Pre-stack multi-parameter inversion method with balanced model constraints

Abstract

The present invention provides a pre-stack multi-parameter inversion method with balanced model constraints. The method includes: Step 1, constructing an initial iterative model and a broadband balance constraint model by using seismic structure interpretation and logging information; Step 2, constructing an objective function with a double loss function of seismic residual and broadband balance model residual; Step 3, optimizing and solving the objective function according to the inversion wavelet, pre-stack seismic and constraint model to obtain broadband P-wave and S-wave velocities and density. The pre-stack multi-parameter inversion method with balanced model constraints adds broadband multi-parameter model constraints on the basis of pre-stack seismic constraints. By adjusting the balance parameters, the contribution sizes of the two constraint information can be adjusted, which is of great significance for expanding the seismic inversion frequency band and improving the inversion resolution by using seismic, logging and full waveform inversion information.

Description

Technical Field

[0001] The present invention relates to the field of seismic data processing, and particularly to a pre-stack multi-parameter inversion method constrained by a balanced model. Background Art

[0002] Well-log constrained seismic inversion is an important reservoir prediction and description technology in oil and gas exploration and development. Most of the conventional inversion objective functions are constructed by using the forward seismic of the parameters to be inverted and the residual of the actual seismic as the loss function, and then optimizing and solving. Therefore, seismic information is the main known information in the inversion process. Seismic is a kind of data with strong three-dimensional space description ability, but limited seismic frequency band and geological resolution. In conventional inversion, only seismic information is used, and the final inversion result will tend to the seismic frequency band, with limited inversion frequency band and resolution, which is not conducive to the identification and description of high-resolution reservoirs. With the improvement of the degree of oil and gas exploration, in addition to seismic data, there are also a large number of well-log data in many oil and gas exploration and development areas, and a large amount of geological structure and sedimentary understanding and interpretation results have been accumulated, which can be used to construct a broadband model that reflects the changes of thick strata and thin reservoirs longitudinally, but the model changes are not natural and accurate enough in the transverse three-dimensional space, and three-dimensional seismic information is needed to play a role. Therefore, an inversion method that can efficiently use seismic and model information simultaneously is needed to broaden the inversion frequency band, improve the inversion accuracy and reservoir resolution ability.

[0003] In the Chinese patent application with the application number: CN201911315267.9, a pre-stack seismic inversion method for shale reservoirs is involved. The specific implementation steps are as follows: Step 1: Obtain the pre-stack seismic data of the research work area, and this data is the gather data containing full-wave effects; Step 2: Set the depth range for which inversion needs to be carried out, and use the collected logging data to establish an initial model of elastic parameters in the depth domain within the set depth range; Step 3: Based on the initial model, calculate the reflection coefficient sequence in the time-angle domain using the recursive matrix; Step 4: Based on the pre-stack seismic gather and the obtained reflection coefficient sequence, calculate the forward simulation record gather corresponding to the initial model, and calculate the residual term between the pre-stack gather and the simulated record data; Step 5: Calculate the partial derivatives of the forward propagation of the recursive matrix using the initial model parameters; Step 6: Calculate the Jacobian matrix and the pseudo-Hessian matrix of the objective function based on the objective function and the partial derivatives of the forward operator; Step 7: Calculate the update direction of the model and update the model parameters; Step 8: Repeat Steps 3 to 7 to perform the next inversion iteration until the model error drops to the preset range and the iteration stops, and output the inversion results of the parameters. The objective function of this patent method is constructed with the residual between the pre-stack gather and the forward simulation seismic record of the logging model as the loss function, uses seismic information to update the inversion model of the seismic frequency band part, and completely retains the low-frequency and high-frequency information of the logging model. Although the inversion result also has the characteristic of wide frequency, the inversion result does not consider the construction accuracy of the logging model in the case of complex geology and less logging information, especially the reliability of high-frequency information. Although the final inversion result has high resolution, there is a lot of uncertain information, resulting in multiple solutions for reservoir prediction.

