最新刊期
卷 50 ， 期 16 ， 2026
- 摘要：New power systems are evolving into complex giant systems, characterized by strong uncertainty, weak disturbance resilience, and intertwined economic objectives of multiple stakeholders. Rapid and accurate construction of transient stability boundaries is essential for efficient dispatch and real-time control. Based on the practical requirements of dispatch departments, this paper first proposes a general research framework for transient stability boundary characterization and calculation. It then reviews the concept and computation methods of dynamic security regions, identification of critical and correlated sections, assessment of decision-variable impacts on transient stability boundaries, and calculation methods for transmission section power limits. The advantages and disadvantages of various models and methods, along with their applications in bulk power systems, are summarized. Finally, considering the new characteristics of transient stability boundaries in new power systems, the paper thoroughly analyzes the challenges in their characterization and calculation, and outlines future research directions.关键词：new power system;transient stability;security region;transmission section;dispatch;operation7|1|0更新时间：2026-08-21
Review·Perspective
- 摘要：High proportion of renewable energy and low-carbon electrification of end-use energy have become critical features in the evolution of the supply-demand dynamics in the future new power systems. However, the seasonal mismatch contradiction between clean heating demand and renewable energy output has become increasingly severe. This paper focuses on achieving long-term thermal storage through seasonal thermal storage technology to facilitate large-scale accommodation of renewable energy and cross-seasonal utilization of renewable thermal energy. First, a refined seasonal thermal storage model for the optimal planning is established to describe the time-varying heat-loss characteristics of thermal storage tanks across seasons. On this basis, the established seasonal thermal storage model is embedded into the power planning problem to achieve optimal configuration of seasonal thermal storage within the system. Finally, a modified HRP-38-bus test system is used to analyze and validate the effectiveness of the proposed modeling method, and the role of seasonal thermal storage in enhancing renewable energy accommodation and renewable heating is discussed.关键词：new power system;renewable energy;seasonal thermal storage;heating;electricity supply-demand balance;optimal planning;refined model3|1|0更新时间：2026-08-21
- 摘要：Regarding the frequency security issues caused by lack of conventional power generation support in the remote renewable energy bases, this paper proposes an electricity-hydrogen energy storage planning model considering virtual inertia response and frequency regulation parameter optimization. It leverages the complementary advantages of fast inertia response of electric energy storage and long-term large-scale energy time-shifting of hydrogen energy storage. At the planning level, a grid-forming electricity-hydrogen energy storage allocation model in DC transmission scenarios is established. By analyzing the virtual inertia characteristics of hydrogen energy storage systems and electric energy storage systems, a multi-device collaborative inertia support model is formulated. At the operation level, for three safety indicators—maximum frequency deviation, rate of frequency change (RoCoF), and quasi-steady-state deviation—the virtual inertia time constant and primary frequency regulation coefficient of grid-forming control are dynamically optimized to maximize the frequency support capability of energy storage systems in various operation scenarios. Simulation studies based on a renewable energy base in Northwest China demonstrate that the proposed model improves frequency security indicators and reduces energy storage allocation costs through coordinated frequency regulation and adaptive parameter adjustment of electricity-hydrogen energy storage.关键词：renewable energy base;grid-forming control;electricity-hydrogen energy storage;frequency security;virtual inertia5|1|0更新时间：2026-08-21
