenues from the energy market, frequency regulation market, and green electricity certificate transactions. To handle price uncertainties, a robust optimization approach was employed. Furthermore, an incentive-compatible mechanism based on the Vickrey-Clarke-Groves (VCG) theory was designed to eliminate the motivation for users to misreport their private information. To enhance model accuracy, a non-linear battery degradation model was introduced, and the game equilibrium was solved using backward induction combined with professional linearization techniques. [Results] Simulation results demonstrate that the proposed mechanism reduces the system peak load by 17.4% and lowers the comprehensive costs for users by 16.27%. The analysis verifies that truthful reporting of private information is the dominant strategy for users under the VCG-based incentive mechanism. Notably, the aggregator can reduce physical execution deviations by over 80% by conceding less than 10% of its potential profits, indicating a high efficiency in risk mitigation. [Conclusions] The study concludes that the VCG mechanism effectively achieves risk isolation on the user side, ensuring fair participation. The coupling of multiple markets and the application of robust optimization significantly bolster the aggregator’s resilience against market volatility. Moreover, the findings clarify that a battery cost below 1000 CNY/kWh serves as the economic tipping point for the large-scale deployment of vehicle-to-grid (V2G) applications.

【Objective】 Aiming at the problems that traditional power system transient stability assessment methods are difficult to handle high-dimensional nonlinear dynamic behaviors and have insufficient efficiency and generalization ability, this paper improves the accuracy and topological adaptability of large-scale power grid transient stability assessment to provide a reliable decision-making basis for power grid emergency control and security defense. 【Methods】 An intelligent transient stability assessment method based on Graph Transformer is proposed, which constructs node input features by fusing electrical quantities and fault information at the initial moment of short-circuit faults. It uses graph neural networks to model grid topology, combines Laplacian position encoding to inject structural information, and adopts Transformer multi-head self-attention to capture global complex dependencies among nodes. Multi-topology and multi-fault samples are generated through the PSASP platform, and the model is built with max-pooling for feature aggregation and verified on the IEEE 39-bus system and a regional 1947-bus system. 【Results】 Feature ablation experiments on the IEEE 39-bus system show that fault duration is the most important factor affecting transient stability assessment. On the 1947-bus validation set, the model achieves 98.33% accuracy and 97.36% recall, outperforming traditional machine learning and mainstream deep learning models. Under grid topology changes, its performance indicators fluctuate by less than 0.25% and maintain stable and excellent results. 【Conclusions】 The proposed method integrates the advantages of graph neural network topology modeling and Transformer global feature capture, with outstanding performance in high-dimensional data processing, complex disturbance adaptation and topological generalization, high accuracy and strong robustness, which can provide reliable decision support for power system emergency control and security defense.

【Objective】Aiming at the problems that traditional power system transient stability assessment methods are difficult to handle high-dimensional nonlinear dynamic behaviors and have insufficient efficiency and generalization ability, this paper improves the accuracy and topological adaptability of large-scale power grid transient stability assessment to provide a reliable decision-making basis for power grid emergency control and security defense. 【Methods】An intelligent transient stability assessment method based on Graph Transformer is proposed, which constructs node input features by fusing electrical quantities and fault information at the initial moment of short-circuit faults. It uses graph neural networks to model grid topology, combines Laplacian position encoding to inject structural information, and adopts Transformer multi-head self-attention to capture global complex dependencies among nodes. Multi-topology and multi-fault samples are generated through the PSASP platform, and the model is built with max-pooling for feature aggregation and verified on the IEEE 39-bus system and a regional 1947-bus system. 【Results】Feature ablation experiments on the IEEE 39-bus system show that fault duration is the most important factor affecting transient stability assessment. On the 1947-bus validation set, the model achieves 98.33% accuracy and 97.36% recall, outperforming traditional machine learning and mainstream deep learning models. Under grid topology changes, its performance indicators fluctuate by less than 0.25% and maintain stable and excellent results. 【Conclusions】The proposed method integrates the advantages of graph neural network topology modeling and Transformer global feature capture, with outstanding performance in high-dimensional data processing, complex disturbance adaptation and topological generalization, high accuracy and strong robustness, which can provide reliable decision support for power system emergency control and security defense.

