Abstract:  This paper proposes a fast and quantitative reliability evaluation method for medium-voltage distribution network by using data correlation analysis. First, we establish “performance-element” double-deck evaluation system based on distribution network reliability improving methods, which symbolizes the reliable performance and construction status of the system, respectively. Second, through virtual distribution network simulation system we calculate enormous indexes of different kinds of distribution network to provide data support. Last, we establish stepwise regression model to analyze the correlation among indexes, which transforms the multivariate nonlinear regression problem into the multi-step simple regression problem with considering distribution network failure mode, and finally obtains index correlation explicit expression.  The validity of the proposed model has been verified through the reliability evaluation on three test distribution network and one real distribution system. 

Key words:  medium-voltage distribution network, reliability evaluation, enormous data generation, correlation analysis