- 摘要：With the large-scale integration of distributed photovoltaics (DPVs), distribution networks face increasing uncertainty and challenges in renewable energy accommodation. Rational planning of load-storage resources is crucial for enhancing system renewable energy accommodation capacity and flexibility. Therefore, a coordinated load-storage planning method considering uncertainty of DPV coupling is proposed. Firstly, the Copula function is employed to characterize the spatio-temporal coupling characteristics of DPV output, and a Copula-Wasserstein distance-based ambiguity set construction method is proposed to describe the coupling uncertainties among multiple DPVs. Then, considering the complementary characteristics between physical energy storage and demand response, a load-storage coordinated distributionally robust optimization planning model considering uncertainties of DPV coupling is formulated, and a solution method combining McCormick relaxation and strong duality theory is proposed. Finally, two actual 10 kV distribution networks in a certain area of China are used for cases. The results demonstrate that the ambiguity set construction method based on Copula-Wasserstein distance can effectively exclude probability distributions that violate coupling characteristics between DPVs, significantly reducing system planning and operation costs. Moreover, the coordinated load-storage planning can effectively solve problems of local DPV accommodation and regional heavy overload.关键词：distributed photovoltaic (DPV);distribution network;uncertainty;Copula-Wasserstein distance;coupling;energy storage;load-storage coordination;distributionally robust optimization planning1|2|0更新时间：2026-08-21
- 摘要：In the context of carbon emission peak and carbon neutrality as the core areas of energy production and consumption, cities are facing multiple challenges such as municipal solid waste treatment and local consumption of renewable energy. Therefore, a low-carbon scheduling method for virtual power plant (VPP), which considers the urban biomass resource recovery and the coordinated control of photovoltaic-storage-charging systems, is proposed. Firstly, by taking “carbon treatment” as the bridge, a coordinated treatment path for dry and wet biomass waste and a multi-source energy supply structure are constructed, and an energy flow model of the biomass power plant is established. Secondly, a coordinated vehicle-to-grid (V2G) charging and discharging strategy for photovoltaic-storage-charging stations is designed based on user preferences. Finally, a low-carbon optimization scheduling model for the VPP integrating both supply-side and demand-side strategies is constructed. The operating effect and mechanism effectiveness of the proposed method are verified through multi-scenario case studies.关键词：virtual power plant;biomass energy;photovoltaic-storage-charging station;low-carbon scheduling;vehicle-to-grid (V2G);renewable energy0|1|0更新时间：2026-08-21
- 摘要：Electric transfer vehicles (ETVs), which serve as critical logistics equipment connecting quay areas and yard areas, pose severe challenges to port microgrids in cost reduction and energy consumption minimization due to their uncoordinated charging. To address these challenges, this paper explores the charging flexibility of ETVs and proposes a charging scheduling method for ETVs considering conflict-free path constraints and battery health constraints, with the objective of minimizing task delay penalties and charging costs. Meanwhile, to mitigate battery life degradation caused by frequent and deep charging-discharging cycles, a linear battery healthy operating region is constructed via hyperplane projection and convex approximation techniques, which is compatible with the proposed port ETV charging model without introducing extra computational complexity. Furthermore, to handle the combinatorial decision-making complexity arising from the strong coupling among path, charging, and energy constraints, the port ETV scheduling process is formulated as a multi-agent Markov decision process under a rolling-horizon framework. A Q-learning-based multi-agent reinforcement learning framework is then employed to realize collaborative optimization between vehicle-level decision-making and system-level optimization, thereby effectively reducing the solution scale and improving overall scheduling efficiency. Finally, numerical analyses based on a practical port are carried out to verify the effectiveness and practicability of the proposed method.关键词：microgrid;electric transfer vehicle (ETV);path constraint;reinforcement learning;battery health;multi-agent;charging cost;Markov decision process0|1|0更新时间：2026-08-21
- 摘要：In the context of extreme disasters, enhancing the resilience of power-transportation coupling networks has become a key concern for ensuring the functionality of critical urban infrastructures. To account for user behavior in transportation networks, this paper proposes a resilience improvement method for power-transportation coupling system based on a two-stage robust optimization framework that incorporates variational inequalities. First, a tri-level resilience improvement model is formulated, encompassing pre-disaster investment, mid-disaster response, and post-disaster operation, all subjected to user equilibrium constraints. Second, to tackle the challenge that user-equilibrium-based transportation subproblems cannot be directly dualized, a variational inequality approach is introduced to equivalently capture the equilibrium state, thereby establishing the robust optimization-variational inequality-based resilience framework. To effectively solve the resulting problem, it is proven that, under given first-stage decisions, the inner-layer model satisfies strong duality, and a two-level solution strategy integrating column-and-constraint generation with fixed-point projection is designed. Finally, numerical simulation verifies the effectiveness of the proposed model and solution method, demonstrating the necessity of resilience improvement investments to maintain stable operation of power-transportation coupling networks in extreme disaster scenarios.关键词：power-transportation coupling system;extreme disaster;resilience;robust optimization;variational inequality0|1|0更新时间：2026-08-21
