Residual static correction nonlinear optimization method and device, medium and equipment
By using a genetic algorithm for nonlinear optimization, the problem of high-frequency residual static correction under complex surface conditions was solved, improving the quality of seismic imaging and the continuity of the phase axis. This overcame the limitations of traditional methods and achieved higher precision static correction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA PETROLEUM & CHEMICAL CORP
- Filing Date
- 2024-11-19
- Publication Date
- 2026-05-19
AI Technical Summary
Existing technologies are insufficient to effectively solve the problem of high-frequency residual static correction under complex surface conditions, especially in low signal-to-noise ratio regions. Traditional linear reflected wave residual static correction methods are ineffective and cannot accurately determine the residual static correction amount.
A genetic algorithm is used for nonlinear optimization. By generating an initial population, calculating fitness values, performing gene crossover and sorting, the residual static correction of the reflected wave is optimized to achieve global optimization and avoid local extremum traps and wavelet half-period limitations.
It improves the imaging quality of seismic data, enhances the continuity and energy focusing of the same phase axis, improves the characterization of underground structures, and meets the static correction requirements under complex near-surface conditions.
Smart Images
Figure CN122063671A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of petroleum geophysical exploration technology, specifically relating to a nonlinear optimization method, apparatus, medium, and equipment for residual static correction quantities. Background Technology
[0002] Seismic exploration is the most important technical method for locating underground oil and gas resources and serves as a crucial basis for decisions regarding oil and gas exploration and development deployment. Currently, onshore seismic exploration in China has shifted from the east to the northwest and southwest regions. These areas are mostly characterized by mountainous, desert, and loess plateau terrains, with complex surface and subsurface structures, making oil and gas exploration challenging and placing higher demands on the accuracy of seismic imaging. High-precision seismic imaging requires accurate data preprocessing, including static correction.
[0003] Static correction has always been a crucial technical step in the processing of land seismic data. The results of static correction directly affect the accuracy of seismic imaging structures and significantly impact the signal-to-noise ratio and resolution of the processed seismic data. Theoretically, high-frequency static correction is equivalent to adding a low-pass filter to the seismic data, causing phase distortion of the seismic wavelet and attenuation of high-frequency amplitude, ultimately reducing the resolution of the profile. Furthermore, since static correction occurs in the early stages of the processing, the continuity and energy intensity of the phase axes largely depend on the quality of the static correction. This significantly affects the quality of subsequent critical processing steps such as denoising and velocity analysis. For example, the suppression of regular noise heavily relies on the continuity of the phase axes; velocity analysis also depends on the horizontal superposition of phase axes; this influence propagates and amplifies further during processing.
[0004] Currently, both domestically and internationally, the combined application of datum static correction and reflected wave residual static correction is considered a core technical approach for solving static correction problems. Medium- and long-wavelength static correction problems are addressed through primary static correction methods such as elevation static correction, tomographic static correction, and refraction static correction. Short-wavelength static correction problems are then solved through iterative methods of reflected wave residual static correction and velocity analysis. This approach has achieved significant results in practical applications, but it also has limitations, mainly in the following two aspects.
[0005] 1) Limitations of single static correction
[0006] The terrain in complex piedmont areas and loess plateaus is highly undulating with complex near-surface structures, exhibiting drastic variations in longitudinal and lateral velocities, unstable refractive interfaces, and low signal-to-noise ratios. Even after initial static correction based on first arrival travel time, significant residual static correction issues often remain, resulting in low-quality stacked images that cannot meet subsequent processing requirements.
[0007] Existing single-line static correction methods, including elevation static correction, refraction static correction, and tomographic static correction, can all be considered model-based static correction in principle: a near-surface velocity model is inverted using the first arrival wave from the earthquake, and then the static correction is calculated using the velocity model. Model-based static correction has achieved significant results in practical applications and remains the most important single-line static correction method. However, single-line static correction cannot completely solve most static correction problems, especially in areas with dramatic surface elevation fluctuations and abrupt changes in near-surface velocity in both longitudinal and lateral directions. This is because model-based static correction first estimates the near-surface velocity model and then calculates the static correction. The near-surface velocity model is merely an equivalent model of a certain velocity-thickness coupling in the solution space; on this equivalent model, the first arrival obtained by forward modeling can achieve a match within the error range with the observed first arrival. To meet the assumptions, stability, and convergence of the inversion calculation, first-arrival tomography techniques have performed numerous approximations on the model in practical applications, including a series of smoothing, interpolation, and extrapolation calculations. Near-surface velocity models obtained from initial tomography are generally the result of multiple iterations based on an initial layered medium, ultimately still exhibiting typical layered medium characteristics. They possess considerable smoothness in the horizontal direction. However, for complex work areas, the near-surface structure is far more complex than the model, such as limestone outcrops, faults, and lenticular bodies; these special structures bring about severe high-frequency residual static correction problems, which existing model methods struggle to accurately describe.
