With the increasing penetration of renewable energy sources in microgrids (MGs), the load frequency control system is becoming increasingly vulnerable to cyber attacks due to its interaction with the physical entities of the MG through the communication network. False data injection attacks (FDIAs) pose a significant security threat to MGs as they can disrupt the stability of the MG by injecting false data into control instructions or sensor data channels. This paper investigates the defense and protection problems of the LFC system in MGs under FDIAs. Two types of FDIAs, one affecting the control commands and the other affecting the sensor data, are considered. The sorting algorithms k-nearest neighbor and support vector machine, are utilized to detect the presence and type of FDI. Additionally, a detection method for the sensor-side FDIA is proposed based on the control method used for the control-side FDIA. Furthermore, a comprehensive control method is developed by integrating the detection and control methods, which can effectively handle scenarios where both types of FDIAs appear alternately. Simulation results demonstrate that the proposed method accurately detects and identifies FDI attacks within 0.5 s of occurrence, and the frequency deviation is maintained within  ± 0.02 p.u. under alternating FDI scenarios, effectively eliminating the impact of FDI on microgrid stability.

To address the difficulty of accurately characterizing operating-day regulation capability of aggregated resources, the overestimation of sustainable regulation space caused by period-by-period statistical boundaries, and the challenge for a unified response strategy to balance executability and adaptability, this paper proposes a reinforcement learning-based construction method of segmented regulation response functions for virtual power plants. First, a similar-day screening mechanism is established based on holiday attributes, load level, and curve shape to extract local historical samples close to the operating day. Second, a bilateral adjustable capability feasible region is constructed by combining local historical envelopes of similar days with rolling-window cumulative constraints, and deterministic boundaries of the operating day are derived under regulation demand. Then, baseline load is constructed from historical days of the same type, effective regulation periods are screened by regulation demand deviation signals, one-directional net regulation quantities are generated, and a unified segmented regulation response function is established. Finally, reinforcement learning is introduced to optimize segment boundaries and response parameters so as to improve the matching between the baseline strategy and actual regulation conditions. Case studies show that the proposed method can effectively compress nonsustainable statistical boundaries, form deterministic boundaries that better reflect the actual regulation capability of aggregated resources, and generate segmented response strategies with clear structure and good interpretability. Reinforcement learning can further improve triggering performance and overall regulation effectiveness by optimizing key interval parameters.

Distributed secondary control in inverter-dominated microgrids is challenged by multi-variable coupling among frequency, voltage, and power, as well as heterogeneous feasibility limits of inverter-based resources. Conventional Euclidean consensus schemes typically enforce feasibility via channel-wise saturation or heuristic clipping, which may distort regulation directions and degrade performance under near-boundary operating conditions. This paper proposes a Bloch-sphere-inspired geometric coordination framework that represents a coupled frequency–voltage regulation direction by a bounded spherical state with an auxiliary slack coordinate. A projection-based Laplacian interaction is designed on the sphere to guarantee positive invariance of the coordination state and to promote geometric alignment over a connected communication graph, which supports a consistent planar regulation direction for actively participating nodes. Physical feasibility of inverter commands is ensured at the execution layer by computing a locally feasible step size along the agreed direction, yielding heterogeneous participation induced by inverter-specific constraints. Case studies on an islanded inverter-dominated microgrid demonstrate improved transient regulation, substantially reduced infeasible-command tendency under near-boundary overload, and enhanced robustness under cross-coupled regulation compared with a conventional dual-channel consensus baseline.

