A Deep Learning-Based Approach for Comprehensive Rotor Angle Stability ‎Assessment ‎

Unlike other rotor angle stability assessment methods which only deal with either transient or small-signal (SS) stability, in this paper, a new stability prediction approach has been proposed which considers both transient and SS stability status. Therefore, the proposed method, which utilizes Multi-Layer Perceptron-based deep learning model, can comprehensively predict the post-disturbance rotor angle stability. Since the proposed method uses the voltage of the generating units directly measured by WAMS in the early moments after the disturbance occurrence and does not need to calculate the generators rotor angle (which requires a high computational burden), it can timely predict the stability stiffness using data provided by PMUs installed at generators buses. In this respect, this method provides a proper chance for the system operators to take appropriate corrective measures. To evaluate the proposed method s efficiency, it has been implemented and tested on IEEE14-bus and IEEE 39-bus test systems. The dynamic simulation results show that although the proposed method requires fewer PMUs than previous methods that exist in the literature, it can timely evaluate the stability status. Also, to properly show the power system stability stiffness from the transient and SS stability point of view, the suggested method accurately classifies the post-disturbance operating point into Unstable, Alarm, or Normal categories.