A regularization theory-based method for time-frequency analysis and its applications

Time-frequency analysis (TFA) has been widely used in seismic exploration,thus it is crucial to develop a TFA algorithm with high time-frequency resolution.Given the limitations of conventional TFA methods,this study proposed a TFA method based on the regularization theory.The proposed method considers the signal in a short-time window as a superposition of harmonics with different frequencies and takes the TFA problem as an inverse problem.From this perspective,the TFA problem is ill-posed and needs to be solved based on the regularization theory to get a significant time-frequency spectrum.The solution methods under the conditions of L₁ and L₂ norm constraints and the minimum support constraint are commonly used in the regularization theory.This study investigated these solution methods and unified them into the same solution framework.Numerical analysis shows that the TFA method under the condition of the minimum support constraint yielded high time-frequency resolution.This method was systematically applied to the actual data of a specific study area,producing a time-frequency data volume with high time-frequency resolution.Moreover,the planar reservoir distribution was clearly characterized using a single-frequency data volume,demonstrating the promising application prospect of the method.