Increasing photovoltaic utilization has highlighted the complexities of peer-to-peer energy sharing (P2PESh) in distributed networks. These networks' dynamic characteristic leads to identifying and exploring energy-sharing areas, which fosters the evolution of regional energy trade. The current study aims to address this complexity by proposing a novel energy-sharing framework based on the Stackelberg game paradigm. The energy-sharing acts as a leader, where the dynamic pricing method is applied to the energy-sharing regions (ESRs). The PVs are used to intelligently select energy-sharing zones and optimize the flexible loads. A comprehensive profit maximization model is proposed, in which a multi-zone pricing strategy and network usage fees are incorporated. The enhanced artificial lemming algorithm (EALA) is employed to optimize this problem to ensure the efficient exploration and exploitation of the solution space. A utility model is represented for buyers to inform the energy-sharing zone selection and load configuration. The study establishes the existence and uniqueness of the Stackelberg equilibrium within this framework to ensure stability in energy transactions. The effectiveness of the proposed framework is verified in the real-world system, which exhibits increased profits for energy-sharing providers, improved income for buyers, and enhanced grid stability. The effective management of PVs operation and integration into the grid significantly enhanced the framework's flexibility. This research underscores the potential of integrating P2PESh strategies into distribution networks and offers a sustainable and efficient solution for the growing community of solar subscribers.

In the context of dual-carbon goals and the new power system, virtual power plants (VPPs) enhance low-carbon operation and flexibility by aggregating distributed resources. However, existing research is mostly based on time-of-use (TOU) pricing or a single carbon constraint, neglecting the joint effect of electricity and carbon prices and the coupling between demand response and low-carbon operation. To address this gap, this paper proposes a demand response model with electricity-carbon coupling, where a coupling coefficient embeds carbon cost into the TOU framework. Then, an optimal VPP scheduling model integrating ladder carbon trading and demand response is constructed. The Simulation results show that the mechanism effectively shifts loads to low-carbon periods, improves coordination among energy storage, electric vehicles (EVs), and loads, reduces net carbon emissions, and enhances overall operational efficiency.

With the in-depth promotion of the global energy structure transformation and power demand-side management, steel enterprises, as high-energy-consuming industries, are facing an urgent need to participate in power grid peak shaving. In this paper, aiming at the complexity and multi-link coupling characteristics of the steel production process, a modeling and peak shaving optimization method for the steel production process based on Petri nets is proposed. First of all, a hierarchical Petri net modeling method is used to construct a steel production system model that includes key processes such as sintering, rolling, and oxygen production. The production state and material buffer are represented by places, and the production activities and control events are represented by transitions, clearly depicting the parallel and synchronous relationships among various links. Secondly, a peak shaving auxiliary logic transition is introduced, and a load regulation mechanism for each process during peak and off-peak periods is designed. The flexible switching of the production mode is achieved by controlling the flow of tokens among places. On this basis, an optimization model with the goal of minimizing the electricity cost during peak periods is established, comprehensively considering production plan constraints, coupling constraints among processes, and minimum operation time constraints. This method organically combines the discrete production scheduling problem with the continuous load optimization problem, providing theoretical support and decision-making basis for steel enterprises to participate in power demand response.

This paper presents a blockchain-enabled local trading framework for distributed photovoltaic (PV) systems. A consortium blockchain supports peer-to-peer transactions among PV producers, consumers, and storage operators. A multi-agent Nash bargaining model is developed to optimize pricing and energy allocation, while smart contracts automate bidding, matching, and settlement processes. A dynamic clearing mechanism is introduced to handle generation and demand deviations. Simulation results based on a virtual industrial park show that the proposed system improves local PV consumption, supports fair negotiation, and enhances overall trading efficiency.Copyright (c) 2025 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)

With the rapid growth of distributed power generation, the accuracy of forecasting has become increasingly critical. Traditional centralized learning methods rely on data aggregation from multiple regions for unified training, which presents challenges in privacy protection, system security, and data heterogeneity. This paper proposes a distributed power forecasting method based on federated learning. A bi-layer LSTM network is used as the global model, and the FedAvg algorithm is applied for model parameter aggregation, enabling collaborative training among nodes without sharing raw data. Experiments using real-world data from multiple clients (PV sites) systematically compare centralized, local-only, and federated training modes. Results show that the proposed method outperforms other approaches across various weather conditions in terms of accuracy, robustness, and generalization, while ensuring data privacy, thus demonstrating strong practical and scalable potential.

