50results about How to "Improve inversion efficiency" patented technology

The invention provides a combined inversion method of ocean controllable source electromagnetic and magnetotelluric based on a product function, comprising: reading and converting ocean controllable source electromagnetic field and ocean magnetotelluric data; setting joint inversion execution parameters; establishing joint inversion Initial model; constructing joint inversion objective function; nonlinear transformation of inversion parameters; obtaining Jacobian and Hessian matrices of inversion data for formation anisotropic resistivity and formation thickness; calculating model parameter update based on inversion parameters search inversion update amount step size; calculate the inversion iteration fit difference; judge whether the inversion requirements are met; output the final inversion model. The method simultaneously inverts the anisotropic resistivity and thickness of the submarine formation, and the inversion efficiency is higher; at the same time, it combines the high resolution of the oceanic controllable source electromagnetic data for shallow formations, and the ability of magnetotelluric data to detect deep structures. The inversion effect is better than the traditional single electromagnetic data inversion; the proposed joint inversion method has wider applicability.

The invention discloses an underground structure inversion imaging method, which comprises the following steps: S1) inputting background velocity field, observation data and a seismic source, and obtaining an initial reflection coefficient model; S2) carrying out encoding on the observation data to obtain a super shot gather, and carrying out encoding on the seismic source to obtain an encoded seismic source; S3) obtaining forward modeling data according to the encoded seismic source and the initial reflection coefficient model; S4) obtaining data residual between the forward modeling data and the super shot gather; and S5) according to the data residual, determining the gradient, and carrying out reflection coefficient model iteration based on the gradient to obtain a final reflection coefficient model. By carrying out segmented step coding on the observation data and the seismic source, difference between multi-seismic resource shot data is increased fully, interference between adjacent shots is reduced, crosstalk noise is suppressed well during multi-seismic resource shot data imaging, and imaging precision during multi-seismic resource shot data imaging inversion is improved.

The invention discloses a three-dimensional time-domain aeronautical electromagnetic fast inversion method. The method first constructs the best second-generation wavelet suitable for time-domain aeronautical electromagnetic data, and performs multi-scale decomposition on the time-domain aeronautical electromagnetic observation data; self-adaptive selection Optimize the wavelet threshold, extract the characteristic data of the current scale; conduct multi-scale wavelet analysis of model sensitivity, and adaptively optimize the time-domain airborne electromagnetic inversion grid; Three-dimensional Gauss-Newton inversion; by gradually reducing the scale of data space and model space, time-domain airborne electromagnetic characteristic data on different scales are extracted step by step to participate in the inversion, and at the same time, the inversion grid is updated with variable scales to achieve fast time-domain airborne electromagnetic data Refined inversion. While ensuring the resolution, it can effectively improve the efficiency of time-domain airborne electromagnetic 3D inversion.

The present application relates to a method and device for two-dimensional inversion of electric field data by magnetotelluric TE polarization modes. The method includes: observing electric field data components at different frequencies of multiple observation points along the X axis in the survey line and calculating the electric field data components at different frequencies of the observation points along the X axis, and constructing an inversion objective function under the magnetotelluric TE polarization mode ; By inverting the objective function, the inversion iterative equations corresponding to the inversion objective function are obtained; by introducing the normalized electric field data component error and replacing the measured magnetic field Hy component with the calculated magnetic field Hy component, the actual fitting is Zhang Quantitative impedance data; According to the mathematical expression of the normalized electric field data component error, calculate the Jacobian matrix, calculate the inversion iterative equation group according to the Jacobian matrix and the normalized electric field data component error, according to the inversion iterative equation group The electric field data were inverted in 2D. Using this method can improve the calculation efficiency and inversion accuracy of two-dimensional inversion.

The invention discloses a multi-scale near-surface tomographic velocity modeling method and a modeling system, comprising: step 1: inputting first-arrival travel time and initial velocity model; step 2: calculating theoretical travel time and ray path based on the initial velocity model; Step 3: Obtain the speed modifier, and according to the speed modifier, modify the initial speed model to obtain the current speed model; Step 4: Compare the speed modifier with the preset speed modifier threshold; Step 5: If the speed modifier is greater than or equal to the preset speed To modify the threshold value, compare the maximum offset with the offset threshold; Step 6: If the maximum offset is greater than or equal to the offset threshold, use the current velocity model as the final velocity model. The multi-scale near-surface tomographic velocity modeling method of the present invention improves the inversion accuracy by inputting the first-arrival travel time and the initial velocity model, and modifying the initial velocity model according to the velocity modification value, and can establish a high-precision near-surface velocity model.