Moisture is one of the important factors causing the degradation of transformer oil insulation performance. This paper proposes a method for predicting the water content in oil based on ultrasonic envelope analysis features and machine learning algorithm. First, the prepared oil samples with different water contents were subjected to ultrasonic testing. Then, the ultrasonic features were extracted based on envelope analysis method, and the relationship between the envelope analytical features and the water content in the oil was analyzed. Finally, a prediction model of water content in oil was constructed by random forest. The results show that the extracted ultrasonic envelope analysis features can effectively characterize the water content in the oil. The average prediction accuracy of the test set is higher than 90
This paper presents a high power density DC-DC boost converter based on the proposed multiple-switch topology and soft-switching technique. The proposed multiple-switch DC-DC boost converter (DBoC) employs a half-interleaved PWM switching technique to reduce the inductor current ripple and required inductance, enabling smaller inductor size under high switching frequency operation. To reduce the semiconductor losses caused by high-frequency operation, a lossless snubber circuit is implemented to achieve soft-switching operation. A 400 W 3-switch DBoC and 5-switch DBoC are designed, simulated, and experimentally verified. The results show that the required inductor volume of the 3-switch DBoC and 5-switch DBoC is reduced to 43.87
To enhance the operational performance of new energy generation systems under non-ideal grid conditions, this paper proposes a unified power quality conditioner (UPQC) based on a nine-switch converter (NSC). Unlike existing literature that merely uses NSC as a direct replacement for traditional back-to-back converters, this work focuses on the power quality conditioning scenario and solves the critical issues of insufficient DC voltage utilization and weak fault ride-through capability in conventional NSC applications. First, the topology structure, working principle, and mathematical model of NSC-UPQC are analyzed. Then, an optimized carrier-based modulation strategy with dynamic modulation ratio allocation is proposed to improve DC bus voltage utilization and realize seamless switching between voltage compensation and current compensation modes. Moreover, a double closed-loop control scheme combined with feed-forward is designed to enhance the stability and dynamic response under symmetrical and asymmetrical grid voltage sags. Finally, the effectiveness of the proposed topology and control strategy is verified by simulations and experiments under voltage sags and nonlinear load conditions. The results show that the proposed NSC-UPQC reduces the number of switching devices by 25
Parallel operation of transformers is widely used in substations and power plants to increase capacity and operational flexibility. However, small mismatches in voltage conversion ratios can generate circulation currents that do not contribute to load power transfer and may cause excessive copper losses, reactive power exchange, overheating, and reduced reliability. Most existing analyses focus on two-transformer configurations and rely on simplifying assumptions such as known load-terminal voltage, negligible load-impedance influence, and neglect of magnetizing current and iron losses. This study presents a generalized full analytical method for circulation current calculation in multiple parallel-connected transformers. The formulation is developed for three single-phase transformers as the most general practical configuration and is extendable to an arbitrary number of parallel units. Two complementary cases are derived: (i) neglecting magnetizing current and iron losses, and (ii) including magnetizing current and iron losses. Closed-form expressions are obtained for load-terminal voltage, winding currents, and circulation current components, explicitly accounting for load impedance. Comparative MATLAB-based analyses demonstrate that voltage-ratio mismatch produces significant imbalance in current and power sharing, increases RMS winding currents, distorts reactive power distribution and transformer power factors, and degrades efficiency effects that become substantially more pronounced when magnetizing current and iron losses are included.
Transformer bushings are critical components in power systems, and their failures may lead to severe operational and safety consequences. Although numerous studies have investigated bushing design and aging, the influence of environmental surface pollution on the internal electric field of oil-impregnated paper (OIP) bushings has received limited attention. This paper presents a 3D FEM analysis of a 145 kV OIP capacitor-type bushing to investigate how external pollution affects the internal electric field distribution. The pollution layer is modeled as a conductive film on the porcelain surface with three severity levels (light, medium, heavy), four thicknesses (0.5–2 mm), with both uniform and non-uniform distributions. Several realistic contamination patterns, including top-surface pollution, vertical gradients, and fan-shaped wind-driven deposition, are considered. The results show that severe uniform pollution can produce an excessive local electric field increase, along the inside path relative to the clean condition, while the electric field near the outermost capacitor foil can also increase significantly. Among the non-uniform cases, a top-heavy pollution gradient and large-angle fan-shaped contamination produce the most significant field enhancement. These findings highlight that both pollution severity and spatial distribution strongly influence the internal electric field of OIP bushings and should, therefore, be considered in their design and reliability assessment in polluted environments.
