Research on Fault Scenario Modeling and Planning Technology for Distribution Networks Based on Monte Carlo and Clustering Methods

To support resilience planning for distribution networks with high proportions of renewable energy integration, this study addresses the problem that existing fault scenario modeling methods fail to accurately capture the spatial-temporal correlation of faults and generate insufficiently representative fault scenarios. A hybrid modeling method that integrates meteorological and equipment status data is proposed. This method constructs a spatiotemporal fault rate model based on a non-homogeneous Poisson process, combining basic fault rates, meteorological sensitivity increments, and equipment aging increments, and generates massive random fault sequences through Monte Carlo simulations. Furthermore, an adaptive density peak clustering (CFSFDP) algorithm based on K-nearest neighbors is introduced to automatically extract typical fault patterns and build a probabilistic fault scenario library. Simulations based on the improved IEEE 33-node system show that the generated fault scenario library exhibits good validity and robustness. Stress testing of planning schemes based on this scenario library indicates that compared to the traditional “$\mathrm{N}-1$” scheme, the enhanced scheme incorporating distributed photovoltaic and energy storage significantly improves system resilience. In meteorological-dominated and extreme clustering fault scenarios, the system's average outage duration index (SAIDI) decreases by 42.7 % and 58.3 %, respectively, while the expected energy not supplied (EENS) decreases by 52.1 % in agingdominated scenarios. The study demonstrates that this method can generate fault scenarios that closely resemble real-world conditions with clear physical meaning, providing a quantitative analysis tool for detailed distribution network resilience planning.