Spectral Inversion Method Analysis Based On Simulated Annealing

With gas-oil exploration and development of in-depth, Find low-amplitude structure, thin reservoir, subtle reservoirs, carbonate reservoirs and other complex geological, has become the main objective of the exploration and development. Continental oil and gas deposits of China are mostly thin Layer Type, Oil exploration at this stage mainly thin layer of the main reservoir. How to determine the spatial distribution and property of the reservoir has become an urgent problem. Address the situation currently facing, High-resolution seismic exploration technology has become an important research direction.Improve the resolution has been Exploration Geophysicists the permanent goal. This paper summarizes several concepts of resolution predecessors had put forward and analysis of several factors affecting the resolution of seismic. Second, the spectral characteristics of the reflection coefficient discussed from the theoretical. And we establish a wide range of thin layer reflectivity model combining seismic subsidence frequency. Meanwhile, we also analyze their time domain and frequency domain characteristics. Finally, for detecting the location of reflection coefficient, we summed up the law of corresponding through theories and models.Most geophysical inversion belong to the objective function of optimization of more extreme. Making based on linear or quasi linear theory of inversion often leads to inversion into a local optimum. In this paper, we combine some of the latest cutting-edge interdisciplinary research. First of all, we improved the traditional simulated annealing algorithm and combined heuristic and non-heuristic algorithms against each of the characteristics to develop a new nonlinear inversion algorithm-Simulation of Conjugate Gradient Method. It succeed to global optimization of simulated annealing and conjugate gradient method to calculate speed and high precision, can quickly search the global optimal solution. Tests show that, this method is an efficient inverse algorithm, especially suitable for nonlinear, multi-extreme optimization problems and geophysical inverse problems.Time-frequency analysis as a powerful tool for non-stationary signals of time-varying, has become a hot topic of modern signal processing. Seismic signal is a non-stationary signal. Time-frequency analysis providing the joint distribution of the information of time domain and frequency domain allow us to clearly observe the relations the frequency of seismic signal changes with time. In this thesis, we focus our mind on generalized S transform, systematic introduced to the concept and property of time-frequency analysis, introduces several kinds of time-frequency analysis methods and analysis of its characteristics. Given the characteristics of generalized S transform, this article will be used to detect the location of thin-layer reflection coefficient, to provide the better initial model for the spectral inversion.Spectral inversion is the latest pop up a new geophysical inversion. It can be considered a sparse spike inversion. Its main feature is that use only part of the spectrum data for inversing sparse reflection coefficient or thin layer thickness. In this paper, we analyzes briefly the causes of the spectral inversion improving the resolution. The objective function of spectral inversion are derived in detail in this paper. And we also analyzed the ill-posed of the objective function. Through the numerical test discussed the impact of various factors spectral inversion and also verified the capacity of improving resolution. It is proved that spectral inversion is an inversion of sparse, and inversion result can be better in the sparse conditions. Finally, the actual data processing combined with the information of well proved once again the ability of spectral inversion to improve the resolution.