A Data Driven Knowledge Acquisition Method and Its Application in Power System Dynamic Stability Assessment

In this paper, a novel data driven knowledge extraction scheme is proposed and applied to realize power system stability estimation since power system stability assessment can be treated as a typical classification problem (stable/unstable). The strategy is composed of three cascading layers, including the feature selection for choosing an optimal subset from candidate inputs, pattern discovery layer for identifying the latent structure of samples in the selected feature space, and the decision tree layer for generating the self-contained production rules based on the pattern discovery results. Application results on IEEE test system show its merits as a knowledge extraction method, thereby the proposed approach can be widely used in other engineering domains.