Plug-in electric vehicles (PEVs) have emerged as pivotal response loads on the demand side in power grid, capable of providing essential services such as frequency regulation, load following, and other ancillary grid support. However, the efficient aggregation and control of multiple PEVs present significant challenges, primarily due to the inherent uncertainties in their availability and energy demands, as well as the need for real-time coordination across a distributed network. This paper addresses these complexities by introducing a novel two-layer interactive time-scale optimization and control structure for demand response load following services. The upper layer employs a stochastic chance-constrained optimization to deliver reliable and optimized power schedules for load utilities. Meanwhile, the lower layer leverages a distributed pinning control algorithm to equitably distribute and track the optimized power schedule among multiple PEV agents, ensuring network-wide stability and efficiency. The proposed framework was validated through simulations involving an twenty-PEV agent network, and the simulation results demonstrating notable improvements in response accuracy and robustness against variability, establishing the effectiveness of the distributed control strategy in real-world demand response scenarios.

This study proposes an innovative decentralized energy management framework for multi-energy microgrids (MEMGs) that integrates combined heat and power (CHP) units, hydrogen storage systems (HSSs), battery electric vehicles (BEVs), and thermal storage to improve operational efficiency and flexibility. The research addresses critical gaps in the existing literature concerning the limitations of centralized control and the inadequate handling of uncertainties in multi-energy systems. By introducing a decentralized multi-agent optimization model, the proposed approach enables local decision-making, reduces communication overhead, and enhances system scalability. A robust stochastic programming technique is employed to capture uncertainties in renewable generation and load demand. Furthermore, the model incorporates a dynamic demand response (DR) strategy that coordinates both thermal and electrical loads. Also, a modified version of the Flow Direction Algorithm (FDA) is proposed to address the problem at hand, offering significant advantages in both local and global search capabilities, thereby enhancing solution quality and optimization performance. Simulation results reveal that the proposed method achieves a 27 % reduction in total operational cost, a 19 % increase in hydrogen utilization efficiency, and a 14 % reduction in peak electricity demand, thereby advancing the state of the art in multi-energy microgrid management.

In the ongoing global energy transformation, the proportion of new energy in the electricity market is continuously increasing. This paper conducts an in-depth exploration of the influence of varying new energy market shares on the operation trends of the electricity market. By integrating evolutionary game theory and system dynamics, a simulation model is established. Based on the actual data of Jiangsu Province, empirical analysis is carried out to reveal the impact mechanisms on key indicators, providing practical suggestions for electricity market regulation and the promotion of energy transition.

This study introduces an innovative energy management strategy for hydrogen-integrated micro combined heat and power (MCHP) microgrids, addressing the urgent need for sustainable energy systems. The proposed model uniquely combines fuel cell units, hydrogen fueling stations (HFS), and multi-energy storage systems to enhance the efficiency of energy generation and consumption. Utilizing a multi-objective optimization approach based on the Salp Swarm Algorithm, the research effectively manages uncertainties in renewable energy generation and fluctuating demand. The framework facilitates dynamic coordination of electrical and thermal energy production while optimizing operational costs, allowing for a seamless transition between energy sources and enhancing system resilience. Additionally, the implementation of demand response programs enables strategic shifting of electricity loads to off-peak periods, significantly improving overall efficiency. Results indicate a substantial reduction in total operational costs of up to 78% compared to traditional methods, alongside a 22% increase in hydrogen utilization efficiency. Furthermore, the demand response strategy successfully reduced peak electricity demand by 12%, contributing to further cost savings. This research highlights the potential of hydrogen technologies in microgrid applications and paves the way for future integration with broader energy markets, advancing the transition toward low-carbon energy solutions.