- 摘要：As the DC source-load equipment such as distributed photovoltaics and electric vehicles increases, the stability analysis of DC distribution systems becomes more complex, where in the DC-side impedance of a voltage-source converter (VSC) interacts with its AC-side counterpart and may cause system instability. Most existing studies on small-signal stability are only qualitative, and quantitative evaluation of system stability is rarely addressed. Taking droop-controlled DC distribution systems as the research object, this paper establishes the VSC DC-side equivalent impedance and the load-side impedance models, and further reduces the order of the DC-side equivalent impedance model. Based on the impedance-ratio forbidden-region criterion and the phase constraint condition, and with consideration of the gain margin and phase margin, a quantitative evaluation index for system stability is developed. Meanwhile, the stable intervals of the index are depicted to intuitively illustrate the influence of various parameter variations on system stability. Finally, the theoretical analysis is verified through simulation and hardware-in-the-loop experiment.关键词：DC distribution system;small-signal stability;equivalent impedance;short-circuit ratio (SCR);phase constraint;stable interval;quantitative evaluation0|1|0更新时间：2026-08-21
- 摘要：Against the backdrop of the accelerated energy transition, the application of grid-forming capabilities of fuel cells holds great potential. This paper proposes a small-signal stability analysis method based on dynamic theory for the multi-timescale stability problem of grid-forming fuel cell power generation systems. Firstly, a small-signal model of the grid-forming fuel cell power generation system is established. Secondly, by introducing the quantitative indicators of the synchronous component and the damping component, the dynamic model of the system is constructed, revealing the stability evolution law under various operating conditions. Then, by combining the participation factor analysis method with the dynamic characteristics, the internal dynamics of the fuel cell and the dynamic coupling mechanism between the cell-side DC voltage control parameters and the power synchronous control and AC voltage control of the grid-forming converter are deeply analyzed. Finally, the time-domain simulation based on the MATLAB/Simulink platform and the hardware-in-the-loop experiment verification show that the proposed analysis method can accurately predict the stability boundary of the system.关键词：fuel cell;grid-forming;multi-timescale;small-signal stability;dynamic model;coupling0|1|0更新时间：2026-08-21
- 摘要：Grid-connected operation of doubly-fed induction generators (DFIGs) easily triggers wide-band oscillations. In order to adapt to complex grid operating conditions and scenarios involving DFIG parameter deviations, this paper proposes an adaptive impedance reshaping method that combines coordinated rotor-side and grid-side regulation with segmented frequency control. First, an analytical model of the frequency coupling sequence impedanc for DFIGs is built to clarify the dominant factors affecting impedance characteristics over a broad-band frequency range. Second, a virtual impedance controller is designed for the rotor side with gains dynamically adjusted to the frequency, while employing active damping control at the grid side to adapt to parameter variations,which reshapes the impedance characteristics within sub-synchronous and super-synchronous and intermediate frequency bands, respectively. On this basis, the established impedance model is revised, and the discretization design of the control algorithm is carried out. The feasibility of its engineering application is verified. Finally, Simulation case studies validate the practicability of the proposed control method in various operating scenarios, including series compensation and shunt compensation scenarios.关键词：doubly-fed induction generator (DFIG);broad-band oscillation;frequency coupling;impedance reshaping0|1|0更新时间：2026-08-21
