High precision ultrasonic measurement method for thin-walled parts

Abstract

Thin-walled parts are extensively utilized in the aerospace industry, where over thinning defects in these parts would significantly compromise mechanical safety. Ultrasound measurement methods are commonly employed for thickness non-destructive testing. Manual calibration of reference signals is usually necessary, negatively impacting automation, robustness, and accuracy. This paper proposes a high-precision ultrasound amplitude method for thin-walled component inspection, getting rid of manual calibration. Firstly, the measured signal is preprocessed to extract the spectrum in the effective frequency range. Then, the spectrum is approximated using the asymmetrical Gaussian amplitude model to estimate the spectrum of reference signal and reflection coefficient. Moreover, initial propagation time is estimated from autocorrelation of the signal. Finally, the accurate thickness is obtained from the numerical calculation of reflection coefficient with the initial time. Simulations and experiments demonstrate that the proposed method can accurately measures metal plates as thin as 0.5 mm. Compared to the traditional correlation method, it greatly improves measurement accuracy and stability, reducing them to 1/2 and 1/4 of the latter, respectively. Additionally, the proposed method also demonstrates high accuracy for curved thin plates, with an average relative error of only 0.57 %. Furthermore, the method exhibits high levels of automation, robustness, and accuracy, and holds significant potential for thickness detection in thin-walled structural components.