Distribution Network Fault Risk Assessment Method Considering Difference in Entropy Value of Rare Factors

External factors such as bad weather and external failure seriously affect the reliability of a distribution network. To comprehensively and accurately assess the risk of faults in a distribution network, this paper proposes a fault risk assessment method that considers the difference in entropy value of rare factors. This method uses K-means clustering algorithm to classify failures based on their consequences and an improved association rule mining algorithm to analyze rare environmental factors and evaluate high-risk and low-probability factors, so as to realize the quantitative analysis of the association between rare factors and risk levels. By combining the Pearson correlation coefficient, the correlation of each environmental feature is analyzed and redundant features at different risk levels are eliminated. Then, the component criticality analysis method is used to adjust the risk weight of rare elements, which can quantitatively measure the degree of correlation between the occurrence of individual elements and the fluctuation of the overall risk of the system. According to the fluctuation difference of different factors, the risk weight optimization matrix considering the difference of rare factors is obtained, and the fault risk assessment model of the distribution network is established, in combination with the information theory. Finally, a distribution network fault risk assessment model is established, of which the accuracy and effectiveness is verified by a case analysis of an actual distribution network in a city.