- 摘要：As a mainstream topology in power-electronic source-load systems, the three-level neutral point clamped converter inherently exhibits third-harmonic fluctuation in the DC neutral-point potential, which causes pronounced low-order harmonic emission on the AC side and has become one of the major harmonic sources in renewable energy plants. In this paper, the impact of the output third-harmonic component on the neutral-point potential in the three-phase four-wire topology and the multi-frequency coupling relationship are incorporated into the analysis. First, a unified analytical fluctuation-loop model is established for various topologies, carrier-based modulation schemes, and output operating conditions, which enables correlation analysis and characteristic interpretation between neutral-point potential fluctuation and AC low-order harmonic emission. Then, the amplification effect and the amplification ranges of passive parameters on neutral-point potential fluctuation and harmonic-emission distortion are revealed, and improvement measures aimed at suppressing harmonic emission or neutral-point potential fluctuation are proposed based on the analytical model. The proposed suppression methods cover a range of typical operating conditions and are easy to implement in engineering. Experimental results validate the accuracy of the developed analytical model and the effectiveness of the proposed suppression methods.关键词：renewable energy integration;converter;carrier-based modulation;neutral-point potential;zero-sequence injection;harmonic suppression;fluctuation suppression0|1|0更新时间：2026-08-21
- 摘要：The large-scale integration of power electronic devices has led to a significant decline in system inertia, severely deteriorating the frequency dynamic characteristics. The frequency nadir (FN), as a critical boundary parameter in the power grid security defense system, urgently requires the development of a mathematical representation method that balances model accuracy and computational efficiency to accurately characterize it, thereby guiding the optimal operation of low-inertia power systems. To achieve high-precision linearization of non-convex FN constraint, a progressive convex hull approximation (PCHA) method is proposed. First, based on the frequency response model of the multi-machine system and the linearization assumption of the primary frequency regulation process of the unit, the FN constraint in the form of mixed-integer second-order cone programming (MISOCP) is derived through segmented analysis. Then, a PCHA method is proposed to progressively approximate the feasible region of the original FN constraint by constructing a convex hull through vertex search, enabling an equivalent transformation of non-convex constraints. Finally, taking the frequency-constrained unit commitment problem as an example, tests are conducted on the modified IEEE 118-bus and IEEE 1354-bus systems. Results demonstrate that the proposed PCHA method can effectively extract high-precision FN constraint. Compared to directly handling the form of original mixed-integer second-order cone (MISOC), the FN constraint obtained by PCHA transformation exhibits significant computational efficiency advantages in solving the large-scale optimization problems, while strictly ensuring dynamic safety of the system frequency.关键词：frequency nadir (FN);inertia;frequency constraint;frequency safety;unit commitment;convex hull approximation0|1|0更新时间：2026-08-21
- 摘要：The wide-area coordination of multi-energy virtual power plant (MEVPP) clusters relies heavily on open communication networks. Consequently, the cyber-physical systems within these clusters face severe challenges arising from non-ideal communication conditions—specifically latency, packet loss, and noise. These impairments often compromise the convergence of traditional distributed strategies and deteriorate economic performance. To solve these problems, this paper proposes a digital twin (DT) enabled cloud-edge collaborative resilient scheduling strategy for MEVPP clusters. Firstly, a communication topology adaptive reconfiguration mechanism based on Pareto optimality is established to enhance the structural resilience of the system against cyber attacks and network failures. Secondly, a hybrid-driven DT model, integrating a “mechanistic model + residual correction network” is constructed to achieve high-fidelity state prediction under communication constraints. These predictions serve as the initial values for the distributed consensus algorithm, thereby minimizing initial errors and mitigating aggregation bias and iterative oscillations induced by information gaps. Finally, a cloud-edge spatiotemporal decoupling mechanism is designed,in which the tiered carbon price is employed to guide optimal electricity-carbon synergistic interactions among clusters. Case studies demonstrate that the proposed strategy maintains rapid and stable convergence even under severe latency and dynamic topology disturbances. Furthermore, it significantly enhances the system operational resilience while effectively reducing both the total operating costs and carbon emissions of the clusters.关键词：multi-energy;virtual power plant (VPP);resilience;communication network;digital twin (DT);cloud-edge collaboration;cyber-physical system1|0|0更新时间：2026-08-21
