Time Series Shapelet Classification Based Online Short-Term Voltage Stability Assessment

摘要

This paper describes an online short-term voltage stability assessment scheme from an overall view of the load area in the power system. In this scheme, data acquisitions are completed by post-contingency phasor measurements and a time series shapelet classification method is employed for classification learning. Combined with decision trees, this novel approach can not only hold a high performance of classification but also offer an acceptable interpretation of classification results. An improved algorithm to speed up shapelet searching is proposed, which makes it more practical. Semi-supervised cluster learning is also adopted in this scheme to mitigate the unreliability of the previous practical criteria. The test results on the Nordic test system demonstrate the effectiveness and reliability of the proposed scheme.