| As a nondestructive testing technology, ultrasound is gradually applied to the thickness detection of metal coating. However, the echo generated by coating always overlaps with the interface echo in the ultrasonic blind zone, leading to the difficulty in extracting the information of ultra-thin coating thickness. Therefore, based on the ultrasonic microscopic measurement, an ultrasonic characteristic parameter, spectral sample entropy, was introduced in our measurement. According to the idea of soft-sensing, a mapping relationship was established between the spectral sample entropy and the ultra-thin coating thickness to achieve precise prediction for the coatings thickness of precious metals. Pure silver samples with hundred micron-sized gold coating were used to perform verification experiments. The results showed that the accuracy of the prediction model based on spectral sample entropy could reach 94.7%, better than the predicting outcomes of analysis using traditional ultrasonic feature, indicating this new quantitative characterization method is effective. |
