Fast Transient Stability Batch Assessment Using Cascaded Convolutional Neural Networks

Abstract

Stability assessment is a regular routine in both power system planning and operation. It is computationally demanding using state-of-the-art approaches, such as time-domain simulation (TDS), as numerous scenarios (e.g., N-1 contingencies) have to be assessed for a large batch of contingencies. The purpose of this paper is to determine stability conclusion before the simulator reaches the end of simulating time windows. Therefore, TDS can be terminated early so as to reduce the average simulation time for batch assessment without losses of accuracy. The idea is to develop a data-driven methodology to “learn” from existing TDS results in the batch and “infer” stability conclusions using current available TDS output. To achieve this goal, cascaded convolutional neural networks are designed to capture data from different TDS time intervals, extract features, predict stability probability, and determine TDS termination. While accumulating more knowledge in batch processing, early termination criterion is refreshed continuously via feedback learning to terminate TDS increasingly earlier, with the increase of existing TDS results in the batch. Case studies in IEEE 39-bus and Polish 2383-bus system illustrate effectiveness of the proposed method.