New Data Driven Scheme for Real-Time Power System Transient Stability Assessment

A computational efficient and precise approach for real-time power system transient stability assessment (TSA), with post disturbance synchronized data is proposed in this paper. An empirical wavelet transform (EWT), Teager Kaiser energy changes (TKE) and Random Forest (RF) based new scheme is proposed to assess TSA. Based on the generator bus frequency and angles, a novel Robust Transient Stability Index (RTSI) is proposed to capture real-time status of power network. The EWT is utilized to decompose the proposed multi-component index into various intrinsic mode functions (modes). Subsequently, instantaneous energy change within each mode is calculated using Teager–Kaiser energy operator (TKEO). The energy changes in each mode defines decision boundary between transient stable and unstable operating scenario. Finally, for each mode, standard deviation of these energy changes is obtained to measure the variations in energy change about mean value. The Random Forest classification scheme is developed to obtain the TSA in real-time. The developed classifier predicts TSA using standard deviation of these energy changes. Simulation results and comparison with relevance minimum redundancy arithmetic, deep imbalanced learning framework and improved support vector machine methods demonstrates capability of the proposed scheme.