The Feature Extraction Technology Of Ultrasonic Signals In Thickness Measurement Based On Wavelet Analysis

Due to the existing regular texture on the cell surface of solid rocket motors, the noise in ultrasonic echo signals ,which has bad influence on extraction of characteristic signals, is inevitable in the process of ultrasonic thickness measurement of heat insulating layer in spining case. In this thesis, the feature extraction technology of ultrasonic signals in thickness measurement based on wavelet analysis is studied to realize accurate thickness measurement of heat insulating layer in spining case.According to the characteristics of the echo signals from interface between heat insulating layer and motor case, a new method based on Daubechies wavelets, Symlets wavelets and wavelet multi-resolution decomposition has been proposed to realize the feature extraction of ultrasonic signals. The results demonstrate that the method based on Daubechies wavelet is effective in de-noising the noise in ultrasonic signals, and accurately identifing the characteristic signals from the first interface and the second interface where the thickness of heat insulating layer is more than 1.7mm. The method based on wavelet multiresolution decompositioncan is effective in extracting the echo signals from the second interface where the thickness of heat insulating layer is less than 1.7 mmAccording to the characteristic signals from the first interface and the second interface, the thickness of the heat insulating layer can be calculated. Comparing between the calculated thickness and the actual thickness, the absolute error is less than 0.03mm, the relative error is less than 1.97%. All the methods metioned above are proved to be effective in extracting the characteristic signals in thickness measurement of heat insulating layer in spining case.,which offer a new idea for extraction of ultrasonic signals in thickness measurement of such heat insulating layer in spining case.