Estimation of the critical clearing time using MLP and RBF neural networks

摘要

This paper presents multi-layer perceptron (MLP) and radial basis function (RBF) neural networks (NNs) based methods for the estimation of the critical clearing time (t_(cr)) as an index for power systems transient stability analysis (TSA). The t_(cr) evaluation involves elaborate computations that often include time-consuming solutions of nonlinear on-fault and post-fault systems equations. Knowing that for a particular fault scenario (contingency), the t_(cr) is a function of the pre-fault system operating point, the objective of this paper is to show how one may develop the MLP and the RBF NNs based methods for estimating the t_(cr) by using only the pre-fault operating conditions as the inputs of the NNs. The paper uses the proposed MLP and RBF NNs based methods to estimate the t_(cr) under different topological as well as operating conditions of the 10-machane 39-bus New England test power system, and results are given. The simulation results show that both NNs are able to retain past learned information almost instantaneously. However, compared to the RBF NN, the MLP NN makes us have a more accurate estimation for the t(cr).