- 摘要：Distribution transformer areas have diverse load characteristics and complex influencing factors. Traditional methods have poor interpretability of forecasting results. To this end, this paper proposes an interpretable load forecasting method integrating multi-dimensional data features. First, interpretability analysis is conducted based on forecasting results of deep learning models. Second, combined with similarity experiments and ablation experiments, the effects of dataset scale and seasonality on architecture forecasting performance are verified from an internal perspective. On this basis, a visualization and quantification method for load dispersion is proposed to verify the effect of dataset dispersion on model forecasting performance from an external perspective. Finally, by synthesizing dataset features from both internal and external perspectives, a “dynamic matching” dataset optimization strategy is proposed, which dynamically selects training samples from the database by integrating seasonal and dispersion features to improve sample representativeness. Test results show that the proposed method can capture key data that improve architecture forecasting accuracy, reduce the time cost of data collection, and improve the credibility of forecasting results.关键词：new power system;load forecasting;deep learning;data feature;interpretability;dynamic matching0|1|0更新时间：2026-08-21
Basic Research
- 摘要：Large language model (LLM) has demonstrated strong generalization capabilities in such areas as few-shot learning and multi-source heterogeneous data fusion, offering a new technical approach for the intelligent prediction of heavy-overload of distribution transformers in distribution transformer areas. This paper proposes a heavy-overload prediction model for distribution transformers in distribution transformer areas based on adaptive prompt engineering using LLM. Firstly, the input processing module is used to convert multi-source heterogeneous time-series data, such as load time series and temperature data from distribution transformer areas, into a representation form recognizable by LLM. Secondly, adaptive prompts based on statistical information and dominant pattern recognition are constructed to enhance the modeling capability of LLM for the load time-series information of distribution transformers. Then, a dual-branch structure combining time series and text is used to learn and train time-series and text features separately through a lightweight LLM. Meanwhile, to reduce the distribution differences between the time-series modality of distribution transformer load data and the text modality of the LLM, the model introduces the cross-modal alignment based on adaptive prompts, intermediate feature regularization loss, and output consistency loss, thereby facilitating the effective transfer of the pre-trained knowledge of LLM and the domain-specific knowledge regarding the heavy-overload prediction task of distribution transformers to the time-series modality. Finally, the proposed model is validated based on an actual distribution transformer load dataset from a certain region in China. Experimental results demonstrate that the proposed model exhibits high prediction accuracy in the heavy-overload prediction task of distribution transformers in distribution transformer areas, thereby validating its effectiveness and feasibility in the practical application scenarios.关键词：distribution transformer;heavy-overload;large language model (LLM);data fusion;few-shot learning;prompt engineering0|1|0更新时间：2026-08-21
- 摘要：The construction of flexible and intelligent new distribution systems requires solving the problems of massive, heterogeneous, and uncertain access and control of new grid-connected entities, which puts forward new requirements for communication model transformation paradigms: the uncertainty challenge of protocol access, the semantic challenge of transformation requirements, and the economic challenge of operation and maintenance modes. However, traditional model transformation is limited to one-to-one or one-to-many hardware implementation forms within limited categories, and cannot meet the new requirements for communication model transformation paradigms. Therefore, a large language model (LLM) method based on retrieval-augmented generation is proposed. Firstly, based on the fine-tuned LLM, protocol fields of the communication model are classified into default fields, calculation fields, and mapping fields. Secondly, combining retrieval-augmented generation technology to generate target protocol messages for mapping fields. Finally, simulation verification is conducted using multi-category actual messages from new distribution systems. The simulation results show that the proposed method achieves accurate classification and mapping of protocol fields for multiple types of communication models in the new distribution system, and the proposed model can be deployed in software form on the edge side. Compared with traditional conversion methods, it is more flexible and economic.关键词：new power distribution system;communication model transformation;large language model (LLM);retrieval-augmented generation0|1|0更新时间：2026-08-21
