A lateral continuous seismic wave impedance inversion method based on knowledge transfer

Abstract

The present application relates to the field of geophysical exploration, in particular to a kind of transverse continuous seismic wave impedance inversion method based on knowledge transfer.The method comprises: obtaining poststack reflection seismic data, for its first seismic data, using wave impedance inversion method based on multiple groups of variant differential evolution algorithm;For each subsequent seismic data, the inversion optimal wave impedance model of its previous adjacent channel is migrated to the evolution of the current seismic channel as useful knowledge vector, and the migration of the optimal solution is constrained by the mapping matrix of structure, forming migration knowledge vector;Wave impedance inversion of the current seismic channel is carried out using multiple groups of variant differential evolution algorithm based on knowledge transfer, and a transversely continuous wave impedance model is obtained.The method of the present application can not only obtain more accurate inversion wave impedance model, but also effectively ensure the transverse continuity of the result;At the same time, the convergence speed and convergence accuracy of the algorithm are greatly improved, and the present application is more suitable for the description of lithologic hydrocarbon reservoir wave impedance parameters.

Description

Technical Field

[0001] This invention relates to the field of geophysical exploration, specifically to a method for inverting the impedance of transverse continuous seismic waves based on knowledge transfer. Background Technology

[0002] Oil and natural gas, resources buried in underground rock formations, are crucial strategic resources for national security and economic development. For a considerable period in the foreseeable future, various industries will remain highly dependent on oil and natural gas resources. Currently, my country's most important oil and gas exploration resources are lithologic oil and gas reservoirs, and seismic exploration is the primary method used in oil and gas exploration. This method reveals the distribution of underground oil and gas through parameters such as the wave impedance of the subsurface medium. However, my country's lithologic oil and gas reservoirs exhibit a thin interbedded structure vertically and strong heterogeneity horizontally. Therefore, accurately characterizing the wave impedance parameters of underground lithologic oil and gas reservoirs is a key challenge in seismic exploration.

[0003] For wave impedance inversion, commonly used inversion methods mainly include linear inversion methods based on gradient information and nonlinear inversion methods based on global optimization. Local optimization inversion methods based on gradient information are efficient due to the utilization of the gradient information of the objective functional, but they are prone to getting trapped in local optima and failing to converge to the optimal parameter model. Nonlinear inversion methods based on global optimization (such as particle swarm optimization and differential evolution algorithms) effectively solve the problems associated with local optimization, avoiding getting trapped in local extrema and improving the accuracy of the inversion results. For example, CN107390269B discloses a particle swarm optimization algorithm for wave impedance inversion constrained by the statistical characteristics of well logging data, specifically including: 1) performing phase correction on seismic data to make it a zero-phase record; 2) extracting the wavelet amplitude spectrum and obtaining the seismic wavelet; 3) analyzing and statistically analyzing velocity information based on existing well logging data within the data volume or profile, establishing a feasible region for inverting the reflection coefficient, i.e., the search space for the reflection coefficient; 4) inverting the reflection coefficient using the improved particle swarm optimization algorithm proposed in this invention; 5) calculating the wave impedance based on the inverted reflection coefficient. CN112182481B discloses a seismic waveform inversion method and system based on an improved differential evolution algorithm, the method including: step 1, calculating and obtaining the calculated seismic data of the area to be explored, and measuring and obtaining the actual seismic data of the area to be explored; using the degree of fitting between the calculated seismic data and the actual seismic data as the objective function; step 2, using the improved differential evolution algorithm to optimize the objective function obtained in step 1, obtaining the physical parameters of the subsurface medium model of the area to be explored, and completing the seismic waveform inversion.

[0004] In practice, since the subsurface medium and seismic wavelets are usually continuously changing, there is an inherent connection between adjacent seismic traces. However, global optimization algorithms based on single-trace inversion strategies ignore these similarities when solving seismic inversion problems, resulting in issues such as lateral discontinuities in the inversion results.

