As the penetration of power electronic converters and renewable generation increases, voltage and current waveforms in power systems are increasingly distorted and nonstationary. Accurate estimation and fast tracking of electric power quantities are essential for energy metering and billing, online monitoring, and safe operation of power systems. However, most existing estimation methods assume that the waveform within an observation window is stationary and periodic, which often leads to a slow dynamic response under nonstationary conditions and introduces estimation errors when the fundamental frequency deviates from its nominal value. To address these challenges, this paper introduces a framework termed operating point fluctuations (OPF). In the OPF framework, a nonstationary waveform is reinterpreted by treating the time-varying fundamental component as a dynamic operating point and describing all distortions as fluctuations around it. Based on this framework, a set of single-phase and three-phase power quantity expressions is derived. Furthermore, an estimation method is proposed. Comprehensive simulation studies demonstrate that the proposed method achieves high accuracy under stationary conditions, fast dynamic tracking under nonstationary conditions, low computational burden, and robustness to noise and fundamental frequency deviation. Validation using field measurement data further confirms these results and demonstrates its suitability for practical applications.
Fast and accurate fault identification is crucial for the safe and stable operation of multi-terminal flexible DC grids. Existing protection methods for flexible DC grids are sensitive to transition resistances, with limited engineering adaptability, making it difficult to balance speed and reliability. To address this issue, this paper proposes a single-ended protection method based on the voltage energy spectrum matrix of the current-limiting reactor. This method uses the difference of high-frequency components of voltage on both sides of the current-limiting reactor at the initial stage of fault, with the energy spectrum matrix similarity as the triggering criterion. Combined with the damping effect of the reactor, a fault criterion based on the high-frequency and low-frequency energy ratio is developed for both internal and external faults. The paper first analyses the voltage transient evolution of internal and external faults under the damping effect of the current-limiting reactor. It then studies the time-frequency characteristics of voltage signals under different operating conditions and, based on this, constructs a complete protection scheme. Finally, a multi-terminal flexible DC system model is built in PSCAD/EMTDC for simulation verification. The results show that high-frequency energy quickly concentrates at the early fault stage. The spectrum matrix is rapidly reconstructed, and the similarity between the two sides drops by more than 20%, which enables protecting fast starting. The high-frequency energy ratio shows a clear difference between internal and external faults (more than 49%). The method can complete fault identification within 3 ms after fault inception. It remains reliable under a 300 Omega transition resistance and 30 dB noise. In addition, it is not sensitive to changes in the reactor inductance and can also effectively identify lightning disturbances. Therefore, it is suitable for fast and high-reliability protection of long-distance multi-terminal flexible DC grids.
Tower spotting design in transmission line planning is a critical component, as it directly affects both project cost and safety. Conventional design approaches often inadequately account for terrain characteristics and global optimisation requirements, resulting in low efficiency or limited applicability in large-scale projects. This paper proposes a practical optimisation strategy for transmission line tower spotting that jointly considers multiple safety constraints and terrain features to achieve coordinated designs of tower locations and heights. First, an optimisation model is developed to minimise total construction costs, incorporating constraints such as horizontal span, vertical span, and ground clearance. Next, a terrain-driven optimisation method is developed, which utilises topographic peak points from the terrain profile as initial candidate locations for tower spotting, thereby substantially reducing computational complexity. The approach employs the Simulated Annealing Particle Swarm Optimisation (SA-PSO) algorithm to achieve global height optimisation. Finally, the proposed method is validated using a case study of 500 kV transmission line planning in Guangdong Province, demonstrating enhanced economic efficiency while fully complying with all safety requirements.