[0004] In the Chinese patent application with the application number: CN201710910382.5, a pre-stack seismic inversion method and system are involved. The method may include: converting the pre-stack seismic gather into an angle gather, grouping and stacking to obtain multiple angle gather groups; establishing a shale gas reservoir model, obtaining the density sensitivity of the elastic impedance equation, and selecting a density inversion equation; performing elastic impedance inversion on the multiple angle gather groups according to the density inversion equation to obtain the elastic impedance data volume of the multiple angle gather groups; establishing a relationship between the P-wave velocity, S-wave velocity, density, and incident angle and linearizing it to obtain a linearized elastic impedance equation; writing the P-wave velocity, S-wave velocity, and density as weighted sums of the elastic impedance respectively; obtaining the elastic impedance weighting coefficients based on the elastic impedance inversion results of the well-side gather and the logging data; and obtaining the reservoir elastic parameters based on the weighted sum and the weighting coefficients. This patent method has many links and a complex process. It cannot directly use the pre-stack seismic data to synchronously invert the P-wave velocity, S-wave velocity, and density. It is necessary to calculate the elastic impedance corresponding to multiple angles, and then use the linearized elastic impedance equation to solve elastic parameters such as the P-wave velocity, S-wave velocity, and density. The actual application workload is large and there are many process errors, which affect the inversion accuracy.

[0005] In the Chinese patent application with the application number CN201611116496.4, a pre-stack seismic inversion method and device for horizontal fractures are involved, including: obtaining the physical property parameters of the underground medium in the target work area according to the logging data of the target work area; the physical property parameters of the underground medium include: the physical property parameters of each formation and the fracture parameters of each fracture; extracting the seismic wavelets of the seismic trace gather beside the well in the target work area according to the obtained physical property parameters of the underground medium; establishing the initial formation model of the target work area according to the obtained physical property parameters of the underground medium; and inversely obtaining the formation model of the true physical property parameters of the underground medium in the target work area according to the initial formation model, the seismic wavelets of the seismic trace beside the well, and the actual seismic trace gather. This patented method is mainly a pre-stack seismic inversion method invented for horizontal fracture prediction. In addition to the longitudinal and transverse wave velocities and density, the horizontal fracture parameters are also obtained as the final inversion parameters. The inversion theory of this method is different from the conventional non-vertically incident seismic wave theory and lacks the verification of actual data. At the same time, when constructing and inversely obtaining the initial model with 5 parameters, the reliability of the initial model needs to be considered more.

[0006] Therefore, we have invented a new pre-stack multi-parameter inversion method with balanced model constraints to solve the above technical problems. Summary of the Invention

[0007] In view of the above problems, the object of the present invention is to provide a pre-stack multi-parameter inversion method with balanced model constraints, which can efficiently utilize seismic and broadband model information to inversely obtain a variety of elastic parameters with strong broadband reservoir space description ability.

[0008] The object of the present invention can be achieved by the following technical measures: a pre-stack multi-parameter inversion method with balanced model constraints, which includes: Step 1, constructing an initial iterative model and a broadband balanced constraint model by using seismic structural interpretation and logging information; Step 2, constructing an objective function with a double loss function of seismic residual and broadband balanced model residual; Step 3, inputting the inversion wavelet, pre-stack seismic and constraint model, and performing optimization solution of the objective function to obtain broadband longitudinal and transverse wave velocities and density.

[0009] The object of the present invention can also be achieved by the following technical measures:

[0010] In Step 1, based on seismic structural interpretation and logging information, an initial iterative model and a broadband multi-parameter balanced constraint model are constructed to provide constraints for the application of this pre-stack multi-parameter inversion method with balanced model constraints.

[0011] In Step 2, the objective function includes seismic residual and balanced model residual terms:

[0012]

[0013] In the formula, S(VP, VS, ρ)Δ is the expected seismic response, D1 is the pre-stack seismic record, D2 is the forward seismic record of the equilibrium constraint model, VP represents the P-wave velocity to be inverted, VS represents the S-wave velocity to be inverted, ρ is the density to be inverted, and β is the balance parameter.

[0014] In step 2, the balance parameter β is set in the objective function to control the weights of the two loss functions to distribute the contributions of the earthquake and balance models. The value range is 0-1. When β = 0, the balance model has no contribution to the inversion. When β = 1, the balance model and earthquake each contribute 50% to the inversion.