[0005] To overcome these problems, model-based impedance inversion methods typically rely on introducing appropriate constraints to make the inversion more stable and obtain transversely continuous seismic impedance. The most common approach is to add a regularization term (e.g., TV regularization, ATV regularization) to the objective function of the specific inversion problem. However, this method depends on prior information, and the form of the regularization is difficult to determine in many inversion problems. In addition, some researchers have used norms and Kalman filters as transverse continuity constraints, but these methods do not achieve good results in global optimization algorithms, and the accuracy of the inversion results cannot be fully guaranteed. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a knowledge transfer-based transverse continuous seismic impedance inversion method (KT-MMDE). It starts directly from seismic data and utilizes seismic impedance inversion and global optimization algorithms. Inspired by multi-task optimization problems in the field of evolutionary algorithms, it introduces the ideas of knowledge transfer and dynamic mapping into multi-channel impedance inversion, and achieves accurate inversion of seismic impedance parameters and ensures their transverse continuity without relying on regularization constraints.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] This invention provides a method for impedance inversion of transverse continuous seismic waves based on knowledge transfer, the method comprising the following steps:

[0009] After acquiring post-stack reflection seismic data, for the first seismic trace, an optimal solution, namely the optimal inverted wave impedance model, is obtained by using a wave impedance inversion method based on a multi-set mutation differential evolution algorithm.

[0010] For each subsequent seismic data trace, the inverted optimal wave impedance model of its previous adjacent trace is used as a useful knowledge vector and transferred to the evolution of the current seismic trace. The transfer of this optimal solution is constrained by the constructed mapping matrix, forming a transferred knowledge vector.

[0011] The wave impedance of the current seismic trace is inverted using a multi-set mutation differential evolution algorithm based on knowledge transfer, resulting in a transversely continuous wave impedance model.

[0012] Furthermore, the method specifically includes the following steps:

[0013] (1) Acquire raw seismic data, then preprocess the acquired seismic data to obtain post-stack reflection seismic data, denoted as d. obs (t,n), where t represents the time variable; n is the common depth point gather index;

[0014] (2) Determine the density value, time sampling interval, forward model parameters, and seismic wavelet;

[0015] (3) For actual seismic data, the wave impedance parameters are extracted from real well logging data; the low-frequency wave impedance model I can be obtained through the Butterworth filter. lf (t,n);

[0016] (4) Determine the search space range of wave impedance parameter I based on the obtained low-frequency wave impedance model, and give the objective function to be optimized in wave impedance inversion.

[0017] (5) Perform multi-channel impedance inversion. For the first seismic trace in the post-stack reflection seismic data obtained in step (1), use multiple sets of variational differential evolution algorithms to optimize the objective function in step (4). Select the individual with the smallest objective function value in the population after the algorithm iteration is completed, and record it as the optimal solution P of the current trace. This optimal solution is the optimal impedance model finally searched for the first seismic trace.

[0018] (6) For each subsequent seismic trace, its inversion with the previous seismic trace can be regarded as two adjacent tasks. The optimal solution of the previous trace can be transferred to the evolution process of the current trace to guide the direction of algorithm mutation.

[0019] (7) The optimal solution transfer in step (6) is constrained by the constructed mapping matrix M, thus forming the final transfer knowledge vector.

[0020] (8) Transfer the knowledge vector from step (7) Used in the G+1th iteration of the current path to form a new neighborhood-based mutation strategy;

[0021] (9) Select the individual with the smallest objective function value among the multiple sets of mutation differential evolution algorithms based on knowledge transfer after the iteration of step (8), which is the optimal wave impedance model of the current channel.

[0022] Furthermore, in step (2), the parameters of the forward model and its seismic wavelet are determined based on the shot-receiver distance, effective frequency band range, and sampling time of the actual seismic data.

[0023] Furthermore, in step (7), the constructed mapping matrix M is dynamically changing during the evolutionary iteration of the current path, given the parameter G. m (1 <G m <Gmax ), used to control whether the mapping matrix changes.

[0024] Furthermore, for the Gth optimization iteration of the current path, when G≠G m At this point, the mapping matrix is ​​an identity matrix, i.e., M = I; applying it to the previous optimal solution P yields the transferred knowledge vector.

[0025] Furthermore, for the Gth optimization iteration of the current path, when G = G m At this point, after the G-th iteration of the current path, a temporary optimal solution can be obtained, denoted as vector Q. G At this point, using the known vectors P and Q... G Using the constructed mapping operator Φ M (·), this operator consists of multiple linear regression operations, which yields the mapping matrix M = Φ. M (P,Q G Applying this to the previous optimal solution P yields the transferred knowledge vector.

[0026] Furthermore, in step (8), the transfer knowledge vector obtained in step (7) is... Used as a basis vector in the algorithm's mutation operation during the evolution of the current path.