- 摘要：To address the challenges of a surge in simulation workload in new power systems, this paper proposes a novel method for the automated generation of power system simulation models based on large language models (LLMs). The proposed method introduces a dual-layer knowledge base, comprising factual data and syntax rules, to impose multi-dimensional constraints on LLM-generated contents, thereby addressing the reliability issues of generative models in engineering applications. To overcome a key bottleneck of LLMs in processing large-scale power grid information, a task-decomposition based simulation model generation strategy for large-scale power grid is further designed. This strategy leverages the LLMs to autonomously plan and decompose the large-scale power grid modeling task, aiming to surpass token length limits and achieve efficient and high-accuracy automated generation of large-scale power grid models. Through simulation validation on the IEEE 9-bus system, New England 39-bus system, and the Southern Xinjiang main ring grid systems, demonstrate that the proposed method can accurately generate power system simulation models based on natural language prompts.关键词：large language model (LLM);automated modeling;power system simulation;agent;retrieval-augmented generation;artificial intelligence0|1|0更新时间：2026-08-21
- 摘要：To address issues of syntactic ambiguity and the high computational complexity of the full attention mechanism in the existing recognition methods for entities in power regulation text, a recognition method is proposed based on improved dependency forest and adaptive sparse attention mechanism. The method adopts an improved dependency forest algorithm based on dynamic weight adjustment, which utilizes the semantic patterns inherent in power regulation text to solve syntactic ambiguity problems. An adaptive sparse attention mechanism and a global-local-strided three-layer masking strategy are designed to reduce the algorithm complexity. A hierarchical knowledge fusion architecture is further constructed to integrate the improved dependency forest and the adaptive sparse attention mechanism into the LLaMA2 large language model. Experimental results show that compared with the large language models with the same parameter scale, the F1-score of the proposed method is improved by 1.56% to 2.13%. In terms of inference efficiency, compared with the full attention mechanism, the proposed method significantly reduces the number of floating-point operations per second, memory usage, and algorithm complexity. Finally, a power regulation knowledge graph is drawn to verify the practicality of the proposed method.关键词：large language model;power regulation;entity recognition;knowledge graph;dependency forest;sparse attention mechanism0|1|0更新时间：2026-08-21
Application Research
- 摘要：With the increasing proportion of integrated renewable energy making intra-day power balance more difficult, the significant errors in ultra-short-term forecasting remain one of core challenges for dispatch. The hybrid look-ahead dispatch framework driven by reinforcement learning and stochastic optimization improves the adaptability of intra-day dispatch plans to uncertain scenarios through multi-scenario simulation and rolling-horizon optimization, and has been commissioned in an actual provincial power grid. To meet the requirements of engineering application, this paper first elaborates on the decision-making paradigm and technical challenges of look-ahead dispatch. Subsequently, three practical key technologies are proposed: 1) massive power grid operational scenario generation within look-ahead horizon and balance assessment, 2) reinforcement learning-based intelligent multi-scenario simulation, 3) look-ahead decision-making considering both intra-provincial regulation and inter-provincial trading. Then, a sequential “forecast-simulation-decision” software architecture is designed and a model-data interactive-driven look-ahead dispatch technology framework is established. Finally, the test using operation data from Gansu Power Grid in China demonstrates the computational efficiency of the look-ahead dispatch software and analyzes its enhancement effects on power balancing capabilities in challenging scenarios of “ensuring supply security and promoting renewable energy accommodation”.关键词：renewable energy;look-ahead dispatch;reinforcement learning;stochastic optimization;spot market;forecasting;simulation;decision0|1|0更新时间：2026-08-21
Engineering Technology Frontier
- Postal code：310000
- Tel：13632102679
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0