A Study on Building Personalized Information Profiling for Distribution Network Faults

Constructing personalized information profiles for distribution network faults across the diverse sub-regions of Guizhou Province, characterized by its mountainous terrain, complex topography, and variable weather, is of significant practical importance. However, leveraging local on-site monitoring data for this purpose poses considerable challenges. This study proposes an innovative three-level hierarchical regional division method integrated with the K-prototypes clustering algorithm, successfully constructing a personalized information profile for distribution network faults. Results show three distribution network fault clusters exist in Guizhou Province. Class A (0.52 indeterminate failures, 34,581.26 customer outage hours, 12.70-hour duration, 6.24-year line lifespan) needs prioritized fault investigation and network optimization. Class B (0.08 unknown failures, 1,178.95 outage hours, 2.37-hour duration, 8.61-year lifespan, 0.66 grid loop rate) requires monitoring for rising failures. Class C (0.10 moderate unknown failures, 1,563.07 outage hours, 2.35-hour duration, 10.76-year lifespan, 0.68 grid loop rate, 0.50 customer reporting rate) needs focus on high reporting rates and network upgrades.