[0015] In step 3, the expected model response is Taylor expanded and the high-order terms are omitted to obtain the linearized expression of the initial model response, Jacobian matrix, and expected model response of the model disturbance:

[0016] S(VP,VS,ρ) Δ =S(VP,VS,ρ)+GΔM=WR(VP,VS,ρ)+GΔM (2)

[0017] Where W is the wavelet extracted from the actual pre-stack seismic, G is the Jacobian matrix composed of the partial derivatives of the P-wave velocity, S-wave velocity, and density of the iterative model forward seismic; ΔM is the matrix composed of the P-wave velocity, S-wave velocity, and density perturbation at each sampling point, expressed as:

[0018] ΔM=[ΔVP1 ΔVS1 Δρ1 ΔVP2 ΔVS2 Δρ2 … ΔVP n ΔVS n Δρ n ] T .

[0019] In step 3, substitute equation (2) into equation (1), and calculate the first-order partial derivative, setting it to zero, and the solution formula for the inversion parameter model perturbation quantity can be obtained:

[0020]

[0021] The initial iterative model is updated using the obtained disturbance. The formula is as follows:

[0022]

[0023] Where m=(VP i , VS i , VP i+1 , VS i+1 ,ρ i+1 ,ρ i ), represents the P-wave velocity, S-wave velocity and density parameters of the i-th sampling point and its next adjacent sampling point; λ is the regularization parameter, and k is the number of iterations.

[0024] In step 3, the iterative model is continuously updated. When the number of iterations or the objective function meets certain conditions, the iteration terminates, and finally, the inversion of the P-wave velocity, S-wave velocity, and density is achieved.

[0025] In the present invention, due to the adoption of the above technical solutions, the following advantages are achieved: Aiming at the problems of high dependence on conventional inversion seismic, narrow frequency band, and low accuracy of inversion results, the inversion objective function adds a balanced model constraint term, and the balanced parameter is used to control the contribution of the model to the final inversion. This inversion method can better play the role of broadband modeling of well logging and full waveform inversion information, further expand the inversion frequency band, and improve the inversion resolution and reservoir prediction ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flowchart of a specific embodiment of the pre-stack multi-parameter inversion method with balanced model constraint of the present invention;

[0027] Figure 2 It is a schematic cross-sectional view of the broadband multi-parameter balanced constraint model constructed in an embodiment of the present invention;

[0028] Figure 3 It is a cross-sectional view of the multi-parameter inversion result and seismic stack when the balanced parameter β = 1 in an embodiment of the present invention;

[0029] Figure 4 It is a comparison cross-sectional view of the inverted P-wave velocity when the balanced parameters are β = 1 and β = 0 respectively in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0031] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, and / or combinations thereof.

[0032] As Figure 1 shown, Figure 1 It is a flowchart of a specific embodiment of the pre-stack multi-parameter inversion method with balanced model constraint of the present invention. The pre-stack multi-parameter inversion method with balanced model constraint of the present invention includes the following steps:

[0033] Step 1: Based on seismic structural interpretation and logging information, construct an initial iterative model and a broadband multi-parameter balanced constraint model.

[0034] Step 2: Construct an objective function with a double loss function of seismic residual and broadband balanced model residual;

[0035] The inversion objective function is as follows:

[0036]

[0037] In the formula, S(VP, VS, ρ) Δ is the expected seismic response, D1 is the pre-stack seismic record, D2 is the forward seismic record of the balanced constraint model, VP represents the longitudinal wave velocity to be inverted, VS represents the shear wave velocity to be inverted, ρ is the density to be inverted, β is the balance parameter, controlling the weights of the two loss functions, and the value range is 0 - 1.

[0038] Step 3: Input the inversion wavelet, pre-stack seismic, initial iterative model, and broadband model, perform optimization and solution of the objective function, and obtain broadband longitudinal and shear wave velocities and density.

[0039] Performing Taylor expansion on the expected model response and omitting the high-order terms, the linearized expression of the expected model response of the initial model response, Jacobian matrix, and model perturbation amount can be obtained:

[0040] S(VP, VS, ρ) Δ = S(VP, VS, ρ) + GΔM = WR(VP, VS, ρ) + GΔM (2)

[0041] Among them, W is the broadband wavelet for pre-stack seismic energy matching, G is the Jacobian matrix composed of the partial derivatives of the forward seismic of the iterative model with respect to the longitudinal wave velocity, shear wave velocity, and density. ΔM is the matrix composed of the perturbation amounts of the longitudinal wave velocity, shear wave velocity, and density at each sampling point, expressed as:

[0042] ΔM = [ΔVP1 ΔVS1 Δρ1 ΔVP2 ΔVS2 Δρ2 … ΔVP n ΔVS n Δρ n T

[0043] Substitute equation (2) into equation (1), take the first-order partial derivative, and set it to zero, the solution formula for the perturbation amount of the inversion parameter model can be obtained:

[0044]

[0045] Use the obtained perturbation amount to update the initial iterative model, and the formula is as follows

[0046]

[0047] where m = (VP i , VS i , VP i+1 , VS i+1 , ρ i+1 , ρ i ), representing the P-wave velocity, S-wave velocity, and density parameters of the i-th sampling point and its next adjacent sampling point. λ is the regularization parameter, and k is the number of iterations.

