Pattern recognition adaptive full waveform inversion method

Abstract

The invention discloses a pattern recognition adaptive full waveform inversion method. The method comprises the following steps: 1, completion of each shot gather record comprises the following sub steps: 1-1, the positive propagation wave field at each moment is calculated; 1-2, an adjoint source deltas is calculated and is backpropagated to obtain back propagation data; and 1-3, the correlationbetween the back propagation wave field and the positive propagation wave field is calculated to obtain the gradient of the single shot; 2, gradient superposition is solved from all shots, and a global gradient of the model space is thus obtained; and 3, the modification amount of a speed model is obtained through a steepest descent method or a local inversion algorithm, and an optimized speed model is further obtained. The method disclosed in the invention has no limitation of only using the low frequency component of seismic data, can use the intermediate frequency and high frequency components of the seismic record, such as a frequency component in a main frequency range, and does not have the defect of wrong solution caused by cycle jump and the like by the conventional full waveform inversion technique.

Description

technical field [0001] The invention relates to a pattern recognition self-adaptive full waveform inversion method, which belongs to the application field of oil and gas exploration. Background technique [0002] Oil and gas resources are not only of great significance to the country's economy but also to the country's strategic significance. my country is a country with a large demand for energy, and spends a lot of money to buy foreign oil every year. In addition to causing huge economic losses to the country, it also poses a huge challenge to the country's energy strategy. How to find oil and gas resources buried in the depth of thousands of meters to tens of thousands of meters underground is a huge problem and goal that major oil companies in the world are currently facing. It requires first to obtain a three-dimensional image of the underground. This process is commonly referred to as geophysical exploration, and then judges whether there are oil and gas resources und...

AI Technical Summary

Problems solved by technology

How to obtain a high-precision underground velocity model has always been the goal of people's efforts, but due to the high difficulty of its technology, there has been no breakthrough in this field

[0004] With the deepening of geophysical exploration, people are trying to study how to obtain a high-precision velocity model. For this reason, the technology of full waveform inversion has been proposed. This technology has achieved certain results to a certain extent, but this technology can only use seismic The recorded low-frequency components are used to invert the subsurface velocity model, otherwise there will be a very troublesome problem of cycle jump, but unfortunately, the low-frequency components of seismic records are often greatly affected by interference waves, and the signal-to-noise ratio is relatively low, such as Interference waves such as surface waves or surges generated by sea waves appear in the form of strong energy and low frequency, which affect the low frequency components of seismic signals, while conventional full waveform inversion technology can only use low frequency components of seismic waves, which is obviously not what we want of

In addition, the low-frequency component of the seismic wave cannot obtain the high-frequency component of the velocity model, that is to say, the detailed changes of the velocity model cannot be obtained, which also restricts the accuracy of the velocity model

Images