This paper presents a robust method for detecting anomalous electricity consumption using spectral clustering, tailored for real-world smart meter data from a Russian region. The study analyzes hourly readings from 1745 customers in the North Caucasus, specifically addressing challenges like significant data gaps and zero values. Our three-stage method involves constructing aggregated consumption profiles (daily, weekly, monthly), filtering out customers with negligible consumption, and applying spectral clustering with t-SNE visualization. This approach effectively categorizes customers into three distinct groups: those with stable load profiles, anomalous consumption, and insignificant consumption. The highest clustering quality was achieved using the filtered monthly consumption profile, with a Davies-Bouldin Index of 0.9741 and a Silhouette Coefficient of 0.4535. The results confirm the method's effectiveness for monitoring and analyzing electricity consumption. The developed universal method is applicable to any energy consumption dataset. Furthermore, we provide an open dataset of 27-month hourly consumption data from customers to support further research in electricity theft detection. The proposed method serves as an efficient preliminary screening tool to identify suspicious customers for subsequent expert verification.
Rapid growth of rooftop photovoltaic (PV) installations has increased the need for secure, privacy-preserving, and scalable Renewable Energy Certificate (REC) trading mechanisms. Currently, none of the blockchain-based REC trading frameworks has privacy-preserving federated intelligence, temporal modeling using transformers, and communication-efficient IoT support, which prevents scalability and secure REC verification. This paper presents an energy-aware federated transformer learning (EA-FTL)-assisted blockchain framework to overcome the above limitations and achieve privacy-preserving REC verification and trading while simultaneously maintaining energy efficiency. The framework combines the differential privacy (DP), secure aggregation (SA), trust-weighted proof-of-authority (TW-PoA) consensus, and adaptive compression-aware federated aggregation (ACFA). Rooftop PV units are equipped with IoT devices that gather the PV generation data via edge gateways and the federated models are trained by learning the energy trading patterns without sharing any raw user information. Hyperledger Fabric enables REC issuance, verification, and trading with low latency transactions with tamper resistance capabilities. The framework has been tested with experimental data from rooftop PV, Household Power Consumption (UCI) dataset, Photonics generation datasets (NREL), blockchain transaction logs, and edge communication datasets in a co-simulation scenario in MATLAB/Simulink, TensorFlow, and Hyperledger Fabric. The experimental results showed the 96.8
To address the challenges posed by intermittent power supply and random load fluctuations to frequency stability in microgrids, and to effectively utilize the energy storage potential of electric vehicles (EVs), a prior knowledge-enhanced method for EV load dynamic characteristics is proposed to enhance microgrid frequency resilience. By quantifying the dynamic characteristics of EV charging and discharging loads, prior knowledge is derived and transformed into constraints. A two-stage microgrid frequency resilience enhancement model is then constructed. In the first stage, each sub-microgrid invokes the prior knowledge of EV charging and discharging loads. With the objectives of minimizing load shedding and achieving cost optimization, the frequency resilience of the sub-microgrid is preliminarily improved by scheduling local resources and optimizing EV charging/discharging tasks, subject to various constraints including those of diesel generator sets and vehicle-to-grid (V2G) capabilities. In the second stage, by aggregating the data reported from each sub-microgrid and considering the constraints of available EV capacity and cross-microgrid transmission power, system load shedding and cost are further optimized through cross-regional mutual assistance utilizing EV energy storage. Ultimately, frequency resilience is significantly enhanced under coordinated multi-microgrid operation. Experimental results demonstrate that this method can effectively guide EVs to participate in microgrid charging and energy storage, releasing electrical energy during peak periods to buffer frequency fluctuations based on prior knowledge. It accurately manages the timing of charging and discharging, handles peak fluctuations, and shortens frequency recovery time. Moreover, the maximum value of the frequency resilience evaluation index for the multi-microgrid system remains around 14, indicating an effective balance between cost and power supply capacity, improved frequency resilience, and stable operation. It is verified that this method significantly enhances microgrid frequency resilience through structured knowledge-driven optimization, thereby achieving the predetermined research objectives.
