Intelligent prediction of out‐of‐step condition on synchronous generators because of transient instability crisis

摘要

This paper presents an adaptive scheme for predicting out-of-step (OOS) condition of synchronous generator based on the Bayesian technique. The proposed scheme performs as an intelligent OOS method for synchronous generators from which by using training variables, the tripping signals are estimated. For classifying target classes between stable and OOS conditions, a series of measurements are derived under various fault scenarios including topological and operational disturbances. The tripping signals are estimated by using feature selection technique based on the Bayesian technique. In this procedure, the data of input variables and corresponding output target classes are implemented as input-output pair data for Bayesian training and testing. For this propose, the ability of the OOS protective scheme is examined for a number of unseen samples in working mode. The proposed approach is applied on IEEE 39-bus test system from which by using trained variables, the tripping signals are estimated online. Furthermore, to evaluate the proposed protective scheme in real-time environment, a 2-machine experimental case is used to assess the effectiveness of the proposed scheme. The results show a promising performance of proposed protective scheme for proper estimating of tripping signals.