[0008] On the other hand, single-stage static correction based on near-surface velocity models cannot solve static correction problems caused by non-velocity factors. For example, differences in excitation and reception conditions between the shot and receiver, as well as differences in ray paths, often cause high-frequency residual static correction problems.
[0009] 2) Limitations of the linear reflected wave residual static correction method
[0010] In industry, traditional residual static correction methods for reflected waves are linear residual static correction methods, represented by cross-correlation time difference picking, also known as linear reflected wave residual static correction. These methods can only solve the problem of residual static correction for short wavelengths at high frequencies. Time difference picking is very difficult on data with low signal-to-noise ratios and large residual static correction values, resulting in ineffective residual static correction. Theoretically, when the residual static correction value is greater than half a period of the wavelet, the waveform distortion of the model channel is severe, the extrema of the cross-correlation function are not the true time delay, and cycle skipping occurs, causing linear reflected wave residual static correction to fail.
[0011] The reflection wave residual static correction method is used to solve the high-frequency static correction problem existing after the datum plane static correction, and it belongs to the data smoothing method. The biggest problem with the reflection wave residual static correction lies in its heavy dependence on the accuracy of velocity and also on the signal-to-noise ratio of seismic data. In areas with complex near-surface and low signal-to-noise ratio, there are great difficulties in the velocity analysis work itself. In the case of inaccurate velocity or too low signal-to-noise ratio, it is difficult for the reflection wave residual static correction to obtain the correct residual static correction amount. Summary of the Invention
[0012] The object of the present invention is to solve the above-mentioned problems existing in the prior art, and provide a non-linear optimization method, device, medium and equipment for residual static correction amount, which is used to realize non-linear reflection wave residual static correction, improve the residual static correction problem of complex near-surface, and is also an update and development of the traditional linear reflection wave residual static correction method.
[0013] The present invention is realized through the following technical solutions:
[0014] In the first aspect of the present invention, a non-linear optimization method for residual static correction amount is provided, which specifically includes the following steps:
[0015] The first step is to generate an initial population;
[0016] The second step is to calculate the fitness value of each individual;
[0017] The third step is to sort all individuals in descending order according to the fitness value;
[0018] The fourth step is to obtain the grid index of each shot point and geophone point according to the geodetic coordinates X and Y of the shot point and geophone point;
[0019] The fifth step is gene crossover to obtain individuals after gene crossover;
[0020] The sixth step is to calculate the fitness value of the individuals after gene crossover and sort the individuals after gene crossover in descending order according to the fitness value;
[0021] The seventh step is to compare the average value en2 of all fitness values in the sixth step with the average value en1 of all fitness values in the second step. If (en2 - en1) < en1 * 0.01, then enter the ninth step; otherwise, enter the eighth step;
[0022] The eighth step is to judge whether the current genetic generation is greater than the set maximum genetic generation N. If so, enter the ninth step; otherwise, replace the genes of the (RstMax / SampInt - 2)th group of individuals as a whole with the genes of the individual with the largest fitness value in the initial population, and return to the second step;
[0023] Step 9: Output the individual with the largest static correction value from step 6, which is the optimal result.
[0024] A further improvement of the present invention is that:
[0025] The first step is to generate the initial population, which includes the following steps:
[0026] (1) Assume that the maximum time shift for calculating the remaining static correction in the work area is RstMax, and the sampling interval is SampInt;
[0027] (2) The correction values of the shot points and receiver points in the entire work area are taken together as a single residual correction value individual, where each shot point and receiver point correction value is a single gene in the individual;
[0028] (3) For each shot point and receiver point, the random number generator rand is used to obtain an array of RstArr, which consists of RstMax / SampInt random numbers between [0,RstMax].
[0029] (4) Take out the i-th value from all random number arrays RstArr and combine them to form the i-th correction value individual; there are a total of RstMax / SampInt individuals, forming the initial population.
