GridWaveLoc: A Fault Location Algorithm Integrating Fault Traveling Wave Distribution and Network Dependency Graphs for Transmission Grids

Abstract

Accurate fault localization in transmission grids is essential for reducing downtime and maintaining power system stability. Conventional fault location techniques frequently struggle with the intricacy of contemporary transmission grids, which have intricated dependencies and variable fault characteristics. Problem Statement: Previous fault location techniques used limited indicators, like wave arrival times or impedance-based measurements, while ignoring amplitude variations and propagation speed. Numerous also use oversimplified network models that assume a uniform topology and ignore grid node dependencies. These constraints result in delays, localization errors, and ineffective grid restoration. The difficulty is to combine fault wave propagation and a realistic network structure to enhance accuracy and response time. Objective: The objective of this research is to enhance fault localization accuracy and response time in transmission grids by using fault traveling wave distribution and network dependency graphs. To accomplish this, the research creates GridWaveLoc, a fault location algorithm that incorporates wave propagation characteristics into the grid's dependency structure, resulting in quicker fault detection and increased grid reliability. Methodology: The GridWaveLoc algorithm executes realtime data such as fault type, wave arrival time, amplitude, transmission line characteristics, and network load. Mean and mode imputation for missing values, label encoding for categorical variables, and MinMax normalization for continuous features all fall under data preprocessing. The algorithm uses fault wave propagation times and network dependency graphs to narrow down possible fault locations. The Euclidean distance technique is employed to detect the nearest grid node to the fault origin, guaranteeing accurate fault location prediction. Results: Experiments were performed on the Transmission Fault Localization Dataset, which contained 11 features and 2000 records, to evaluate three types of faults: short-circuit, open-circuit, and ground faults. GridWaveLoc obtains 98.5% accuracy, surpassing the Traveling Wave Method (92.3%), the Impedance-Based Method (89.5%), and Artificial Neural Networks (85.7%). GridWaveLoc also has the lowest mean absolute error (MAE) of 0.12 km and root mean square error (RMSE) of 0.15 km, substantially enhancing fault localization precision compared to other techniques. These results emphasize the potential for real-time fault detection in massive transmission networks. Conclusion: The GridWaveLoc employs a novel approach to fault location in transmission grids by integrating fault traveling wave evaluation and network dependency data. This technique improves the dependability and effectiveness of power grid functions by offering a solid solution for real-time fault localization.