Probability Analysis of Voltage Collapse Based on Ranking Fuzzy Rough-set Load Clustering

A novel non-mechanistic approach for voltage collapse is proposed. By incorporating the notions of ranking clustering and fuzzy weighting with rough-set clustering, a new algorism for load clustering is presented, and then the probability model of load distribution is constructed. In the basis of fuzzy regression analysis of load improving pattern, the most probable direction of load increasing is calculated by sampling the data of load climbing periods, and the corresponding distance from saddlenode bifurcation point on the singularity boundary to the original operation point is figured out. The conception of voltage stability distance is defined, and by undergoing the most likelihood estimation of the parameters of its distribution function and the Kolmogorov test to goodness-of-fit, the distribution function of voltage stability distance is formed. Accordingly, the technique of probability estimation, considering the rising pattern of active power and reactive power, is advanced. By applying the approach in IEEE RTS and comparing of different clustering methods and diverse probability means of voltage stability, the rationality and applicability of proposed method are attested.