Stochastic unit commitment with AC network constraints is intractable for realistic systems. We benchmark machine learning surrogates for AC feasibility classification-Kolmogorov-Arnold networks (KANs), multilayer perceptrons (MLPs), gradient-boosted trees and support vector machines (SVMs)-on three power systems (118-2848 buses) under two feature tracks: pre-solve (load multipliers, renewable factors and contingency identity) and dispatch-aware (adding generator setpoints). KAN, MLP and tree-based models all achieve area under the ROC curve (AUROC) above 0.99 on the pre-solve track. On the dispatch-aware track, gradient-boosted trees lead with AUROC 0.997, while KAN with weight decay reaches AUROC 0.992, a residual gap of approximately 0.005. Violation-type error decomposition identifies which constraint categories drive false-feasible predictions. Split-conformal prediction with group-conditional calibration provides distribution-free coverage guarantees and flags uncertain points for full power flow verification. Speedups of three to five orders of magnitude over Newton-Raphson enable real-time screening.
The growing penetration of renewables and prosumagers in the 2030s will require deeper monitoring and control in medium- and low-voltage grids. Wireless technologies, particularly 5G, offer possibilities for cost-effective, flexible, and scalable connectivity in grid automation. Power system communication is highly heterogeneous, involving diverse traffic types with varying latency and reliability requirements. Prior research on 5G technology enablers, such as carrier aggregation, dual connectivity, and network slicing (NS), has focused on other verticals and relied mainly on simulations, leaving a gap in experimental validation for grid protection and automation under realistic conditions. This paper maps 5G technology enablers to IEC 61850 traffic types and evaluates their performance in commercial 5G networks for line differential protection and virtualised fault passage indication. Results show that NS improves reliability compared to traffic prioritisation, but does not yet meet ultra-low latency requirements for line differential protection. Minimum latency for mission-critical protection traffic reaches approximately 11 ms, and round-trip times are 32% lower than prior works. These findings provide the first end-to-end empirical evidence of 5G technology enabler deployment for smart grid applications and offer practical guidance for future implementations.
Bipolar hybrid cascaded multi-terminal high-voltage direct current (MTDC) systems feature the power density of LCCs, combined with the flexible control capabilities of the MMCs, which are suitable for long-distance bulk power transmission and can integrate renewable energy sources on a large scale, including the Baihetan-Jiangsu hybrid cascaded MTDC project proposed in China. The co-existence of these converter technologies, however, leads to complicated DC fault behaviour that is very challenging to manage and recover from after faults. This paper will provide a detailed discussion of DC faults in the bipolar hybrid cascaded MTDC systems, the faults in the line-commutated converter (LCC) section, the hybrid LCC-MMC interface, and the modular multilevel converter (MMC) section. The electrical characteristics of DC faults are determined with the help of electromagnetic transient simulations, which consider the voltage collapse, fault current propagation, and converter stress. After being determined to have common electrical signatures and dynamic behaviours, a new classification scheme is proposed, which classifies faults that share common characteristics, dynamic responses, and system-level responses. The control actions needed to alleviate the effects of DC faults, such as firing-angle control of LCCs, blocking and deblocking MMCs, and coordinated AC and DC circuit breaker operation, are also another focus of this paper. It pays special attention to the effective DC fault current suppression, voltage stabilisation, and minimisation of stress on LCC and MMC converters. Moreover, the recovery strategy is worked out, and three recovery cases are investigated for each of the major categories of faults. In these cases, different control sequences, converter re-energisation schemes, and breaker operation to stabilise the system and resume normal operation were involved. The effectiveness of the proposed fault classification, control actions, and recovery strategies is confirmed on the background of a PSCAD/EMTDC model of the Baihetan-Jiangsu hybrid cascaded MTDC system.
