Intelligent fault detection in power distribution systems using thermos-grams by ensemble classifiers

Abstract

In today’s world, many companies use the thermal imaging (infrared), in order to prevent failures and improve the reliability of the electrical networks. In fact, the technical inspection of the electrical equipment using thermal cameras, is the most effective method for preventive defect detection. This contribution deals with, a systematic method in which, areas suspected of failure, are identified through computer-aided thermal image processing. To this end, the candidate areas are determined, using adaptive threshold and, a number of features are extracted from them. Next, using a genetic algorithm (GA), the irrelevant features are omitted. Finally, by means of a hybrid classifier, the pattern of positive and false positive areas, have been identified. This classifier can also be used as a filter, after extracting the candidate areas. This method is tested on images taken from Tehran northwest substations. As a result, applying the feature selection algorithm leads to a faster intelligent fault detection and higher Reliability, especially in widespread networks, which is known as an effective validation for the proposed method.