[Objective] In order to reduce total cost of energy consumption of commercials, meet the responsibility weight of renewable energy power consumption, improve the utilization rate of new energy, and alleviate reverse overload of distribution transformers, a configuration strategy of self-owned power sources including wind, solar and energy storage under time-of-use price curve, load guideline and load baseline demand response (DR) is constructed. The constraint of allowing reverse transmission is considered. The effectiveness of participating in three types of DR to reduce total cost of energy consumption is evaluated and compared. [Methods] Taking the minimum total cost of energy consumption as the target, a two-layer optimization model is established, including alternating iterations of self-owned power source configuration during the planning phase and load baseline during the operation phase. The upper model considers the maximum of multiple predicted new energy curves at any time, establishes a self-owned power source configuration model, and solves the capacity of new energy, energy storage, and nominal discharge time. The lower model calculates the mathematical expectation of total cost of energy consumption based on the outcome of the upper model and the probability of multiple predicted new energy curves. The mathematical expectation of transmission curves is used as the load baseline and fed back to the upper model until the results of two consecutive self-owned power source configuration and the load baseline are identical. Via setting the transmission constraint of allowing reverse transmission of not exceeding 20% of the generated electricity of photovoltaic, and adjusting time-of-use price, benchmark incentive price and load guideline, the influence on economic indicators is studied. Taking a certain large commercial complex in a coastal province of China as an example for calculation and evaluation, the effectiveness of the strategy is verified from the perspective of reducing total cost of energy consumption. [Results] From the perspective of mathematical expectation, compared with the control schemes without self-owned power sources, total costs of energy consumption utilizing the proposed model are reduced by 2% to 10%. The similarities when participating in baseline DR are 98% to 100%, which are significantly higher than 82% to 86% when participating in guideline DR. Compared to guideline DR, participating in baseline DR can reduce total cost of energy consumption by 6% to 10%. The proportions of reverse transmission of photovoltaic are 0% to 10%. [Conclusions] The proposed strategy in this paper satisfies the responsibility weight of renewable energy power consumption of provincial power grids, while achieving high similarity between the load baseline and transmission curve. The set constraint of allowing reverse transmission increases the scale of self-owned power sources, significantly reducing total cost of energy consumption of commercials.

[Objective] In response to the increasingly diversified charging demands arising from the rapid development of electric vehicle (EV), this paper investigates a planning method for charging stations based on the collaboration of multiple types of charging posts. [Methods] From the perspectives of EVs, transportation networks, and power grids, a siting and sizing planning method for charging stations under vehicle-road-grid coupling is first established based on graph theory. The charging behavior characteristics of EV users are then explicitly modeled, and four types of charging posts—Slow Charging Post (SCP), Fast Charging Post (FCP), Mobile Charging Post (MCP), and Ultra-fast Charging Post (UCP)—are selected as the main facilities. A planning model is constructed with the objective of minimizing the annualized total social cost, incorporating constraints from multiple scenario conditions and multiple charging post types. The planning problem is then reformulated as a Mixed-Integer Second-Order Cone Programming (MISOCP) problem via scenario transformation and second-order cone relaxation techniques, and solved using the Gurobi optimizer. [Results] Simulation results demonstrate the high efficiency and effectiveness of the proposed model. The results indicate that the planning solution considering SCP, FCP, UCP, and MCP is optimal. Notably, the integration of MCPs provides effective emergency response during peak charging demand periods and reduces the overall planning cost by 17.82%. [Conclusions] In the proposed planning model, EV users can select among multiple types of charging posts based on specific principles. The coordinated configuration of diverse charging posts offers greater flexibility compared to single-type configurations, enabling the satisfaction of charging demands while reducing the annualized total social cost.