Safety distance monitoring is a critical requirement for live substation operations, where inaccurate distance perception may lead to severe safety hazards. Existing vision-based approaches commonly estimate distances using single representative points, such as detection box centres, which are sensitive to irregular object geometries, depth noise, and threshold-induced decision instability in complex operational environments. This paper proposes a vision-based safety distance monitoring framework that integrates image segmentation and binocular vision for robust spatial distance perception. Operational elements are modelled as three-dimensional regions rather than isolated points, and a mask-constrained region-to-region distance formulation is introduced by combining candidate point selection with robust statistical aggregation. An enhanced segmentation-based perception module is adopted to improve planar localisation accuracy under scale variation and deformation, while a lightweight pruning strategy enables real-time deployment. Experimental results on live substation operation scenarios show that the proposed method reduces the distance estimation error from 2.52 m to 1.04 m in MAE and from 3.48 m to 1.68 m in RMSE while improving the safety decision correctness from 88.6% to 92.4% and reducing both false alarms and missed warnings near safety thresholds.
Electromagnetic compatibility (EMC) is an essential requirement for the disturbance-free operation of equipment connected to electricity networks. The process of EMC coordination always starts with a detailed characterisation of the interference mechanism that has to be protected. Based on this characterisation, appropriate indices for the quantification of the mechanism have to be introduced. Finally, based on these indices, compatibility levels, immunity testing procedures, including test levels and emission limits, are established that form the best possible trade-off of sharing costs, risks and responsibilities between all involved stakeholders. Recently, flicker in LED lamps caused by interharmonic distortion was identified as a new interference mechanism which is not yet covered by the existing EMC coordination in the IEC 61000 series of standards. This article proposes a novel EMC coordination framework to address this interference mechanism. It shall serve as information for standardisation committees in IEC TC77/SC77A and IEC TC34 to extend or update the relevant standards. The article follows the process for EMC coordination for this new mechanism, starting from a detailed characterisation of it, proceeding to define suitable indices and measurement methods to capture it and finally proposing compatibility levels, immunity testing procedures, test levels and emission limits.
Connecting decentralized renewable production to the electricity distribution networks can generate voltage and/or current constraints, often requiring network reinforcement or renewable generation curtailment. Coupling electricity and heating networks through power-to-heat technologies offers an alternative to alleviate these constraints. This paper analyzes the impact of such coupling on the integration of renewable electric energy generation with a heat pump connected to a district heating grid to absorb surplus renewable generation. This coupling alleviates constraints on the electric grid and offers a renewable decarbonized alternative heat source to the coupled district heating grid, originally supplied by a CHP. A bi-level optimization framework is proposed to solve both operational and planning problems of the coupled electric and thermal networks. Results demonstrate that the proposed coupling method can increase renewable hosting capacity of distribution grids while avoiding network reinforcement and curtailment of renewable generation. A sensitivity analysis is conducted to assess the robustness of this approach. In the studied use case, the photovoltaic generation hosting capacity can be doubled while maintaining the same total system costs, or alternatively, total costs can be reduced by up to 50% for a given renewable production capacity. Adding thermal storage can further reduce these costs by up to 40%.
Power system restoration (PSR) is a time-critical and complex combinatorial optimization problem, particularly in the transmission path selection phase following a blackout. When applied to large-scale power grids, traditional metaheuristic algorithms often suffer from slow convergence and entrapment in local optima. To address these limitations, this paper proposes a novel hybrid ACO-A* algorithm, which embeds the deterministic A* search heuristic into the probabilistic transition rules of ant colony optimization (ACO). A comprehensive comparative analysis is conducted against the standard genetic algorithm (GA), standard ACO, hybrid GA-A*, Dijkstra shortest path and mixed-integer linear programming (MILP) via the IEEE 39-, 68- and 118-bus test systems. The methodology incorporates a hierarchical Dijkstra-based subsystem partitioning strategy to decompose complex networks into manageable restoration zones. All restoration paths are validated using AC power flow analysis to ensure voltage, thermal and transient stability constraints are satisfied. The simulation results demonstrate that the proposed hybrid ACO-A* significantly outperforms other methods, achieving a 100% success rate and reducing computation time by approximately 69% compared with standard ACO on the IEEE 118-bus system, demonstrating favourable scaling characteristics. The method shows potential for online decision support pending validation on utility-scale systems. Furthermore, this study extends the algorithm's application to strategic infrastructure planning, identifying optimal black-start (BS) unit locations that reduce the total system restoration impedance by 18% compared with betweenness-centrality-based approaches. Statistical validation using 20 independent runs per test case confirms result robustness. The findings demonstrate that the hybrid ACO-A* is a robust, scalable and superior solver for both operational recovery and resilience planning.
