Contingency screening for steady-state security analysis by using FFT and artificial neural networks

A new approach based on artificial neural networks (ANNs) combined with fast Fourier transform (FFT) is developed for single line contingency screening in steady-state security analysis. The offline fast decoupled load flow calculations are adopted to construct two kinds of performance indices, PI_p (active power performance index) and PI_v (reactive power performance index) which reflect the severity degree of contingencies. The results from offline calculations of the load flow are used to train a multilayered artificial neural network for estimating the performance indices. FFT is used for preprocessing the inputs to improve and speed up the ANN training. The effectiveness of the proposed method is demonstrated by contingency ranking on two IEEE test systems and comparisons are made with the traditional method. Good calculation accuracy, high contingency capturing rate and faster analysis times for contingency screening are obtained by using the ANNs