Iet Generation Transmission & Distribution最新文献

Peer-to-peer (P2P) energy trading has emerged as a promising approach for managing energy produced by prosumers. However, the influence of diverse prosumer behaviours on P2P energy trading remains underexplored. This paper provides a comprehensive analysis of the various behavioural patterns shown by sellers in the energy trading market. Unlike previous studies that have examined only limited aspects of prosumer behaviour, this paper, in addition to behaviours such as competition and coalition, also examines the likelihood of strategic behaviours arising in sellers' decision‑making processes. To achieve this, game theory, a robust framework for modelling individual and collective behaviours, is employed. In the proposed model, buyers act as price proposers, while sellers serve as energy suppliers. The behaviour of sellers is modelled under different conditions: competition, coalition, and coalition suspension. The analysis reveals that coalition formation among sellers yields higher payoffs compared to competition behaviour. However, it is also demonstrated that coalitions can be suspended (violated) because prosumers can achieve greater gain by suspending their coalitions. Additionally, prosumers employ the grim trigger strategy to prevent the suspension of coalitions. Additionally, in another section of this article, a new bilateral negotiation mechanism is presented, which is designed to be implemented in a distributed manner within the structure of P2P energy trading. This market-clearing mechanism is designed to consider, in addition to economic constraints, technical and operational constraints and the matching of buyers and sellers. Moreover, the mechanism includes a constraint to prevent the emergence of market power and to address coalition and strategic behaviour by sellers. That constraint is applied so as, on the one hand, not to reduce sellers’ participation and, on the other, to remain effective. To evaluate the proposed approach, the performance of the bilateral negotiation mechanism, the economic aspects of the suggested method, the analysis of prosumer behaviour, the impact of coalition suspension and the effects of purposeful behaviours on P2P trading have been examined.

Accurate inertia estimation for grid-connected doubly fed induction generator (DFIG)-based wind farms is essential for providing inertia support and maintaining frequency stability. However, due to the diversity of operating conditions and control parameters among individual wind turbines, a DFIG-based wind farm cannot be represented as a single equivalent device. In this paper, a time-varying inertia estimation framework based on sensitivity-guided clustering and aggregation is proposed. First, a sensitivity analysis framework is proposed to analyse the impact factors affecting the time-varying inertia of individual DFIGs, using the extended Fourier amplitude sensitivity test. After that, the dominant factor, identified as the virtual inertia control parameter, is selected as the clustering indicator. Subsequently, DFIGs with similar dominant factors are clustered based on limited measurement data, utilizing the unscented Kalman filter to reduce the requirement for extensive measurement devices. The time-varying inertia of the wind farm is then estimated using streaming dynamic mode decomposition with information from each cluster. Simulation and experimental results demonstrate that the proposed framework achieves high-accuracy inertia estimation with limited measurements, reducing the relative error by nearly 10% compared with existing methods. Moreover, it exhibits strong robustness to noise and disturbances, confirming the effectiveness of the inertia estimation.

This study focuses on islanded AC microgrids and addresses the dynamic stability challenges caused by renewable energy variability and uncertain load demand. The work specifically targets the secondary control layer, which plays a critical role in restoring voltage and frequency while ensuring accurate power sharing among distributed generators (DG). The primary control relies on conventional voltage–current (V–I) droop characteristics to provide decentralized operation, but this approach alone leads to steady-state deviations and limited power-sharing accuracy. To overcome these limitations, we propose an advanced distributed secondary control strategy based on consensus algorithms. At this layer, the two fundamental parameters of the droop characteristic—its slope and offset—are dynamically tuned in a coordinated manner. Active power sharing is improved by adjusting both the slope and the offset of the d-axis droop, while reactive power control is refined through modifications to the q-axis slope. This dual-parameter adaptation ensures robust proportional power sharing, precise voltage regulation at the point of common coupling, and resilience against communication delays. The effectiveness of the proposed secondary control scheme is validated through detailed simulations in MATLAB/Simulink, demonstrating enhanced stability, faster transient recovery, and improved power quality under varying load conditions.