[0048] By using the above objective function and the initial three-parameter model iterative update formula, the initial model is continuously updated and iterated. When the number of iterations or the objective function meets certain conditions, the inversion iteration is terminated, and the final iteratively corrected model parameters are output, and the P-wave velocity, S-wave velocity, and density inversion results can be obtained.

[0049] In the specific embodiment 1 of applying the present invention, the prestack multi-parameter inversion method with balanced model constraints of the present invention includes the following steps:

[0050] Step 1, based on the structure and logging, construct a broadband multi-parameter balanced constraint model. According to the actual reservoir thickness response frequency, process the logging high cut-off frequency at 120 Hz to reduce the high-frequency random information of thin layers and improve the stability of the broadband model. At the same time, use the structure and logging to construct an initial iterative model, and set the logging high cut-off frequency at about 10 Hz, and only use the low-frequency information of logging to construct. Figure 2 The schematic diagram of the broadband multi-parameter balanced constraint model constructed in an embodiment of the present invention is shown. They are the P-wave velocity, S-wave velocity, and density models respectively. The model is constructed by using the interpolation method with 120 Hz high cut-off frequency logging information under the constraint of geological structure. The model has a wide frequency band and high resolution. However, due to the small number of logging samples, the model has poor representation ability for the lateral variation of the geological space and can be used as the application balanced constraint model of the inversion method.

[0051] Step 2, construct an objective function with a double loss function of seismic residual and broadband balanced model residual.

[0052] The objective function is as follows:

[0053]

[0054] Derive and construct the inversion iteration formula:

[0055]

[0056] D2 represents the forward seismic record modeled using the broadband balance constraint model and the broadband inversion wavelet W, and can be expressed as D2 = WR(VP, VS, ρ). β is the balance parameter, which controls the weight of the two loss functions and ranges from 0 to 1. The initial setting of the balance parameter should be determined based on the correlation between the forward seismic modeled with the balance constraint model and the actual earthquake. A high correlation coefficient indicates a strong reliability of the balance model, and the balance parameter value should be increased. The balance parameter should also be gradually decreased with increasing inversion iterations. In this example, two parameter values, β = 1 and β = 0, were selected and applied.

[0057] Step 3: Input the inversion wavelet, prestack seismic, initial iteration model and broadband constraint model, perform objective function optimization and solve, and obtain broadband P- and S-wave velocities and densities. Figure 3 This is the multi-parameter inversion result and seismic stacked section for the balance parameter β = 1 in one embodiment of the present invention. The broadband balance model and seismic each contribute 50% to the inversion. Compared to broadband models constructed using only well logging information, the inversion adds spatial constraints and seismic information, resulting in natural spatial variations in P-wave velocity, S-wave velocity, and density. Furthermore, the inversion has a wider bandwidth than seismic, resulting in higher reservoir resolution.

[0058] Figure 4 Figure 1 shows a comparative profile of the inverted P-wave velocity when the balance parameter β = 1 and β = 0, respectively, in one embodiment of the present invention. When the balance parameter β = 0, the broadband balance constraint model does not work. This means that conventional pre-stack multi-parameter inversion based on seismic data only produces low-frequency information from the initial iteration model of the well logging data and seismic frequency band information. However, when β = 1 is used, the inverted P-wave velocity has a higher resolution due to the addition of the broadband balance model constraint information.

[0059] In a second specific embodiment of the present invention, the prestack multi-parameter inversion method constrained by the equilibrium model of the present invention includes the following steps:

[0060] Step 1: Based on the structure and well logging data, a low-frequency model and a broadband multi-parameter balanced constraint model were constructed. First, a low-frequency P-wave velocity model and a broadband balanced P-wave velocity constraint model were constructed using the 10Hz and 150Hz high-cutoff P-wave velocities, S-wave velocities, and density information from well logging. In this example, the broadband model has a wider high-frequency band, resulting in better thin-bed resolution but increased uncertainty.