This paper presents a new Python-based methodology that combines the advantages of two established techniques for analysing electrical power systems (i.e., (i) harmonic domain (HD) modelling and (ii) PSS/E software tool). By integrating these two approaches, the Python-based program enhances the accuracy of the solution while maintaining computational resources.The proposed methodology primarily employs the harmonic domain admittance matrix (HDAM) that describes the system's behaviour in terms of harmonics. PSS/E can then analyse the system's power flow while considering the effects of harmonics. The paper showcases the proposed algorithm when using two specific scenarios (i.e., (i) integrating high-voltage direct current (HVDC) systems into the grid and (ii) integrating photovoltaic (PV) solar power systems into the grid).The proposed model was validated on the IEEE 39-bus system incorporating HVDC and PV connections. Benchmarking against detailed PSCAD time-domain simulations demonstrated high accuracy, with maximum voltage magnitude errors maintained below 1.5 & times; 10-3 pu and power mismatches not exceeding 2.5 & times; 10-3 pu. Furthermore, the proposed method significantly improved computational efficiency, reducing simulation time by over 98% (e.g., from 1760 s in PSCAD to 5.94 s with the proposed method) while achieving strict convergence (10-6) in fewer than 18 iterations.
Existing research has enhanced short-term photovoltaic (PV) power forecasting through weather-type classification frameworks; yet two limitations remain: (1) Numerical weather prediction data are overly smoothed, and errors are amplified under complex weather conditions, causing multi-stage error accumulation; and (2) weather type labelling methods are constrained - manual labelling is labour-intensive and subjective, while clustering based on statistical features struggles to capture nonlinear patterns, making accurate labelling difficult for complex weather. To address these issues, we propose a short-term PV power classification forecasting method leveraging spatio-temporal irradiance correlations and contrastive learning. First, a temporal convolutional network corrects predicted irradiance at the target site using correlations with neighbouring sites. Next, high-dimensional features extracted from the corrected irradiance via a TS2Vec contrastive learning model enable robust weather type classification. A Timemixer++-based model then predicts weather types for the target day. Finally, a classification-based forecasting framework uses the predicted weather types to generate PV power forecasts. Simulation results demonstrate that the proposed framework significantly improves the accuracy of short-term PV power forecasting.
Energy transmission lines are critical for reliable electricity delivery, but faults in these lines can disrupt power systems, leading to instability and outages. Rapid fault detection, accurate classification, and timely isolation are essential to maintain grid resilience. Traditional distance relays are widely used for fault detection. However, due to dynamic variations in system conditions and external disturbances, they face limitations in complex scenarios. This situation highlights the need for more advanced and precise fault detection methods. In this study, parametric analyses were conducted on a 400 kV medium-length transmission line, and various fault scenarios were created using system characteristic data with various program software to generate a dataset. The performance of deep learning techniques in fault detection was comparatively analysed based on the obtained datasets. Notably, the proposed hybrid deep learning algorithm outperformed other classification algorithms in predicting critical information such as fault type and zone detection. The proposed hybrid framework demonstrates strong predictive performance, ensuring high accuracy in fault type classification, 99.994%, and zone localisation, 99.981%. By systematically integrating conventional protection schemes with advanced computational methodologies, the approach is anticipated to significantly enhance the reliability and operational efficiency of fault management in high-voltage transmission networks.
