Machine Learning Algorithm Based Condition Monitoring of Electrical Components of Solar PV Plant with Ultraviolet Image Processing Observed Through Unmanned Aerial Vehicle
摘要
The identification of defects at an early stage in the components is a significant task to avoid terrible failures in the process maintenance of an electrical apparatus. Physical examination of electrical power supplied apparatus can be very hazardous owing to the presence of high voltage and placed at a greater height and vision than human intervention. Hence, in this scheme of condition monitoring, present an ultraviolet image based non-invasive imaging system for component health status and visual inspection of electrical components observed through an unmanned ariel vehicle for a small distribute generation plant. The ariel vehicle collects thermal images in the solar panels, fittings, connections, and their accessories to identify the defects. The ultraviolet images taken were processed through different methods of image processing techniques to get the required geometrical and statistical features, thus obtained data are done with latest image extraction techniques. Based on the information of processed image features will be implemented through build classifier algorithm with support vector machine is formed for classification of defected area analyzed with classification algorithm. Hence based on the results of the algorithm, a condition signal for strategy has been generated, based on the generated approach the equipment can be monitored/repaired before the defect in the component can become an unsafe fault. So that the faults can be identified at an early stage and the reliability of the system is increased operational as well as an economic point of view.