In this study, the IEEE 14-bus test system is employed to evaluate the proposed energy management strategy for Virtual Energy Hubs (VEHs). The results demonstrate significant cost reductions with the integration of the interactive Energy Market Management (EMM) system. In the baseline scenario, operating costs were reduced by 10.01% when the EMM was introduced, and further reduced by 13.11% with the addition of direct load control programs. The most significant cost reduction of 56.39% was achieved in scenarios incorporating both EMM and ancillary service demand response programs. Additionally, the use of direct load control programs alone resulted in a 6.02% reduction in operating costs, while ancillary service demand response programs contributed an additional 2.29% cost savings. These findings underscore the substantial potential for cost reduction and efficiency improvements through advanced energy management strategies.

The virtual power plant (VPP) integrates electric vehicle (EV) parking lots as both flexible consumers and prosumer, leveraging their bidirectional charging capabilities to improve grid stability and profitability. This paper defines a novel method to the improve economic aspect of distributed energy resources (DERs) in a distribution network through a VPP framework, actively contributing in day-ahead and regulation reserve markets. One of the main novelties of this study is using a forecasted price-based unit commitment approach for VPPs in microgrids with the aim of determining an optimal pricing strategy though addressing real-world operational complexities. Also, this study integrates a dual-role EV parking lots, acting both as a consumer and electricity provider, and explores its potential to minimize costs while optimizing charging and discharging agendas. The proposed optimization model tries to obtain maximum VPP profits in day-ahead and reserve markets by controlling the complexities of distributed thermal and electrical production, energy storage limits, and power balance restraints. By implementing an efficient model based on the a mixed-integer linear programming (MILP), a higher solution speeds, global optimality, and scalability for larger problems overcoming traditional limitations such as local optima and infeasibility in large-scale scenarios is achieved. By considering the uncertainties of solar and wind sources, a spinning reservation technique is used to increase microgrid stability. This study also examines how demand response programs help gas stations operate better and facilitate effective energy transfers between VPPs and the upstream network. As a major step toward increasing microgrid profitability and operational efficiency, the results demonstrate the superiority of establishing a strategic pathway for VPPs to optimize energy transactions, set competitive reserve market pricing, and handle market uncertainties.

Building a new power system dominated by renewable resource is the main approach to make the dual carbon goal come true. Nevertheless, deviation exists in the forecast of large-scale grid-connected wind power output, which imposes severe challenges for the electricity sales of load aggregators bearing certain absorption responsibility weight. Hence, demand response is introduced into the electricity sales strategy of load aggregators to help guide user's consumption behavior and effectively reduce the problems from deviation in clean energy forecast. First taking into account the hourly correction of wind power forecast, this paper proposes a carbon emission reduction integration model oriented to users' electricity consumption behavior which considers circumstance of wind power absorption. Secondly, a load aggregator - user model is constructed based on the rolling correction WPC-CERI (Wind Power Consumption - Carbon Emission Reduction Integration) for hourly correction of aggregator's electricity sale price and users' electricity consumption. Then, the two-layer model is converted into a single-layer one through KKT, and optimize the solution through IPOPT solver. The final example suggests that it is a feasible model to optimize the demand response package for load aggregators, mitigate the customer-side load demand, further absorb wind power and diminish carbon dioxide discharge.

In the realm of microgrid (MG), the distributed load frequency control (LFC) system has proven to be highly susceptible to the negative effects of false data injection attacks (FDIAs). Considering the significant responsibility of the distributed LFC system for maintaining frequency stability within the MG, this paper proposes a detection and defense method against unobservable FDIAs in the distributed LFC system. Firstly, the method integrates a bi-directional long short-term memory (BiLSTM) neural network and an improved whale optimization algorithm (IWOA) into the LFC controller to detect and counteract FDIAs. Secondly, to enable the BiLSTM neural network to proficiently detect multiple types of FDIAs with utmost precision, the model employs a historical MG dataset comprising the frequency and power variances. Finally, the IWOA is utilized to optimize the proportional-integral-derivative (PID) controller parameters to counteract the negative impacts of FDIAs. The proposed detection and defense method is validated by building the distributed LFC system in Simulink.