To address the issues of high current ripple and strong parameter dependency in finite control set model predictive current control (FCS-MPC) for permanent magnet synchronous motors (PMSM), an improved discrete space vector model predictive control method based on prediction error compensation is proposed. First, the vector space is equidistantly partitioned into multiple equilateral triangles. An extended vector set is constructed based on the triangle vertices. Vertices are represented using barycentric coordinates and numbered, while the nearest neighborhood of each vertex is defined based on these coordinates. The value function optimization is transformed into determining the neighborhood of the reference voltage, thereby reducing computational complexity. Second, an improved nonlinear extended state observer (NESO) estimates voltage errors caused by equivalent lumped disturbance. This error is fed forward to compensate the reference voltage, enabling precise selection of the optimal voltage vector. Finally, experiments validate the proposed method’s effectiveness. Results demonstrate reduced computational load and improved robustness against parameter mismatch.
Alternating-current (AC) copper loss in stator windings of large AC machines originates from eddy-current phenomena in the alternating slot-leakage field: Within individual strands, it appears as skin effect and proximity effect, while at the winding level the same leakage-field mechanism induces electromotive-force differences among parallel strands and thereby drives inter-strand circulating currents. Winding transposition can suppress circulating-current loss by balancing leakage-field exposure among strands; however, comparing candidate schemes in early design requires a dimension-reduced model that preserves main in-slot electromagnetic features. This study proposes a two-dimensional electromagnetic finite element method combined with external-circuit reconstruction. The winding is discretized axially into representative cross sections. Current-density distribution and copper loss are calculated from the two-dimensional field solution, while external-circuit constraints reconstruct the series–parallel strand topology. A stator winding with ten solid strands is investigated, and four schemes are compared under the same conditions: no transposition, end-side resistance connection, in-slot 180 degree transposition, and in-slot 360 degree transposition. Numerical results show that the 360 degree complete-transposition scheme gives the most uniform strand-current distribution and lowest AC resistance factor over 70–150 Hz. A double-slot stator-core platform is used to test three physical windings: no transposition, 180 degree transposition, and 360 degree transposition. At 90 Hz, the relative errors between simulation and experiment are 7.82
With the integration of large-scale renewable energy into the power grid, the intermittency and randomness of its output significantly enhance the dynamic and uncertain characteristics of regional power flow distribution. Traditional iterative power flow calculation methods are facing challenges such as poor convergence, strong dependence on initial values, long computation time. To meet the requirements of fast and accurate power flow analysis under high renewable penetration scenarios, a data-driven method is proposed for constructing a power flow proxy model (PFPM) for PFPM renewable energy integration areas. The proposed method first constructs high-dimensional input–output vectors incorporating dynamic characteristics of sources, grids, and loads. And then an improved deep neural network is adopted for establishing PFPM. To improve training efficiency and alleviate the gradient vanishing problem caused by the combination of traditional Sigmoid activation and mean squared error loss, a cross-entropy loss function is introduced into the network architecture of PFPM. Moreover, two key physical constraints including Kirchhoff laws and voltage magnitude limits are embedded via a penalty function approach to enhance the physical consistency, and a firefly algorithm is used to dynamically optimize the penalty coefficients to better balance the prediction accuracy with the satisfaction of soft physical constraints. Finally, the improved IEEE 118-node system is opted for analyzing the performance of the proposed method considering the multi-scenario training samples. Simulation results show that the proposed model achieves a voltage prediction error of 1.3
Existing deep learning models for electricity price forecasting using multidimensional exogenous variables are often constrained by static feature fusion mechanisms. This limitation makes the models difficult to capture deep temporal dynamics and achieve explicit alignment. To address these limitations, this paper proposes a novel Context-Enhanced NBEATSx (CE-NBEATSx) model based on a Temporal Convolutional Network (TCN) Context Encoder and a Multi-Head Cross-Attention (MHCA) mechanism. First, the TCN Context Encoder is employed to efficiently extract multi-scale temporal features from historical exogenous variables. Subsequently, the MHCA mechanism dynamically aligns these features with the target price series, allowing the model to adaptively focus on the most relevant historical contextual information. Finally, the TCN Context Encoder and the MHCA mechanism are integrated within the doubly residual decomposition framework of NBEATSx. Experimental validation on three datasets from the EPEX market in Germany (DE-I and DE-II) and the PJM market in the United States (US) shows that CE-NBEATSx outperforms the evaluated forecasting models. Specifically, relative to the five baseline models, it reduces the Mean Absolute Error (MAE) by 14.92
This paper presents a Python-based tool designed for network service providers to estimate the frequency regulation and voltage support services available from the distribution networks to the upstream networks (e.g. transmission networks). The proposed tool includes a novel three-step time aggregation method to obtain representative profiles from historical time series data of loads, solar generation, wind generation, and the charging characteristics of electric vehicles. At first, the tool checks the security of distribution network considering its operating conditions. Then, it estimates available ancillary services, i.e. supports for voltage and frequency regulations, which can be provided in a safe manner for the upstream network. The performance of the designed tool has been evaluated using the real-world data from a network service provider in Australia.