[0027] Furthermore, based on the new mutation strategy, the mutation vector of the i-th individual in the G-th generation population... It can be represented as:

[0028]

[0029] In the formula, Let g represent the current individual population, a∈(0,1) be a constant, F∈[0,2] be the perturbation factor, and n1 and n2 be the number of the two temporary variant individuals, respectively. CCDE (·) and g DE / rand / 1 (·) represent neighborhood-based mutation operations in the "DE / rand / 1" method and the Cooperative Mutation Differential Evolution Algorithm (CCDE), respectively. and These represent the variant individuals generated by DE / rand / 1 and CCDE based on the new mutation strategy, respectively.

[0030] Furthermore, and Represented by formulas respectively:

[0031]

[0032]

[0033] In the formula, and and Let r1 and r2 represent individuals randomly selected from the current population, satisfying r1≠r2≠i and m1≠m2≠i.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] Compared with commonly used impedance inversion methods based on global optimization algorithms, the knowledge transfer-based lateral continuity seismic impedance inversion method of this invention not only yields a more accurate inverted impedance model but also largely ensures the lateral continuity of the results. Furthermore, the convergence speed and accuracy of the algorithm are significantly improved. Therefore, this method is more suitable for characterizing the impedance parameters of lithologic oil and gas reservoirs. The advantages of this invention are as follows: First, because a global optimization algorithm is used to solve the impedance inversion problem, this method can invert to obtain globally optimal impedance parameters. Second, this invention improves upon the global optimization method for impedance inversion by incorporating knowledge transfer and dynamic mapping mechanisms, strengthening the correlation between adjacent seismic traces, and allowing the optimization algorithm to evolve and mutate towards greater accuracy, thereby better constructing a seismic impedance model with good lateral continuity. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the process of this invention;

[0037] Figure 2 The following is a comparison of wave impedance models obtained using different algorithms in a target area with gentle lateral changes: (a) the actual model; (b) the impedance model obtained by MMDE; (c) the impedance model obtained by KT-MMDE (without mapping); (d) the impedance model obtained by KT-MMDE.

[0038] Figure 3 These are convergence curves of different algorithms in a target region with gentle horizontal changes;

[0039] Figure 4 The following is a comparison of wave impedance models obtained using different algorithms in a target area with drastic lateral changes: (a) the actual model; (b) the impedance model obtained by MMDE; (c) the impedance model obtained by KT-MMDE (without mapping); (d) the impedance model obtained by KT-MMDE.

[0040] Figure 5 These are convergence curves of different algorithms in target regions with drastic lateral changes. Detailed Implementation

[0041] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0042] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, 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.

[0043] To enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below with reference to specific embodiments.

[0044] Example 1

[0045] like Figure 1 As shown, the specific steps of the knowledge transfer-based transverse continuous seismic wave impedance inversion method are as follows:

[0046] 1) Acquire raw seismic data, then perform routine preprocessing on the acquired seismic data, and obtain post-stack reflection seismic data through migration imaging, denoted as... Where t represents the time variable; n is the common depth point gather number.

[0047] 2) Based on the actual seismic data, such as shot-receiver distance, effective frequency band range, and sampling time, a fixed density value of 1.0 g / cc is set, and forward model parameters including time sampling interval, seismic wavelet used in forward modeling, and number of sampling points are set.

[0048] 3) For model seismic data, the actual wave impedance model is known; for actual seismic data, wave impedance parameters are extracted from actual well logging data. The low-frequency wave impedance model I can be obtained through a Butterworth filter. lf (t,n), in this invention, the cutoff frequency of the Butterworth filter is set to f0 = 3Hz.

[0049] 4) The search space for wave impedance parameters is determined by using the low-frequency wave impedance model obtained in step 3) as a reference value, and adding and subtracting certain set values ​​from the reference value to form the upper and lower bounds of the search space, respectively. The objective function for wave impedance inversion is the sum of the absolute errors between the observed post-stack seismic data and the calculated post-stack seismic data synthesized from the model. For one post-stack seismic data set, the objective function is expressed as:

[0050]

[0051] In the formula, d cal (t) represents forward-modeled single-channel post-stack reflection seismic data obtained with given wave impedance parameters.