[0030] A further improvement of the present invention is that:
[0031] The second step is to calculate the fitness value for each individual. The specific operations include:
[0032] Each individual correction value is applied to the pre-stack CMP gather, and the absolute values of all trace amplitude values are summed to obtain the relevant stacking energy value of the gather, which is used as the fitness value of that individual.
[0033] A further improvement of the present invention is that:
[0034] The fourth step is to calculate the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y). Specific operations include:
[0035] The grid indexes are iline and icdp, where,
[0036] iline = (Y - Y0) / gridY, where iline takes values in the range [0, gridNLine].
[0037] icdp = (X - X0) / gridX, where the value range of icdp is [0, gridNTrace].
[0038] Where X0 and Y0 are the coordinates of the origin of the observation system grid, gridX is the grid spacing in the east-west direction, and gridY is the grid spacing in the north-south direction.
[0039] A further improvement of the present invention is that:
[0040] The fifth step is gene crossover, which yields individuals with the resulting gene crossover. Specific operations include:
[0041] For the i-th individual, a random number generator is used to generate a random number kcdpi within the range [0, gridNTrace] and a random number klinei within the range [0, gridNLine]. The random number seed is the current iteration number. After sorting, the individuals are subjected to gene exchange between adjacent individuals within the range defined by klinei, gridNLine, kcdpi, and gridNTrace. That is, the i-th individual and the (i+1)-th individual are exchanged within the ranges [klinei, gridNLine] and [kcdpi, gridNTrace] to obtain the gene-crossovered individual.
[0042] A second aspect of the present invention provides a nonlinear optimization device for residual static correction, comprising:
[0043] The initial population generation module is used to generate the initial population;
[0044] The computation module is used to calculate the fitness value for each individual.
[0045] The sorting module is used to sort all individuals in descending order of their fitness values;
[0046] The grid index retrieval module is used to retrieve the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y).
[0047] The gene crossover module is used for gene crossover to obtain individuals with the resulting gene crossover.
[0048] The sorting module is used to calculate the fitness value of individuals after gene crossover and sort the individuals in descending order of fitness value.
[0049] The first judgment module is used to compare the average value en2 of all fitness values obtained in the calculation and sorting module with the average value en1 of all fitness values obtained in the calculation module, and to determine whether (en2-en1) is less than en1*0.01;
[0050] The second judgment module is used to determine whether the current generation number is greater than the set maximum generation number N.
[0051] The output module outputs the individual with the largest static correction value from step six, which is the optimal result.
[0052] A further improvement of the present invention is that:
[0053] The grid index retrieval module is used to derive the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y). Specific operations include:
[0054] The grid indexes are iline and icdp, where,
[0055] iline = (Y - Y0) / gridY, where iline takes values in the range [0, gridNLine].
[0056] icdp = (X - X0) / gridX, where the value range of icdp is [0, gridNTrace].
[0057] Where X0 and Y0 are the coordinates of the origin of the observation system grid, gridX is the grid spacing in the east-west direction, and gridY is the grid spacing in the north-south direction.
[0058] A further improvement of the present invention is that:
[0059] The gene crossover module is used for gene crossover to obtain individuals with the resulting genes. Specific operations include:
[0060] For the i-th individual, a random number generator is used to generate a random number kcdpi within the range [0, gridNTrace] and a random number klinei within the range [0, gridNLine]. The random number seed is the current iteration number. After sorting, the individuals are subjected to gene exchange between adjacent individuals within the range defined by klinei, gridNLine, kcdpi, and gridNTrace. That is, the i-th individual and the (i+1)-th individual are exchanged within the ranges [klinei, gridNLine] and [kcdpi, gridNTrace] to obtain the gene-crossovered individual.
[0061] A third aspect of the present invention provides a computer-readable storage medium storing at least one computer-executable program, which, when executed by the computer, causes the computer to perform the steps in the nonlinear optimization method for residual static correction.
[0062] A fourth aspect of the present invention provides a computer device including a memory and a processor, the memory storing a computer program, characterized in that the processor executes the computer program to implement the steps of the nonlinear optimization method for residual static correction.
[0063] Compared with the prior art, the beneficial effects of the present invention are:
[0064] This invention introduces a genetic algorithm into the optimization of the residual static correction of reflected waves, which can solve the residual static correction of reflected waves. It has good noise resistance and accuracy, and largely overcomes the local extremum trap and wavelet half-period limitation of the correction amount in traditional linear residual static correction, thus improving the imaging quality of seismic data. Attached Figure Description
[0065] Figure 1 This is a flowchart of a nonlinear optimization method for residual static correction in an embodiment of the present invention;
[0066] Figure 2 The image shows the pre-stack common midpoint seismic data after static correction, linear reflection residual static correction, and dynamic correction.
