This work proposes an edge-oriented hybrid fault-space restructuring methodology that utilizes UMAP-based embeddings and hierarchical clustering to group overlapping fault conditions into robust hybrid representations, ensuring a highly viable execution on resource-constrained devices once the classifiers are trained.
Abstract
Fault diagnosis in power electronic systems is challenging when fault categories geometrically overlap within the measurement space, limiting class separability and introducing classification ambiguity. This work proposes an edge-oriented hybrid fault-space restructuring methodology that utilizes UMAP-based embeddings and hierarchical clustering to group overlapping fault conditions into robust hybrid representations. Subsequently, supervised machine learning models execute the final classification over this optimized space. Validation was conducted using a large-scale synthetic dataset generated via real-time hardware-in-the-loop (HIL) simulation, evaluating electrical measurements from three-dimensional RMS values to 60-dimensional instantaneous waveforms. Tested with Decision Tree and Random Forest algorithms, the restructuring strategy significantly improves robustness under geometric ambiguity compared to conventional classification without space restructuring. Specifically, low-dimensional measurements achieved F1-score improvements of approximately 72% and 46% for the Decision Tree and Random Forest algorithms, respectively, while high-dimensional measurement configurations still exhibited significant improvements of 36% and 52%. Consequently, these results confirm that the combined restructuring and classification pipeline is highly effective across the analyzed measurement dimensionalities, establishing a dependable cluster-based diagnostic strategy that enhances classification robustness while accepting a trade-off in individual fault-isolation granularity. Finally, hardware deployment experiments on a Raspberry Pi 4 platform demonstrated the feasibility of executing the trained classifiers for real-time inference under constrained computational environments. The experimental evaluation validated real-time execution capabilities, achieving sub-millisecond inference latencies (as low as 0.32 ms), a memory footprint under 0.14 MB, and processing rates exceeding 2600 inferences per second using lightweight Decision Tree classifiers. Ultimately, these findings indicate that the proposed strategy improves fault detection across the evaluated measurement configurations while ensuring a highly viable execution on resource-constrained devices once the classifiers are trained.
A hybrid two-stage machine learning pipeline that decouples detection from classification is proposed, and the direction of the zero-sequence signature is found to be system-dependent, motivating a learned decision boundary in place of a fixed relay threshold.
Sahil Manikshete, A. Gujarathi, Thanh Long Vu et al.· 0 citations
The results show that the proposed Condition Monitoring (CM) approach significantly reduces resource waste and prevents costly downtime, offering a practical and scalable asset management model for industrial applications.
A. Oner, Meral Bayraktar· Italian National Conference...· 0 citations
The increasing integration of Inverter-Based Resources (IBRs) into modern power grids has introduced new challenges for conventional fault-location techniques, particularly for renewable sources such as wind and solar. To address this issue, this paper proposes a robust Machine Learning (ML)-based framework for accurately locating electrical faults in wind farm collector networks. The proposed methodology comprises multiple stages. Initially, 14 different regression models were implemented and evaluated using the Scikit-Learn library to assess their suitability for the task. Subsequently, a comprehensive hyperparameter optimization phase was conducted using Optuna, aiming not only to enhance model accuracy but also to reassess previously under-performing algorithms. The resulting models were then validated under a wide range of operating conditions, including variations in fault resistance, fault location, fault inception angle, and wind farm generation level. By exposing these models to 18,600 distinct fault scenarios, the simulations provide a comprehensive assessment of their generalization capability and demonstrate the effectiveness of the proposed approach. Among all models tested, the Multi-Layer Perceptron Regressor (MLP), Support Vector Regressor (SVR), and Kernel Ridge Regression (KRR) achieved the highest accuracy, with prediction errors not exceeding 2%. Conversely, ensemble-based methods such as Random Forest, AdaBoost, and Gradient Boosting exhibited noticeable limitations, struggling to capture the complex nonlinear relationships between fault characteristics and their corresponding locations in the context of IBRs.
Miguel R. Fonseca, M. Davi, M. Oleskovicz· IEEE Access· 0 citations
This study proposes an artificial neural network-based approach for the detection and classification of faults occurring in high voltage alternating current (HVAC) power transmission lines. The study considers 12 classes comprising 11 fault types and one healthy state. Unlike traditional approaches that rely on extensive feature-extraction procedures, this study directly employs measured three-phase voltage, current, and phase-angle quantities as ANN inputs, thereby avoiding computationally intensive signal decomposition and handcrafted feature extraction stages. The model was evaluated using regression-oriented metrics, including mean squared error (MSE) and correlation coefficient (R). Furthermore, 5-fold cross-validation showed that the proposed ANN achieved better regression performance than GPR, SVR, and Kernel Regression models. Additional robustness analyses performed under different loading conditions and fault resistance values further demonstrated the generalization capability of the proposed ANN models under varying operating conditions. To evaluate the practical contribution of phase-angle information, a classification-based ablation study compared a 6-input ANN using only three-phase voltage and current measurements with a 12-input ANN including phase-angle measurements. Under identical test conditions, the 6-input and 12-input classifiers achieved accuracies of 88.51% and 84.73%, respectively, with macro F1-scores of 0.8789 and 0.8374. Repeated-training analysis further showed that the six-input configuration achieved higher mean performance and lower variability. The results indicate that phase-angle information provides supplementary and class dependent discriminative value, but does not consistently improve all fault classes. Conventional voltage and current measurements alone therefore represent a simpler and more stable alternative, whereas phase-angle measurements may be incorporated when synchronized phasor information is already available.
Zeynep Bala Duranay, İsmail Anıl Avcı, Mohammed Bushra Mohammed et al.· Symmetry· 0 citations
The integration of inverter-based renewable energy sources into electric grids challenges conventional power system protection. Machine learning-based solutions can address these challenges by utilizing available data in modern smart grids. However, the lack of open datasets prevents reproducibility and fair comparisons between different approaches and their results, which hinders further progress. Therefore, this paper presents EvEMTBench, a synthetic dataset of faults and events generated using electromagnetic transient simulations. The physical plausibility of the power system simulation is ensured by validating the simulation parameters against established literature and providing comprehensive documentation of the simulation procedure. The dataset is designed for training, fine-tuning, and benchmarking machine learning models by providing synchronized point-on-wave voltage and current measurements at 9600 Hz across a diverse set of topologies and voltage levels. The inclusion of a wide range of fault and operating events allows the utilization of EvEMTBench for different tasks like incipient fault detection, fault localization, or event detection.
Georg Kordowich, Jonathan Loebel, J. Oelhaf et al.· 0 citations