Feature Selection Method Based on Overlapped Probability for Transient Voltage Stability Assessment

In accordance with the high dimensional characteristics of power system voltage stability, this paper proposes a kind of feature selection method for voltage stability analysis, which is based on the principle of minimum probability of categories´ overlapped area. The key of the algorithm is determining the criterion, which is based on the probability of samples falling in the categories´ overlapped area. The algorithm is simple and intuitive. An original feature set of transient voltage stability assessment has been set up. After the feature selection step, it can be taken as the input of decision tree. The 10-fold cross-validation is used for the validation. Two case studies show the advantage of the overlapped probability criterion when applying in the transient voltage stability issues.