[0052] 5) For the first seismic trace in the seismic data, this invention uses the Multiple Mutation Differential Evolution (MMDE) algorithm to optimize the objective function defined in equation (3) and solve for the optimal wave impedance parameters. The MMDE algorithm mainly includes initialization, mutation, crossover, and selection operations. This algorithm solves the optimization problem based on a population of size NP, where each individual in the population is a vector with the same dimension as the model parameters. The basic process of the algorithm is briefly described as follows:

[0053] Step 1: Randomly generate NP initial individuals within the search space to form a population;

[0054] Step 2: Perform mutation operations on each individual in the population;

[0055] Step 3: Perform cross-operations on each individual in the group;

[0056] Step 4: Select each individual in the group;

[0057] Step 5: Determine if the algorithm termination condition (maximum number of iterations G) is met. max If the optimal individual in the output population is the solution to the optimization problem, then proceed to the next iteration; otherwise, jump back to the second step. The specific steps are as follows:

[0058] Initialization operation: For a D-dimensional optimization problem, if the upper and lower bounds of the search space are x min ={x 1,min ,x 2,min ,...,x D,min} and x max ={x 1,max ,x 2,max ,...,x D,max}, then initialize the j-th dimension of the i-th individual in the population. It can be represented as follows:

[0059]

[0060] In the formula, rand i,j (0,1) is a random number uniformly distributed between [0,1]. After initialization, multiple sets of mutation differential evolution algorithms will iteratively update and evolve all individuals in the population according to the three basic operations of mutation, crossover, and selection until the iteration stopping condition is met.

[0061] Mutation operation: This operation generates a mutated individual for the current individual. The mutation strategy in the multi-mutation differential evolution algorithm combines the "DE / rand / 1" method and the mutation strategy in the cooperative mutation differential evolution algorithm (CCDE). The mutation vector of the i-th individual in the G-th generation population is... It can be represented as:

[0062]

[0063] In the formula, f CCDE (·) and f DE / rand / 1 (·) represent the mutation strategies in the "DE / rand / 1" method and the Cooperative Mutation Differential Evolutionary Algorithm (CCDE), respectively. Let N represent the current individual population, a∈(0,1) be a constant, F∈[0,2] be the perturbation factor, and N1 and N2 be the number of the two temporary variant individuals, respectively.

[0064] Cross operation: This operation is performed on the current individual and its corresponding variant individuals The two individuals exchange "genes" to obtain a test individual. Crossover operation is defined as:

[0065]

[0066] In the formula, rand i,j [0,1]∈U[0,1], Cr∈[0,1] is the crossover rate, and q∈[1,D] is a randomly selected integer that ensures that even if Cr=0, there is still one dimension of information from the mutant individual in the trial individual.

[0067] Selection Operation: This operation is based on the current individual. and testing individuals The objective function value is used to update the current individual, as follows:

[0068]

[0069] In the formula, It is the updated individual, and Ψ(·) represents the objective function of the optimization problem.

[0070] 6) In multichannel impedance inversion, to account for the correlation between adjacent seismic traces, for each seismic trace after the first one, its inversion with the previous seismic trace can be considered as two adjacent tasks. The optimal solution of the previous trace is P = [p1, p2, ..., p...]. D ] T It can be transferred to the current evolutionary process of the path, guiding the direction of optimization algorithm mutation.

[0071] 7) For the previous optimal solution P in step 6), it is often unreasonable to completely transfer it to the inversion evolution of the current trace. The transfer vector needs to be constrained based on the correlation between the two seismic traces. Therefore, a mapping matrix M is introduced to constrain the previous optimal solution P that needs to be transferred. The constructed mapping matrix M is dynamically changed during the evolutionary iteration of the current trace, given the parameter G. m (1 <G m <G max This is used to control whether the mapping matrix changes. For the G-th optimization iteration of the current path, the mapping matrix mainly falls into two categories:

[0072] Scenario 1: G≠G m

[0073] At this point, the mapping matrix is ​​an identity matrix, i.e., M = I. Applying it to the previous optimal solution P yields the transferred knowledge vector. In this case, it is equivalent to completely transferring the previous optimal solution to the evolutionary inversion process of the current solution.

[0074] Case 2: G = G m

[0075] After the G-th iteration of the current path, a temporary optimal solution can be obtained, denoted as vector. At this point, using the known vectors P and Q G Using the constructed mapping operator Φ M (·), we can obtain the mapping matrix M=Φ M (P,Q G Applying this to the previous optimal solution P yields the transferred knowledge vector.