The reverse operation of Under-Load Tap Changers (ULTCs) poses a significant risk to system voltage stability. This study presents a straightforward method to mitigate voltage instability caused by ULTCs' reverse action. As the system approaches instability, power network equipment may fail to transmit reactive power effectively to loads, resulting in a waste of network-generated reactive power. Transformers equipped with tap changers are particularly susceptible to this issue. To address this, a novel voltage stability index is introduced, which uses local measurements to assess variations in reactive power consumption during tap changes. This index identifies transformers that contribute to voltage instability and blocks the tap changer controllers. The effectiveness of the proposed approach is evaluated through simulations on the Nordic test system using DIgSILENT software. The proposed index has been implemented for various scenarios, including generator tripping, load increase, and line outage, under different types of static and dynamic loads and in the presence of measurement noise. It has also been compared with a state-of-the-art method proposed in previous studies. The results confirm the accuracy and reliability of the proposed index in identifying tap changers with reverse performance. Furthermore, by blocking tap changers exhibiting reverse performance, voltage instability can be prevented. The main advantages of the proposed index include its low computational burden, ease of practical implementation, and dependence on local measurements only.
Distribution networks are complex and constantly changing, making accurate monitoring difficult. Effective balancing and management of the distribution networks depend on precise phase identification, but traditional manual or voltage-based methods often lack accuracy and scalability. Using smart meter data with data-driven techniques offers a promising solution. This paper presents a novel method utilizing smart meter data to identify consumer phase connections. The dataset includes 90 consumers from a UK distribution network, evenly divided across three phases. In addition to individual consumption data, total phase consumption was collected every 30 min over a 20-day period. A high-pass filter was applied to highlight dynamic consumption variations, followed by clustering based on the correlation between individual and phase consumption using neural gas network algorithm. The method was extensively evaluated under various conditions, including high-resolution (30-min interval) and low-resolution (daily aggregated) data, as well as simulated missing data with different imputation strategies. The results have been compared with correlation-based K-means and mean shift which indicate that high-resolution data combined with mean imputation achieves the highest clustering accuracy, with K-means and neural gas network outperforming mean shift. Lower-resolution data significantly reduce accuracy due to the loss of temporal dynamics. While increasing the number of consumers slightly affects accuracy, the high-resolution approach remains robust. Furthermore, tuning preprocessing parameters such as filter cutoff frequencies and sampling rates further improves performance.
To enhance the fault handling capacity of low-voltage DC (LVDC) distribution networks and mitigate the detrimental effects of faults on converter apparatus, this research introduces an innovative modular fault suppression and handling strategy. The proposed approach entails the integration of multi-port fault-limiting modules (FLM) with DC circuit breakers (DCCBs) and DC transformers, thereby enhancing the system's current and voltage regulation. This integration allows for the use of economical mechanical DCCBs while improving the fault-ride-through (FRT) capabilities of the conversion equipment. Under normal operating conditions, the FLM is bypassed to reduce system disturbances; in the event of a fault, it is engaged to curtail fault currents and preserve voltage stability. The validity of the proposed method is verified through simulations in PSCAD/EMTDC. Simulation results demonstrate that the proposed strategy effectively reduces the peak fault current and limits the transient voltage compared to conventional schemes. Furthermore, the system successfully achieves a fault ride-through by maintaining the fault current within a safe range of the rated current, ensuring robust protection and rapid recovery.
