Ying Hu

Accurately characterizing complex and unconventional reservoirs is crucial for improving oil and gas recovery efficiency and reducing exploration and development risks. It is feasible to precisely interpret stratigraphic information using time-frequency (TF) characteristics derived from seismic signals. A general chirplet extracting transform (GCET) was developed for stratigraphic interpretation of reservoirs. First, two instantaneous frequency (IF) equations were constructed within the general chirplet domain. From these equations, two distinct extracting operators were derived: one designed for harmonic-like features, and the other for impulse-like features. Each operator selectively extracts TF coefficients that satisfy its corresponding IF equation. Additionally, an adaptive fusion strategy was introduced, which used local sparsity measures to combine the extracting results, thereby enabling the dynamic construction of the GCET. The proposed GCET method simultaneously generated TF representations for harmonic and transient modes, and effectively characterized the TF features of seismic signals with superior noise resistance. Tests on synthetic and field seismic data demonstrated the efficacy of the proposed method for reservoir stratigraphic interpretation.

Carbonate reservoirs exhibit complex internal structures and strong heterogeneity. Their seismic responses commonly show multicomponent superposition and pronounced nonstationary characteristics, which readily induce inter-component interference in time-frequency (TF) representations and pose challenges to the characterization of reflection interfaces. To address this issue, we propose a statistical horizontal squeezing transform (SHST) that reformulates group delay (GD) estimation from a pointwise derivative operation to a statistical estimation problem within local TF neighborhoods. Specifically, a statistical group delay estimator (SGDE) is constructed by introducing a weighted short-time Fourier transform (STFT) spectrum, allowing GD to be inferred through local TF statistical consistency rather than single-point gradients. This strategy enhances robustness against component interference and stabilizes energy reassignment. Experiments on synthetic signals and simulated geological models demonstrate that SHST achieves superior TF concentration and robustness under conditions of closely spaced multicomponent signals and noise. Applications to real deep carbonate seismic data further verify the effectiveness of the proposed method in characterizing the time-depth distribution of reflection interfaces, providing a reliable TF analysis tool for seismic interpretation of complex carbonate reservoirs.

Time-frequency (TF) analysis (TFA) is a powerful tool for handling bearing vibration signals. The polynomial chirplet transform (PCT) can effectively highlight the time-varying characteristics of signals. However, PCT is limited by the uncertainty principle, resulting in lower TF resolution. At the same time, as a parameterized TFA method, PCT is unable to address multi-component signals. Therefore, this paper proposes a novel TFA method called polynomial-demodulated extracting transform (PDET). In this method, the instantaneous frequency (IF) trajectory of the signal is estimated by building the local maximum principle. Then, the TF coefficients are redistributed onto the retrieved IF, yielding satisfactory TFR and accurate IF estimation. Additionally, this study builds a separation strategy for multi-component signals by introducing binary graphs and connectivity structures from image processing technology, which solves the limitation of PDET in handling multi-component signals. The effectiveness of the proposed method in highlighting the time-varying features and noise robustness is verified by the simulated signals compared with advanced TFA methods. Finally, we apply PDET to real vibration data to further demonstrate the potential of PDET for bearing fault diagnosis.

Deep carbonate sedimentary cycle division is a core geological theory for analyzing sedimentary environments and provides important guidance for predicting the distribution of high-quality reservoirs. As a key technique, time-frequency (TF) analysis primarily focuses on the high-resolution characterization of dominant frequency variations caused by changes in stratum thickness. To enhance TF resolution, this study proposes a matching extracting transform (MET) to better characterize sedimentary cycles. Initially, the method constructs a matching chirp rate estimator (MC), leveraging the relationship between the instantaneous frequency (IF) estimator and the group delay estimator within the short-time Fourier transform domain, to capture the slope of the IF trajectory. Subsequently, based on the properties of the MC, a chirp rate valid domain (CVD) for the MC is defined to constrain its computational interval. Within this CVD, a maximum criterion under the chirp transform is then employed to extract the signal’s IF trajectory accurately. Finally, the TF coefficients are reassigned along the retrieved IF trajectory, enhancing TF energy concentration while maintaining signal reconstruction fidelity, as validated numerically. Applications to both simulated sedimentary cycle models and field deep carbonate seismic data demonstrate that the proposed MET is a potential tool for sedimentary cycle division.