[0076] The mapping matrix M is defined as a tridiagonal square matrix, meaning that the elements on the main diagonal and the two secondary diagonals have values, and the elements in other positions have values ​​of 0. The mapping matrix M is represented as follows:

[0077]

[0078] To ensure the accuracy and effectiveness of the transferred knowledge vector for the current path inversion task, the transfer loss function is defined as follows:

[0079]

[0080] When the migration loss is minimized, the optimal mapping matrix M between the two seismic traces can be obtained. * At this point, equation (10) must hold true:

[0081] Q G =M *·P (10)

[0082] Therefore, the temporary optimal solution Q obtained at the end of the Gth iteration of the current seismic trace is... G This can be represented as the linear multiplication and addition of matrix elements, i.e.:

[0083]

[0084] Wherein, the multiplication coefficient m 11 ,m 12 ,...,m D,D-1 ,m D,D This refers to the element values ​​in the mapping matrix M that we need to find. Because vectors P and Q... G Given this, the element values ​​of each row in the mapping matrix M can be obtained using the linear regression operator, as shown below:

[0085]

[0086] In the formula, Φ M (·) denotes a mapping operator, which consists of D sub-mapping operators. Each sub-mapping operator corresponds to a linear regression operation. Based on this mapping operator, the mapping matrix M of the constraint transfer can be obtained.

[0087] In summary, the knowledge vector transferred between inversion tasks of two adjacent seismic traces can be represented as:

[0088]

[0089] 8) Transfer the knowledge vector obtained in step 7) Used in the evolutionary process of current path inversion as a basis vector in the mutation operation, thus better guiding the mutation direction of candidate solution individuals. Based on the new mutation strategy, the mutation vector of the i-th individual in the G-th generation population is... It can be represented as:

[0090]

[0091] In the formula, Let g represent the current individual population, a∈(0,1) be a constant, F∈[0,2] be the perturbation factor, and n1 and n2 be the number of the two temporary variant individuals, respectively. CCDE (·) and g DE / rand / 1 (·) represent neighborhood-based mutation operations in the "DE / rand / 1" method and the Cooperative Mutation Differential Evolution Algorithm (CCDE), respectively. and These represent the variant individuals generated by DE / rand / 1 and CCDE based on the new mutation strategy, respectively, as shown below:

[0092]

[0093]

[0094] In the formula, and and Let r1 and r2 represent individuals randomly selected from the current population, satisfying r1≠r2≠i and m1≠m2≠i.

[0095] 9) Select the individual with the smallest objective function value in the global optimization algorithm based on knowledge transfer after the iteration in step 8), which is the optimal wave impedance model for the current channel.

[0096] Example 2

[0097] The method described in Example 1 was applied to the Marmousi II geological model. After downsampling, the model had a horizontal span of 5670 meters and a temporal span of 0.8 seconds. The model was discretized with a spatial sampling interval of 5 meters and a temporal sampling interval of 0.008 seconds. The discretized wave impedance model had 1134 channels, each with 100 dimensions. Channels 1-300 showed gentle horizontal changes, with a low-velocity anomaly around channel 200. Channels 400-700 showed drastic horizontal changes, indicating faults. The method of this invention was used to test the target area with gentle horizontal changes in channels 1-300 and the target area with drastic horizontal changes in channels 401-700.

[0098] Figure 2 This is a comparison chart of wave impedance models obtained using different algorithms in a target area with gentle lateral changes. The models represent the true wave impedance model, the wave impedance model obtained by MMDE inversion, the wave impedance model obtained by KT-MMDE (without mapping), and the wave impedance model obtained by KT-MMDE inversion. Figure 2 As shown in (b), the traditional global optimization algorithm is based on single-channel inversion, therefore the inverted wave impedance will exhibit obvious transverse discontinuities; while Figure 2 (d) The KT-MMDE inversion results employing knowledge transfer and dynamic mapping not only significantly improved inversion accuracy but also effectively ensured the transverse continuity of the wave impedance model. Furthermore, to test the effectiveness and necessity of incorporating dynamic mapping into the algorithm, Figure 2 (c) shows the inversion results of the KT-MMDE algorithm without mapping, which is an improvement over the traditional method, but a large error occurs near the low-speed anomaly.

[0099] Figure 3These are convergence curves of different algorithms in a target region with a gentle horizontal change. Because the method proposed in this invention introduces a knowledge sharing and dynamic mapping mechanism between multiple channels, the method has greatly improved in terms of convergence accuracy and convergence speed.

[0100] Figure 4 This is a comparison chart of wave impedance models obtained using different algorithms in a target area with gentle lateral changes. The charts represent the true wave impedance model, the wave impedance model obtained by MMDE inversion, the wave impedance model obtained by KT-MMDE (without mapping), the wave impedance model obtained by KT-MMDE inversion, and the difference profiles between the latter three and the true impedance. It can be seen that in target areas with drastic lateral changes, the method proposed in this invention can still obtain wave impedance inversion results with high accuracy and good lateral continuity, while traditional global optimization algorithms are not applicable.