The load frequency control (LFC) scheme, as a vital application in power systems' stability, makes the power system susceptible to cyber-attacks due to its dependence on information technologies and communication networks. This paper studies the LFC performance of Kundur's 4-unit-12-bus power system under false data injection (FDI) attacks. The available defence schemes are either based on the system's model or data-driven. The effectiveness of these schemes depends on the precise mathematical modelling or the extensive historical data of the power system. Therefore, it is necessary to design a defence strategy without depending on the mathematical model and the historical data of the system. To this end, this paper proposes a model-free resilient defence strategy, comprising a model-free detection scheme and an event-triggered mechanism. The presented detection scheme accomplishes the manipulated signal estimation using the measurement and control signals and compares the difference between the estimated and observed signals with a predefined threshold value. When the difference exceeds the threshold value, the detection scheme announces that an attack has occurred on the system. After detecting an attack, the event-triggered mechanism is activated to mitigate the attack's effect on the system frequency response. Accordingly, the event-triggered mechanism blocks the falsified signal and submits the estimated signal to the LFC controller. The presented scheme is independent of the system's mathematical model and historical data and can be employed in any cyber-physical power system. The design process of this strategy is simple and independent of the size and complexity of the power system. A deep reinforcement learning algorithm is also employed to tune the adjustable parameters of the proposed method. The real-time results obtained by the OPAL-RT simulator show that the developed scheme can timely identify FDI attacks and completely mitigate the attack's effect on the system's dynamic performance.

The increasing frequency of extreme cold waves exacerbates wind power uncertainty, intensifying the trade-off between robustness and economy in high wind penetration power systems. To address the problem, this paper proposes a DRO method based on distributionally robust Bayesian inference (DRBI). An ambiguity set defined by the Wasserstein metric is first constructed utilising historical wind data. Secondly, the likelihood distribution of wind power output is predicted using an XGB-transformer model. To accurately characterise wind power output during cold waves, a posterior distribution is then constructed using the proposed DRBI framework. Next, a DRO dispatch model is constructed to ensure operational robustness while minimising total operating cost. Constraints include power balance, wind power uncertainty and system security requirements. The model is solved based on strong duality theory. Finally, the model is validated on a regional 30-bus system and a modified IEEE 118-bus system. Experimental results show that, compared to stochastic optimisation and robust optimisation models, the proposed model effectively balances robustness and economy under cold waves. Besides, accounting for wind power uncertainty, experimental results suggest maintaining wind power penetration at 10–20%. Moreover, the economic efficiency of the optimal schedule can be further improved by adjusting the sample size of cold-wave scenarios.

The ever-increasing demand for electricity necessitates constant innovation in the electric utility sector. In this regard, high surge impedance loading (HSIL) transmission lines can be a promising technology. While conventional HSIL designs rely on more subconductors located symmetrically on circular bundles with a larger radius, unconventional HSIL lines can achieve even more natural power by optimally positioning subconductors in space. This paper focuses on determining the optimal location and number of shield wires for a newly designed 500 kV unconventional HSIL line, whose surge impedance is reduced to 141.5$���\Omega �$, resulting in a 74% increase in SIL compared to a conventional configuration (996 MW). New equations for calculating line inductance and capacitance, considering transposition for both phase and bundle arrangements, are developed. The shielding design aims to maintain an SFFOR of 0.05 flashovers per 100 km-years. Numerical analysis indicates that placing the shield wire at $�x$ = 7.2 m yields SFFOR = 0.047 under high lightning activity (Td = 30), while $�x$ = 6.35 m maintains SFFOR < 0.05 under low activity (Td = 5). These results are validated through geometric circle diagrams, demonstrating effective shielding for all three phases without increasing tower height. This study presents a practical shielding method for unconventional HSIL lines that have the potential to revolutionize bulk power transmission.