[Objective] Under extreme ice disaster weather conditions,the power generation capacity of wind and solar energy experiences a drastic decline,while the probability of power equipment failures increases significantly,leading to severe supply-demand imbalances persisting for multiple days. In order to analyze the complex interdependent relationships among source-grid-load components in high-penetration renewable energy systems during ice storms and identify vulnerable nodes prone to failure propagation,this paper investigates the temporal-spatial distribution patterns,correlations,and stochastic characteristics of source-grid-load systems under ice disaster conditions. The study reveals probabilistic features of both power supply and demand sides,establishing a probabilistic model for source-grid-load systems that incorporates key ice disaster impact factors. [Methods] Based on nonlinear dynamic system theory,we propose three novel indicators：passive/active voltage fluctuation indices for grid nodes,a composite network fluctuation index,and a node voltage violation index. A dynamic impedance matrix and source-load frequency coupling coefficient are introduced to develop an analytical framework for assessing source-grid-load coupling characteristics and vulnerability under ice disaster scenarios. This methodology enables systematic investigation of the synergistic evolution mechanisms and variation patterns within source-grid-load systems during ice events. [Results] Case studies using the modified IEEE 39-node system and a regional grid demonstrate that system-wide passive/active fluctuation indices progressively increase with enhanced nodal interdependencies. The amplified coupling effects from renewable energy fluctuations under ice storms substantially degrade grid resilience. [Conclusions] The proposed approach effectively reveals dynamic evolution patterns of source-grid-load coupling characteristics and identifies critical vulnerable nodes,providing an effective tool for in-depth research on operational characteristics and optimal dispatch strategies for power systems under ice disaster conditions.

With the drive of the “dual carbon” goal and the deepening of the construction of new energy system, renewable sources and novel loads, such as distributed power supply, electric vehicles and controllable user-side resources, have developed rapidly and their proportion has reached new heights. The fluctuation and randomness of these new sources and loads pose new challenges to the safe operation and flexible regulation of the distribution networks, and it is urgent to upgrade the distribution network towards modernity and intelligence. In response to the requirements of the development of modern smart distribution networks, the connotation and characteristics of modern smart distribution networks are analyzed, and the intelligent demands and development priorities of distribution networks are elaborated. In view of the diversity and difference of distribution network construction, the key technologies for upgrading traditional distribution network to modernity and intelligence are preliminarily explored by combining five typical scenarios: coordinated development of micro-grid, efficient carrying capacity of charging facilities, efficient utilization of new energy storage, upgrading of urban and rural distribution network, and efficient coordination of generation, network, load and storage. Finally, based on the connotation, characteristics, and development priorities of modern smart distribution networks, the technical development direction and construction priorities of future distribution networks are prospected.

In the context of "dual carbon", in order to further reduce the carbon emissions of the microgrid system,a low-carbon operation strategy of multi-micro-grid integrated energy system is proposed, which takes into account the coordinated charging of electric vehicles and the reward and punishment type of ladder carbon trading mechanism. Firstly, a dynamic urgency index reflecting the user's charging demand is proposed for the EV load in the microgrid, and a coordinated EV charging model is established to minimize the pek-valley difference of the total electric load in the microgrid. Then, a two-stage power-to-gas, distributed power supply and Cogeneration unit supply cooperative operation model is added to the microgrid. Finally, considering the existence of electrothermal interaction in the multi-microgrid, the carbon emission of the multi-microgrid is restricted by adding the reward and punishment type of ladder carbon trading mechanism, and the low-carbon and economic performance of the multi-microgrid system is further improved. Based on this, a low-carbon economy operation goal with the minimum cost of multi-micro online shopping energy, carbon emission and wind and light abandonment is constructed, and the original problem is transformed into a mixed integer linear programming problem for solving. By comparing cases, the effectiveness of the proposed strategy is analyzed and verified.