Low-voltage flexible interconnection systems (LVFISs) face challenges such as transient impact, unreliable seamless switching, weak fault ride-through capability, and a lack of accurate in-loop test methods. Existing research mainly optimizes control strategies but fails to coordinate transient response, switching, and fault ride-through. To address this, a Hardware-in-the-Loop (HIL) test platform integrating core LVFIS modules is developed. A Fault-aware Seamless Switching Control (FSSC) module, combining a virtual synchronous generator and flexible impedance, is embedded using a virtual synchronous machine and transient optimization. Key innovations include an adaptive transient suppression loop, fault-responsive switching, and an engineering parameter self-calibration module. Experimental results show that transient current impact is limited to 120
With the continuous expansion of power grids, power grids have become more and more complicated, and their vulnerability has increased significantly. It is known that node failures may trigger large-scale blackouts; hence, the accurate identification of critical nodes is crucial for safeguarding power grid security. Traditional identification methods either over-rely on topological structures or focus on single electrical features, which limits their ability to capture the comprehensive impact of critical nodes on power grids and results in incomplete characterization of node importance. To overcome these limitations, this paper proposes an electrically guided dual-stream convolutional neural network (EGDS-CNN) framework to identify the critical nodes of power grids. Specifically, we develop an electrically guided neighborhood sampling strategy that prioritizes nodes with high power flow centrality (PFC) to be selected as the generalized neighbors and extracts their features; then, we use position encoding to weight the node feature, and a structured feature matrix for each node can be constructed. On this basis, EGDS-CNN reformulates critical node identification as a nonlinear regression task via convolutional neural networks. By integrating an adaptive attention mechanism into dual-stream CNN architecture, EGDS-CNN can achieve deep fusion of electrical and topological information, and the precise identification of critical nodes can be obtained. Experimental results show that EGDS-CNN significantly outperforms the existing methods in vulnerability analysis, which confirms that EGDS-CNN is a more effective approach for critical node identification.
Voltage unbalance is one of the major power quality issues in distribution systems. It is primarily caused by the presence of single-phase loads and the increasing penetration of distributed energy resources (DERs) in the network. Proper management and control of DERs can improve the voltage unbalance in the distribution system. However, the side effects of providing this ancillary service on DERs must be considered in management decisions. Otherwise, independently owned DERs might not be too interested in participating in the provision of this service in restructured systems. To this end, this paper proposes an unbalanced power market for generating unbalanced power to compensate for voltage unbalance by three-phase inverter-based distributed energy resources (TIDERs) in a competitive environment. To establish this market, a payment function is required based on the imposed costs on TIDERs to generate unbalanced power for voltage unbalance compensation. Therefore, an expected payment function is formulated for unbalanced power generation by TIDERs to cover the costs imposed on them for generating unbalanced power. Accordingly, the voltage unbalanced compensation problem is formulated as a nonlinear programming problem and implemented in the GAMS software environment on a 25-bus test system. The model is solved using the IPOPT solver. Simulation results indicate that if the costs of producing unbalanced power for compensation are not taken into account, independent resources are unwilling to participate due to the costs imposed on them during the compensation process. In contrast, the proposed structure, together with the implemented payment mechanism, provides effective incentives for resources to actively participate in unbalance reduction. The results further demonstrate that applying the proposed framework not only preserves the participation of independent resources but also reduces the total payment made by the system operator at certain unbalance levels compared to the case in which TIDERs generate balanced power. This reduction in payment is attributed to decreased network losses, resulting from the spatial distribution of resources and their role in improving network balance. Hence, due to the competitive structure of this market, the voltage unbalance is mitigated effectively by imposing the lowest cost on the system operator.