The utilization of data for generation output forecasting in combined cycle power plants (CCPPs) makes room for attackers to exploit and degrade the performance of machine learning models. This work investigates the potential impacts of cyberattacks on energy generation forecasting models for CCPPs. Attacks are implemented on the four top-performing forecasting models: gradient boosting, extreme gradient boosting, Random Forest, and CatBoost, identified through a comparative analysis of 12 models, including various tree-based methods, support vector machines, deep learning models, and linear regression techniques. Scaling, denial of service, fast gradient sign method, and basic iterative method attacks are employed with diverse attack volumes and perturbations to investigate the vulnerability of these models. To counteract these vulnerabilities, a two-layer defence scheme employing ensemble adversarial training is proposed, aimed at mitigating the adverse effects of these cyberattacks. The findings underscore the significance of the proposed robust defence strategy in ensuring the reliability of forecasting models in the presence of cyberattacks.

With the continuous energy transition and digital industrial upgrades, the collection and analysis of electricity consumption data have become the foundation for supporting the intelligent power system operations and load-side energy efficiency optimisation. However, this development raises significant concerns regarding user privacy and potential data leakage. To address the issue, this paper proposes a battery-aided personalised differential privacy framework that generates physically realisable noise through battery operations, thereby overcoming grid stability limitations associated with conventional virtual noise injection methods. The proposed method enables adaptive privacy preservation through bounded noise probability functions with configurable privacy parameters and incorporates time-of-use pricing via a mean-drift mechanism to enhance the economic viability of battery dispatch strategies. A multi-objective sand cat swarm optimisation algorithm is employed to determine the optimal configuration of battery parameters. Numerical experiments demonstrate that the proposed method effectively defends against load forecasting attacks and significantly reduces electricity costs for users, offering a viable solution for balancing privacy protection and economic benefits in smart grids.

The rapid growth of electric vehicles (EVs) has made accurate forecasting of charging station loads essential for ensuring grid stability and supporting infrastructure planning. While previous studies have investigated this problem, correlation-based feature selection remains relatively underexplored, which may lead to redundant features and reduced prediction accuracy. To address this issue, this paper proposes a hybrid forecasting model, RIME-CNN-LSTM-Attention, which integrates correlation-driven feature selection with advanced deep learning. Pearson, Spearman, and Kendall's tau-b analyses are first applied to identify the most influential factors affecting charging demand. The RIME algorithm is then used to optimise the hyperparameters of the CNN-LSTM network, while the attention mechanism dynamically emphasises critical load fluctuation periods. Case studies utilising actual charging station data illustrate that the proposed model substantially surpasses benchmark methodologies, thereby improving the accuracy and resilience of electric vehicle charging load forecasting.

Electric vehicles (EVs) integration into urban distribution system is surging globally, leading to an increase in the demand for EV charging load, which brought significant hosting risk in urban power systems, such as transformer overload. This study aims to characterize and model the spatial-temporal impact of EV charging on hosting risk of urban power systems across various urban scales considering the recent development of fast charging stations. To achieve this, it proposes a novel framework for calculating the hosting risk. This provides sufficient guiding information for maintaining reliability in urban power systems. In detail, we systematically model the EV charging patterns by using high-resolution real-world EV trajectory and charging record data. K-means clustering is utilized to identify typical charging patterns in various urban scales. Then, the novel hosting risk indicators considering the fast charging station development and the EV growth based on urban EV charging pattern distribution are proposed, where random forest is used to model the critical parameters. Finally, the hosting risk indicators under different urban scale and different EV charging conditions are illustrated. The finding indicates that the hosting risk of urban distribution system is closely related to the EV charging patterns, showing significant regional differences. In Shanghai, the overall city charging pattern exhibits a double-peak structure with morning and evening peaks at 10:14 and 21:04, respectively. The maximum 24-h hosting risk indicators for different charging types vary significantly, with Type 2 charging patterns showing the highest risk. The spatial-temporal changes of which can provide significant information for future plans of EV charging stations and reliability maintenance schemes.