[0101] Figure 5 These are convergence curves of different algorithms in a target region with drastic lateral changes. It can be seen that there are significant improvements in both convergence accuracy and convergence speed.

[0102] The wave impedance inversion method proposed in this invention can obtain wave impedance inversion results with good lateral continuity and high consistency with the real wave impedance model without relying on regularization constraints, and is applicable to geological regions with different degrees of lateral variation.

[0103] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for inverting the impedance of transverse continuous seismic waves based on knowledge transfer, characterized in that, The method includes: After acquiring post-stack reflection seismic data, for the first seismic trace, an optimal solution, namely the optimal wave impedance model, is obtained by using a wave impedance inversion method based on a multi-set mutation differential evolution algorithm. For each subsequent seismic data trace, the inverted optimal wave impedance model of its previous adjacent trace is used as a useful knowledge vector and transferred to the evolution of the current seismic trace. The transfer of this optimal solution is constrained by the constructed mapping matrix, forming a transferred knowledge vector. The wave impedance of the current seismic trace is inverted using a multi-set mutation differential evolution algorithm based on knowledge transfer, resulting in a transversely continuous wave impedance model. The mapping matrix is ​​constructed as a tridiagonal square matrix, meaning that there are element values ​​on the main diagonal and the two secondary diagonals, and the element values ​​at other positions are all 0. By defining a migration loss function, when the migration loss value is minimized, the element value of each row in the mapping matrix is ​​obtained using a linear regression operator; The specific steps include: (1) Acquire raw seismic data, then preprocess the acquired seismic data to obtain post-stack reflection seismic data, denoted as . ,in Represents the time variable; n is the common depth point gather index; (2) Determine the density value and the parameters of the forward model; (3) For actual seismic data, the wave impedance parameters are extracted from real well logging data; the low-frequency wave impedance model can be obtained through the Butterworth filter. ; (4) Determine the wave impedance parameters based on the obtained low-frequency wave impedance model. The search space range is defined, and the objective function to be optimized in wave impedance inversion is given. (5) Perform multi-channel impedance inversion. For the first seismic trace in the post-stack reflection seismic data obtained in step (1), use multiple sets of variational differential evolution algorithms to optimize the objective function in step (4). Select the individual with the smallest objective function value in the population after the algorithm iteration is completed, and record it as the optimal solution P of the current trace. This optimal solution is the optimal impedance model finally searched for the first seismic trace. (6) For each subsequent seismic trace, its inversion with the previous seismic trace can be regarded as two adjacent tasks. The optimal solution of the previous trace can be transferred to the evolution process of the current trace to guide the direction of algorithm mutation. (7) The optimal solution in step (6) is transferred by the constructed mapping matrix. This constraint forms the final transfer knowledge vector. ; (8) Transfer the knowledge vector from step (7) Used in the G+1th iteration of the current path to form a new neighborhood-based mutation strategy; (9) Select the individual with the smallest objective function value among the multiple sets of mutation differential evolution algorithms based on knowledge transfer after the iteration of step (8), which is the optimal wave impedance model of the current channel; In step (7), the constructed mapping matrix The evolution and iteration of the current Dao is dynamic, given the parameter G. m ,1<G m <G max This is used to control whether the mapping matrix changes; For the current path In the next optimization iteration, when At this point, the mapping matrix is ​​an identity matrix; applying it to the previous optimal solution P yields the transferred knowledge vector. ; For the current path In the next optimization iteration, when At that time, the current path's first After each iteration, a temporary optimal solution is obtained, denoted as a vector. At this point, using the known vector P and Using the constructed mapping operator This operator consists of multiple linear regression operations, which yields a mapping matrix. Applying this to the previous optimal solution P yields the transferred knowledge vector. .

2. The knowledge transfer-based impedance inversion method for transverse continuous seismic waves according to claim 1, characterized in that, In step (2), the parameters of the forward model are determined based on the shot-receiver distance, effective frequency band range, and sampling time of the actual seismic data.

3. The knowledge transfer-based impedance inversion method for transverse continuous seismic waves according to claim 1, characterized in that, In step (8), the transfer knowledge vector obtained in step (7) is... Used as a basis vector in the algorithm's mutation operation during the evolution of the current path.

Images