A synthesis method of planar antenna arrays for wireless power transmission is introduced. In order to reduce the design complexity of the array antennas, subarray division is used through minimize the difference between the subarrayed excitation and the reference excitation. Also, in order to further improve the beam collection efficiency (BCE) of wireless power transfer systems, the gray wolf optimization method is used to optimize the initial cluster center. The large planar array antennas are divided into several subarrays and the element number as well as the excitation amplitude of each subarray are optimized. Compared with the optimization results of other literature, the proposed method in this paper can obtain higher BCE which shows the effectiveness and efficiency of the proposed method.
This work presents a stratified transactive energy management framework for residential communities that integrates peer-to-peer trading, demand-side management (DSM), and fairness-aware allocation to improve renewable utilization and reduce grid dependence. A real community case study with heterogeneous participants (consumers and prosumers) is evaluated under three operating modes: conventional grid-dependent operation, trading without DSM, and trading with DSM coupled to an enhanced weighted min–max fairness mechanism. Results show progressive improvement from baseline to the proposed configuration, reducing daily grid imports from 160 to 30.5 kWh and grid exports from 30.5 to 5.4 kWh, while increasing solar utilization from 56 to 94
With the large-scale integration of generalized demand-side resources, virtual power plants (VPPs) have become a key means of aggregating heterogeneous resources to participate in power markets. However, existing research faces challenges in unified quantitative evaluation, incentive compatibility, and efficient solution of non-convex models. This paper proposes a VPP aggregation optimization method based on differentiated regulation capabilities. A two-dimensional evaluation model is constructed, incorporating regulation rate per unit time and regulation capacity per unit electricity price, enabling horizontal comparison of wind power, photovoltaics, energy storage, and adjustable loads. Incentive compatibility constraints based on regulation capability ranking are introduced to ensure that resources with better performance obtain higher benefits. A VPP revenue-maximization model is established and solved using the convex–concave procedure (CCP), which avoids linearization errors and guarantees convergence. Case study results show that total VPP revenue increases by 8.74
Amid the current development trend of energy sector toward green future, the increasing integration of renewable energy, advanced grid technologies, and cross-border interconnections has amplified the occurrence and complexity of power system oscillations. Forced oscillation (FO) events, in particular, pose a significant operational challenge because their sustained nature can interact with system modes and propagate across large network areas. Energy-based localization techniques such as dissipating energy flow (DEF) and its projected variants have therefore received considerable attention as measurement-driven diagnostic tools. However, practical applications have shown that purely energy-based indicators can become difficult to interpret under conditions such as resonance masking, weak oscillatory signatures, or ambiguous propagation patterns. Motivated by these challenges, this paper proposes a Multi-Criteria Projected-Complex-Dissipating-Energy-Flow (P-CDEF) framework that extends the classical projected dissipating Energy Flow (PDEF) formulation by integrating complementary diagnostic indicators within a unified scoring architecture. The proposed framework combines temporal injection characteristics, statistical energy descriptors, directional propagation metrics, and modal observability analysis to construct a structured multi-criteria localization strategy. The method is evaluated using the IEEE–NASPI oscillation source localization benchmark Case 7, a diagnostically challenging scenario that has historically exposed limitations of several energy-based approaches. The results demonstrate that the proposed multi-criteria formulation can provide clear source localization signals even in situations where conventional energy-flow indicators alone produce ambiguous rankings. These findings suggest that integrating complementary physical and statistical indicators with the projected energy-flow formulation can improve both the interpretability and robustness of oscillation source diagnostics in modern power system monitoring environments.
