In order to analyze and identify the corrosion of the inner interface of the coupling, perforating and casing, or the collapse and depression of the casing, as well as the reflection form of the dynamic liquid level, the COMSOL finite element analysis software was used to carry out numerical simulation research on the above situation, and its response characteristics were obtained through calculation. In this paper, three-dimensional propagation characteristics of acoustic waves in the annulus of the oil jacket were studied using signal sources of different polarities. The results showed that when the polarity of the signal source was positive (negative), the coupling reflection signal was first positive (negative) and then negative (positive), and the perforation reflection signal was first negative and then positive (positive then negative). The waveform of the reflection signal in the case of corrosion or casing collapse was consistent with the waveform of the perforation reflection signal; the reflection signal of the moving surface was of the same polarity as the emission signal source. When the signal source was a positive-negative superimposed source, the conclusion obtained by applying the above-mentioned positive (negative) polarized signal source was based on the polarity of the first peak of the superposed source. The research results can provide strong support for the popularization and improvement of related theories, and correct reference standards for the performance verification of echo processing methods, as well as important theoretical basis for the correct identification of echo positions on the Dynamic fluid interface.

In practice, there are various noises included in the collected seismic data. As a common background noise, random noise may interfere even annihilate valid signals. In view of this, it needs to adopt effective methods to eliminate random noise as much as possible. This paper proposes a filtering method that is based on pseudo Wigner-Ville distribution (PWVD) and Catte model. PWVD is a windowed form of Wigner-Ville distribution (WVD) which can make the nonlinear signal to be locally linearized in addition to weakening the influence of cross item. And the Catte model is a kind of anisotropic diffusion algorithm which can eliminate the noise while protecting the edge of image. The implementation of the proposed method is to do PWVD for the frequency modulation form of noise-containing signal and then to filter the PWVD via Catte model, at last to adopt peak search to obtain the estimation of the valid signal.

To address the difficulty in identifying liquid-level reflection points arising from the susceptibility of dynamic liquid-level echoes to interference from downhole operating conditions during drilling, this study proposes a method for identifying and extracting dynamic liquid-level signals in the time–frequency (T–F) plane using the synchrosqueezing generalized phase-shifting S-transform (SS-GPST). The method introduces adjustable parameters to regulate T–F resolution and phase-shift correction to improve phase consistency between the S-transform and synchrosqueezing reassignment, thereby obtaining high-resolution T–F distributions of dynamic liquid-level signals. On this basis, two-dimensional Gaussian filtering is applied in the T–F plane to smooth and suppress background noise. The T–F region containing the liquid-level reflection signal is then extracted for inverse-transform reconstruction, and the liquid-level reflection point and its occurrence time are determined from the position at which a pronounced abrupt change first occurs in the reconstructed waveform. The results show that, compared with the T–F distributions obtained using the synchrosqueezing S-transform (SS-ST), synchrosqueezing phase-shifting S-transform (SS-PST), and synchrosqueezing generalized S-transform (SS-GST), the SS-GPST achieves higher T–F resolution and preserves more local details. The measurement error of the SS-GPST does not exceed 1.2 m, and its relative error remains within 0.083%, confirming its ability to identify and extract dynamic liquid-level signals more accurately under complex interference conditions.