[0061] Step 2: Construct an objective function with a double loss function of earthquake residuals and broadband balance model residuals.

[0062] The objective function is as follows:

[0063]

[0064] The inversion iteration formula is derived and constructed:

[0065]

[0066] D2 is the forward seismic record of the broadband balanced constraint model and the broadband inversion wavelet W, which can be expressed as D2 = WR(VP, VS, ρ). In this example, β = 0.5 can be set so that the inversion mainly relies on seismic data to ensure the certainty of inversion, and at the same time, the broadband constraint model is appropriately applied to improve the inversion resolution.

[0067] Step 3: Input the inversion wavelet, pre-stack seismic data, initial iterative model, and broadband constraint model, perform optimization solution of the objective function, and obtain the broadband P-wave and S-wave velocities and density. The certainty and resolution of the inversion results are between the inversion results of the balance parameters β = 0 and β = 1.

[0068] The pre-stack multi-parameter inversion method constrained by this balance model adds a broadband multi-parameter model constraint on the basis of the pre-stack seismic constraint, combines the balance parameter according to the model construction accuracy, and can flexibly and controllably adjust the contribution sizes of the two constraint information of the seismic and logging models, which is of great significance for using logging information to expand the seismic inversion frequency band and improve the inversion resolution.

[0069] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0070] Except for the technical features described in the specification, the rest are the known technologies of those skilled in the art.

Claims

1. A prestack multi-parameter inversion method with balanced model constraints, characterized in that, The pre-stack multi-parameter inversion method constrained by the balance model includes the following steps: Step 1: Construct an initial iterative model and a broadband balance constraint model by using seismic structural interpretation and logging information; Step 2: Construct an objective function with a double loss function of seismic residual and broadband balance model residual; Step 3: Optimize and solve the objective function according to the inversion wavelet, pre-stack seismic and constraint model to obtain broadband P-wave and S-wave velocities and density; In Step 2, the objective function contains seismic residual and balance model residual terms: where S(VP, VS, ρ) Δ is the expected seismic response, D1 is the pre-stack seismic record, D2 is the forward seismic record of the balanced constraint model, VP represents the P-wave velocity to be inverted, VS represents the S-wave velocity to be inverted, ρ is the density to be inverted, and β is the balance parameter; In Step 3, the expected model response is Taylor-expanded and the high-order terms are omitted to obtain the linearized expression of the expected model response of the initial model response, Jacobian matrix and model perturbation amount: S(VP, VS, ρ) Δ = S(VP, VS, ρ) + GΔM = WR(VP, VS, ρ) + GΔM (2) where W is the wavelet extracted from the actual pre-stack seismic, G is the Jacobian matrix composed of the partial derivatives of the forward seismic of the iterative model with respect to P-wave velocity, S-wave velocity and density; ΔM is the matrix composed of the perturbation amounts of P-wave velocity, S-wave velocity and density at each sampling point, expressed as: ΔM = [ΔVP1 ΔVS1 Δρ1 ΔVP2 ΔVS2 Δρ2 … ΔVP n ΔVS n Δρ n T ;​ In Step 3, substitute Equation (2) into Equation (1), take the first-order partial derivative and set it to zero to obtain the solution formula for the perturbation amount of the inversion parameter model: where λ is the regularization parameter and k is the number of iterations.

2. The prestack multi-parameter inversion method with balanced model constraints according to claim 1, characterized in that: In Step 1, based on seismic structural interpretation and logging information, construct an initial iterative model and a broadband multi-parameter balance constraint model to provide constraints for the application of the pre-stack multi-parameter inversion method constrained by the balance model.

3. The pre-stack multi-parameter inversion method with balanced model constraints according to claim 1, characterized in that: In Step 2, set the balance parameter β in the objective function to control the weights of the two loss functions to allocate the contribution sizes of the seismic and balance models. The value range is 0-1. When β = 0, the balance model has no contribution to the inversion. When β = 1, the balance model and the seismic each account for 50% of the contribution to the inversion.

4. The prestack multi-parameter inversion method with balanced model constraints according to claim 1, wherein: In Step 3, continuously update the initial iterative model. When the number of iterations or the objective function meets certain conditions, the iteration terminates, and finally the inversion of P-wave velocity, S-wave velocity and density is realized.

Images