emt fault simulation

This work addresses the lack of standardized, publicly available fault waveform datasets in high-voltage protection research, which hinders reproducible evaluation of data-driven methods. To bridge this gap, the authors introduce PROTECT-90, a benchmark dataset generated through electromagnetic transient (EMT) simulations on a standard 90 kV double-circuit transmission line topology. By systematically randomizing grid operating points, line parameters, and fault conditions, they produce 9,022 physically consistent and scenariowise balanced short-circuit simulation samples. Three-phase voltage and current waveforms, along with structured metadata, are synchronously recorded at eight measurement points. PROTECT-90 represents the first openly shared high-voltage fault dataset, with fully documented modeling assumptions and generation procedures, thereby establishing a standardized and reproducible benchmark for evaluating signal processing techniques and learning-based protection algorithms.

本文评估了多种机器学习方法在电力系统中故障检测和线路识别的有效性，旨在解决传统保护系统面对复杂电网挑战时的不足。

Accurately simulating lightning electromagnetic field (LEMF) propagation remains challenging due to the broadband, transient nature of lightning radiation and the sensitivity of finite-difference time-domain (FDTD) solvers to numerical dispersion, boundary reflections, and ground conductivity modeling. Method: This study systematically evaluates the applicability and accuracy of three open-source FDTD solvers—Elecode, gprMax, and MEEP—for LEMF simulation. Using a unified modeling framework—including perfectly matched layers (PML), dispersion control, and consistent material parameterization—we conduct comparative simulations under ideal and lossy ground conditions for canonical lightning radiation scenarios, validating results quantitatively against analytical solutions or high-fidelity reference data. Contribution/Results: We identify critical impacts of spatial discretization, grid resolution, and PML configuration on far-field waveform fidelity, and characterize typical failure modes arising from suboptimal parameter choices. The work delivers a reproducible benchmarking protocol and an open-source script library, providing practitioners with rigorous, evidence-based guidance for selecting and configuring FDTD tools in lightning electromagnetic compatibility analysis and protection design.

Traditional fixed-threshold protection schemes exhibit insufficient reliability in dynamic short-circuit fault identification and localization under high penetration of distributed energy resources. Method: This study conducts the first systematic, comprehensive benchmarking of classical machine learning models for real-time power grid protection. Using electromagnetic transient simulation data, voltage and current waveform features are extracted via 10–50 ms sliding time windows; model performance is rigorously evaluated across fault classification (F1-score) and fault distance estimation (R²) tasks, with emphasis on accuracy, robustness, and real-time inference latency. Contribution/Results: The best-performing classifier achieves an F1-score of 0.992 ± 0.001; the top regressor attains an R² of 0.806 ± 0.008. Average inference time per sample is only 0.563 ms—well within the millisecond-level response requirement for protective relaying. This work establishes the first unified, reproducible, real-time-constrained benchmark for ML-based protection models, addressing a critical gap in the literature.

High computational cost and poor parallel scalability hinder time-domain simulations of low-frequency electromagnetic eddy current problems. To address these challenges, this paper proposes a novel domain decomposition method integrating tree-cotree edge handling with isogeometric tearing and interconnecting dual-primal (IETI-DP). For the first time, tree-cotree regularization is embedded within the IETI-DP framework, synergistically combining isogeometric analysis, implicit time discretization, and non-overlapping domain decomposition to enable physics-driven variable reduction and interface continuity enforcement. Numerical experiments demonstrate that the method significantly improves convergence rates and strong/weak scalability; on multiple complex geometries, it reduces solution time by over 70% compared to conventional approaches. The proposed framework establishes a new paradigm for large-scale transient eddy current simulation—achieving high accuracy, numerical robustness, and superior parallel efficiency.

This study addresses the lack of a unified framework for assessing the economic impacts of multiple hazards on the U.S. high-voltage transmission network, as existing risk assessments typically analyze disasters in isolation. The authors develop the first multi-hazard risk assessment system that integrates hazard characteristics, grid vulnerability, and macroeconomic propagation, covering nine individual hazards and a compound ice storm–high wind scenario, with space weather incorporated for the first time on an equal quantitative footing with terrestrial hazards. Leveraging national disaster data, a detailed power grid model comprising over 13,000 transmission lines and 10,000 substations, probabilistic failure analysis, and input-output modeling, the study quantifies daily expected losses: tropical cyclones cause the highest direct damage ($137 million/day), while tornadoes trigger the largest downstream economic losses ($4.93 billion/day). The compound event affects 237 million people and results in $85.16 billion/day in lost economic output.

This study addresses the absence of application-level benchmarks with defined temporal and sensing assumptions in power system protection, which hinders evaluating the real-world utility of learning-based models. Building upon the PROTECT-90 dataset, this work proposes the first application benchmark for fault classification and line identification, employing a rigorous disjoint-split strategy to systematically evaluate 1D CNNs and MLPs across varying observation windows. Results demonstrate near-saturated accuracy under full observability. Notably, using current-only inputs maintains 100% line identification accuracy, whereas voltage-only inputs degrade performance to approximately 53%, with CPU inference requiring merely 0.528 ms. By quantifying the impact of sensor deficiencies on model precision, this research confirms that the fundamental bottleneck in such systems lies in measurement information rather than network architecture.

This study addresses the limited engineering applicability of machine learning surrogate models in power system dynamic simulation due to their lack of physical interpretability. For the first time, neural tangent kernel (NTK) theory is introduced into this domain, integrated with small-signal eigenvalue analysis to establish a formal connection between system physical stiffness and neural network optimization stiffness. This linkage reveals the evolution mechanism of error modes during training, enabling the development of an adaptive loss weighting strategy. The proposed approach not only provides modal-level physical explanations for performance differences across network architectures—such as ActNet—but also significantly enhances training convergence and reliability. Consequently, this work lays a theoretical foundation for designing interpretable, structure-aware surrogate models tailored to power system dynamics.

为解决高压输电网络中故障检测与分类问题，提出了一种两阶段混合机器学习流水线，通过结合异常检测和监督分类提高准确性。

This work addresses the spectral degradation and poor solver efficiency arising from discretized Laplacian operators in eddy current problems coupled with external circuits. To overcome these challenges, the authors propose an electromagnetic decoupling (EMD) preconditioner that physically separates the system into vector and scalar potential components, each treated with a tailored preconditioning strategy. The approach synergistically combines algebraic multigrid, domain decomposition, and direct solver techniques, enabling parallel computation across multiple excited conductor domains and seamless integration with mainstream solver frameworks. Numerical experiments demonstrate that, compared to conventional incomplete Cholesky preconditioning, the proposed method reduces iteration counts by up to a factor of 20 and accelerates solution times by as much as 20-fold.