Heliyon (Oct 2023)
Ultrasonic visualization and quantitative analysis of internal defects in RTV coatings
Abstract
Room temperature vulcanised (RTV) silicone rubber coatings effectively enhance the insulation properties of electrical equipment. However, RTV coatings are prone to internal defects caused by the coating process and the effects of aging during service, which can lead to debonding of the coatings. Internal debonding defects are challenging to detect and can ultimately lead to accidents due to a reduction in the insulation capacity of the equipment. To visualize the internal defect morphology of RTV coatings and quantify the defect size, an ultrasonic pulse-echo-based method for detecting and imaging debonding defects is proposed. The method involves the development of a finite element model to investigate how ultrasonic waves propagate in RTV coatings and the influence of ultrasonic probes and inspection conditions on defect echoes. Furthermore, an ultrasonic detection system specifically designed for RTV coating debonding defects is constructed. This system utilizes wavelet packets in the time-frequency domain to analyze the echo signals in both normal and defective regions. The three-dimensional reconstruction of the debonding defect morphology is accomplished by integrating ultrasonic echo amplitude and position information. Finally, the size of the debonding defects is quantified using an adaptive threshold segmentation method. The findings indicate that ultrasound waves reflected in RTV materials propagate as spherical waves, with the acoustic energy primarily concentrated near the acoustic axis. As the propagation distance increases, the sound beam disperses along the axis and extends beyond the transducer, resulting in a decrease in the sound field's directionality. The developed visual reconstruction method in this study offers the capability of three-dimensional visualization for defects present within RTV coatings, including their length, width, and depth. The accurate determination of defect size is achieved through the utilization of the adaptive threshold segmentation method, yielding an average error rate of 5.7 % across different defect types. In comparison, the maximal interclass variance method (OTSU) and the fuzzy C-means (FCM) method produced results with error rates of 9.8 % and 7.9 %, respectively. The research presented in this paper enables precise assessment of debonding defect severity and establishes a reliable foundation for on-site inspection, operation, and maintenance of RTV coatings.
