Order Publications

Reference: ISH2017_371

ISH Collection

Recognition of the degradation level of polymeric insulators using infrared radiation, statistical and ann



The insulation of electric power transmission systems is predominantly made using air and electrical insulators. Faults in electrical insulation cause undesirable consequences, being the main cause of power line faults. In their operating environment, external high voltage insulators are subjected to severe conditions. In the case of polymeric insulators, abrupt temperature variations along the core may indicate phenomena such as core milling, erosion and partial discharges caused by manufacturing defects. Many techniques are used in the inspection of insulator. However, there is still no model that defines the objective criteria used to determine the degradation levels of polymeric insulators. The methodology of the present work consists of the analysis of the infrared radiation emitted by tested insulators and the extraction of information obtained from the images. The application of descriptive statistics techniques are used to reduce and model the input data of the classifier in order to use precise parameters and to avoid the use of unnecessary data. The research used a database of thermographic images. For the formation of the database, 230 kV polymer insulators were used in different degradation conditions, submitted to a phase-to-ground voltage of 133 kV, corresponding to 100% of the nominal voltage. In order to classify the insulators according to their degree of criticality, the research used neural networks with the objective of reducing the subjectivity of the process. Neural networks were implemented in a software Matlab® toolbox using standard type configuration. In general, the research tried to structure a neural network and define reliable attributes used as input to the classifier. By modeling and reduction of infrared data was possible to evaluate the level of degradation of insulators with accurate, as well as to relate the results obtained with classification of the possible cause of defects. The performance of the neural network and the classification of the insulators studied presented results with average hits between 84.97% and 90.10%.

File Size: 666,6 KB

Pages NB: 6

Year: 2017

FREE free
Non member
FREE free

Your AccountYour Account

Password forgotten


may download free of charge all publications including most recent ones

Non Members

may download free of charge publications over three years old and purchase all publications

For any questions in connection with the on-line library, contact:


Publications and Edition Manager