Voltage flicker caused by fast power fluctuations is one of the main power quality concerns in grid-connected wind farms, especially in distribution networks with low reactance-to-resistance (X/R) ratios. Static VAr compensators (SVCs) are commonly used to mitigate flicker by regulating reactive power; however, conventional SVC control strategies usually neglect the effect of active power variations. This assumption can be acceptable in transmission networks with high X/R ratios, but it is not sufficiently accurate in low-X/R distribution networks, where active power variations have a noticeable effect on voltage fluctuations. Therefore, this paper proposes an improved SVC control strategy in which the active power of the wind farm is explicitly included in the compensation process. Moreover, to reduce the adverse effect of the inherent SVC delay caused by power calculation and thyristor firing, a prediction-based delay compensation approach is applied using adaptive filtering methods, including normalised least mean squares and recursive least squares algorithms. The proposed approach is evaluated using real voltage and current measurements obtained from the Manjil wind farm in Iran under different network X/R ratios. The results show that, in low-X/R networks, using SVC without considering active power is not effective for flicker mitigation and can even increase the flicker level compared with the uncompensated condition. A similar behaviour is observed for static synchronous compensator (STATCOM) control when active power is neglected. In contrast, when active power is included in the SVC control, the short-term flicker severity index is reduced by 33% for X/R = 1 and by 44% for X/R = 0.5. By adding the proposed delay compensation controller to the SVC and predicting both active and reactive powers, the flicker reduction is further improved to 44% for X/R = 1 and 54% for X/R = 0.5. The results obtained from STATCOM also confirm the importance of active power consideration, as the flicker reduction reaches 48% for X/R = 1 and 57% for X/R = 0.5 when active power is included in the STATCOM control. For high-X/R networks, the effect of active power becomes negligible, indicating that conventional reactive-power-based compensation is mainly suitable for transmission networks, while active-power-aware control is essential for effective flicker mitigation in low-X/R distribution systems.
Predicting sub-cycle transient peak currents from inverters within the first few milliseconds after fault inception is essential for instantaneous overcurrent protection coordination, yet the most commonly used standard for short-circuit calculation, like IEC 60909-0 and ANSI C037.010 addresses only quasi-steady-state currents. This paper proposes a closedform analytical framework that decouples inverter fault current into hardware natural response and control-commanded response. The hardware natural response, governed by inverter output filter and passive grid resonances, dominates the first transient peak and requires only nameplate and filter data-no proprietary controller information. The controlcommanded response, shaped by the current loop, phaselocked loop, and fault ride-through strategy, governs subsequent waveform evolution. Building on unified hardware modeling, the framework extends separately to Grid-Following and Grid-Forming inverters. Validation against electromagnetic transient simulations and power hardwarein-the-loop tests on three commercial inverters confirms prediction accuracy.
Modern power systems exhibit continuously varying operating conditions. To assess transient voltage stability in this context, the varying operating conditions are characterised as variable parameters in parameter space, enabling the investigation of the transient voltage stability region boundary (TVSRB). The TVSRB represents a hypersurface dividing the parameter space into stable and unstable subspaces in terms of transient voltage stability. To determine the TVSRB in parameter space, first, a smoothness-enhanced transient voltage stability index (SETVSI) based on the transient voltage stability criterion is defined, which has a critical value along with tunable coefficients to enhance the smoothness of the implicit function between SETVSI and variable parameters. Second, polynomial approximation based on collocation point method is adopted to approximate this implicit function with an explicit polynomial. Third, based on the explicit polynomial, the principle of TVSRB determination in parameter space is introduced, and implementation issues are discussed. Finally, the accuracy and practicality of the proposed method are verified in the IEEE 9-bus system and an actual system.
The fault currents generated by direct-drive wind farms and voltage source converter (VSC) stations differ from those of synchronous generators, posing significant challenges to conventional protection methods that rely on current amplitude or phase-angle variations. This paper proposes a pilot protection scheme for transmission lines based on higher-order statistics. The distribution of fault current samples is firstly analysed for both external and internal faults conditions. The differences in the features are quantified by using the third-order cumulant and kurtosis. On this basis, protection criteria are constructed using a two-dimensional Euclidean distance to combine these statistics. PSCAD/EMTDC simulation studies confirm that the proposed scheme can rapidly and reliably identify faults under various fault locations, types, transition resistances, and inception angles. Compared to conventional and alternative pilot protection, the proposed scheme exhibits superior sensitivity, speed, and robustness, particularly in high-resistance fault scenarios and weak-source conditions.
