Improved singular spectrum decomposition method for resonance recognition of air-coupled ultrasonic signals in through-transmission steel plate detection

Abstract Air-coupled ultrasonic resonance detection is an emerging technique that eliminates the need for coupling agents in conventional ultrasonic testing and has a wide range of applications. However, this technique faces challenges, such as weak signal strength and susceptibility to noise, primarily due to the mismatch of acoustic impedance at the air–solid interface and sound attenuation in the air. This study proposed an improved signal processing method based on singular spectrum decomposition (SSD), optimizing the frequency components of each iteration using singular spectrum analysis. Additionally, a weighted Hurst exponent was introduced as a selection criterion to identify useful components and reconstruct the signal. Simulation results demonstrated the effectiveness of this method in denoising components and extracting weak resonance frequency features, especially in low signal-to-noise ratios (SNRs). Compared with conventional SSD and empirical mode decomposition, the proposed method improved the SNR by 39.4% and 301.33%. Furthermore, repeated experimental studies on steel plates with varying air lift-off distances and thicknesses confirmed that the proposed method could effectively reduce noise, thereby enabling accurate measurement of steel plate thickness.