In the monthly trading market, electricity retailers participate in intra-month adjustment volume transactions with trading cycles ranging from ten days, weekly, to daily, where the traded electricity volume is based on the remaining days or specific days within the month. Considering user data security and privacy protection, this paper proposes a horizontally federated learning mechanism based on load decomposition and privacy protection for intra-month adjustment volume load prediction analysis. Firstly, each user locally extracts temporal features from the data using trend decomposition algorithms, decomposing the original load data into trend, periodic, and random components. Secondly, based on federated learning, prediction models are separately constructed for the trend, periodic, and random components. Finally, the electricity retailer aggregates the predicted components from each user to obtain the final prediction result. The proposed algorithm is validated using load data from users of a certain electricity retailer in China. The results demonstrate that the algorithm achieves excellent prediction accuracy and generalization ability while protecting user data privacy.

Regarding the significance of providing electricity to areas that are so remote from power networks, this study focused on the analysis and modeling of an independent micro grid. A Load demand response is a major part of this system to fulfill the electrical energy requirement on consumer end. An energy storage-based control system requires the design and implementation of a power conversion system. Energy storage systems can be used to mitigate the fluctuations from intermittent renewable energy sources. This paper proposes a design of the 8.5 kW wind turbine which incorporates the energy storage system to diminish the fluctuations. The proposed system consists of double conversion, i.e. AC-DC and DC to AC. AC -DC conversion is done by the rectifier and the output is connected to a common DC bus at which the controller and the energy storage system are connected. The controller conditional base algorithm checks the battery parameters, load, and dump load conditions. The SOC of the battery is set at 0.6 and 0.8, so dc bus bar line sets to be reference voltage to avoid the harmonics on output according to the requirement of the load. To achieve the reference voltage from the wind source and battery source then design a controller for the PWM inverter that observes the ac output of the inverter and its control for fluctuations then feeds back to the inverter in the form of gate pulses for smooth output.

With the increasing requirements for power system stability and the rapid development of new energy sources, demand response plays an important role in ensuring the stable operation of power systems. Commercial buildings, as an important part of demand response, can be divided into traditional buildings and intelligent buildings. Traditional buildings mainly rely on manual operation and regular maintenance to meet the needs of users, while intelligent buildings realize the optimization and monitoring of environment, energy and safety through automated systems and intelligent equipment, and automatically meet demand response based on advanced automation control technology. This paper constructs an evolutionary game model of demand response with two types of commercial buildings under the load integrator: intelligent buildings and traditional buildings as the main body; describes the difference between the two types of buildings participating in demand response; and draws the following conclusions through the analysis of the stability of the equilibrium point of the evolutionary game model: intelligent buildings will tend to participate in demand response, while traditional buildings will tend not to participate in demand response. In order to meet the requirements of demand response, commercial buildings will continue to transform into intelligent buildings.

With the increasing popularity of wind and solar generation, the stable and economic operation of the community microgrid system become more challenging. This paper is concerned with the robust scheduling and real-time control of demand response (DR) air conditioning (AC) loads for community microgrid. Firstly, a two-stage robust scheduling model coupled by the day-ahead dispatch and real-time dispatch is established by considering the uncertainties of wind/photovoltaic power generation and loads. Moreover, the incorporation of multiple interval uncertain sets is introduced to characterize the uncertainties related to wind power, photovoltaics, and load, thereby reducing the conservativeness of the model. Secondly, the real-time control of AC loads by adjusting the on/off actions and temperature setpoints with limited impact on user comfort so as to track the optimized electricity consumption curve. Finally, the simulation results show the effectiveness and robustness of the proposed optimization dispatch and real-time control algorithm.