Intelligent identification of flicker source in distribution systems

The authors propose a new intelligent method to identify the source of flicker in distribution systems. By measuring and sampling the bus voltages, a flicker index is extracted using multi-resolution S-transform. Time-frequency resolution characteristic of S-transform is shown to be a suitable candidate for assessment and identification of flicker source. The new flicker index is selected from complex S-transform matrix, which takes into account both amplitude and frequency of flicker. An multi level percepteron (MLP) neural network as a powerful classifier is trained with the bus flicker indices to detect existence of flicker source in contaminating bus. The source of flicker is modelled with an amplitude modulation scheme, and both flickers without harmonics and with harmonics are considered. A 14-bus distribution system was simulated in ATP/EMTP to test the method. Results show that by measuring voltages in proper buses, contaminating bus or zone can be detected using a few measuring points.