Fault Pattern Dataset for a 33 kV Power Distribution Network

This dataset contains historical fault records collected from a 33 kV power distribution network situated within the 104101–104291 zip code region of Lagos State, in the southwest of Nigeria (representing an area characterized by rapid residential and industrial growth) over a 7.58-year period (April 2017 to December 2024). It consists of 23 medium-voltage (33 kV) feeders. Of the 23 feeders, seven supply power to the primary 33 kV/11 kV injection substations that serve as the main distribution nodes, while the remaining sixteen are dedicated to supplying 33 kV/415 V distribution transformers for localized load centers. The data was compiled as part of a study on fault pattern analysis and predictive maintenance using association rule mining. The dataset includes 5,955 fault transaction entries and 7,058 fault occurrences, with detailed attributes such as:

Fault type and cause
Load loss (in MW)
Time of fault occurrence and restoration
Mean Time to Restore (MTTR)
Seasonal classification (dry/rainy)

Derived features include categorized MTTR levels (optimal, acceptable, suboptimal), load loss severity (minimum, moderate, high), and seasonal variations. The dataset supports research in predictive maintenance, reliability analysis, and data-driven fault diagnostics in power distribution systems.

This dataset was used in the paper titled "Fault Pattern Analysis for Enhancing Predictive Maintenance in Power Distribution Systems using Apriori Association Rule Mining".