[0067] Figure 3 This is a diagram illustrating the effect of applying the optimized static correction parameters of this invention to pre-stack common center point seismic data. Detailed Implementation
[0068] The present invention will now be described in further detail with reference to the accompanying drawings:
[0069] Genetic Algorithm (GA) is a computational model that simulates Darwin's natural selection process of biological evolution. In 1975, Professor Holland and others at the University of Michigan proposed the pattern theory, which was extremely important for the theoretical research of genetic algorithms, and published the monograph "Adaptation of Natural and Artificial Systems," which systematically expounded the basic theory and methods of genetic algorithms, laying the foundation for genetic algorithms. As a global optimization search algorithm, genetic algorithm is simple, universal, robust, suitable for parallel processing, and has a wide range of applications.
[0070] The process of genetic algorithm computation optimization is similar to the biological process of biological genetic evolution, and has at least three basic operations (or operators): selection, crossover, and mutation.
[0071] (1) Selection
[0072] Selection, also known as replication, involves choosing individuals from a population that are better adapted to the environment based on their fitness function values. Generally, selection results in individuals with high fitness reproducing more offspring, while individuals with low fitness reproduce fewer offspring or are even eliminated. The most common implementation method is the roulette wheel model.
[0073] (2) Cross
[0074] The crossover operator crosses the gene chains of two selected individuals with a certain probability, thereby generating two new individuals. The crossover position is random. Depending on the problem, crossover can be divided into single-point crossover operator, two-point crossover operator, and uniform crossover operator.
[0075] (3) Variation
[0076] Among the selected individuals, the bits of the new individual's gene chain are transformed in opposite directions according to probability. The simplest way is to change the value at a certain position in the string. Taking binary encoding as an example, this means swapping 0 and 1: 0 mutates into 1, and 1 mutates into 0.
[0077] (4) Eliteism
[0078] Selecting genes solely from the offspring to construct a new population may result in the loss of much information from the previous generation. In other words, when using crossover and mutation to generate a new generation, there is a high probability of losing the optimal solution obtained at some intermediate step. To address this, we use an elitist approach: each time a new generation is generated, the current optimal solution is copied verbatim into the new generation, while other steps remain unchanged. This ensures that any optimal solution generated at any given time can survive until the genetic algorithm terminates.
[0079] The implementations of the aforementioned operators are diverse, and many new operators are constantly being proposed to improve certain performance aspects of genetic algorithms. For example, selection algorithms include hierarchical equilibrium selection, and so on.
[0080] The basic steps of a genetic algorithm are as follows: First, the solution to the problem is represented as "chromosomes," which are binary encoded strings in the algorithm. Before executing the genetic algorithm, a set of "chromosomes" is given, representing hypothetical feasible solutions. Then, the hypothetical feasible solutions are placed in the problem's "environment," and according to the principle of survival of the fittest, the "chromosomes" that are better adapted to the environment are selected and replicated. Through crossover and mutation processes, a new generation of "chromosomes" that are even better adapted to the environment is generated. After this evolutionary process, it eventually converges to the "chromosome" that is best adapted to the environment, which is the optimal solution to the problem. The specific steps are as follows:
[0081] Step 1: Select an encoding strategy and transform the parameter set (feasible solution set) into chromosome structure space;
[0082] Step 2: Define the fitness function to facilitate the calculation of the fitness value;
[0083] Step 3: Determine the genetic strategy, including the population size, selection, crossover, and mutation methods, and determine genetic parameters such as crossover probability and mutation probability;
[0084] Step 4: Randomly generate the initial population;
[0085] Step 5: Calculate the fitness value after decoding the individuals or chromosomes in the population;
[0086] Step 6: According to the genetic strategy, apply the selection, crossover, and mutation operators to the population to form the next generation population;
[0087] Step 7: Determine whether the population performance meets the criteria or whether the predetermined number of iterations has been completed. If not, return to Step 5 or modify the genetic strategy and then return to Step 6;
[0088] Step 8: Output the individual with the optimal fitness value in the population.
[0089] The simplest stopping conditions for the algorithm are as follows: Stop when the pre-given number of generations of evolution is completed; Stop when the optimal individual in the population has not improved for several consecutive generations or the average fitness has not improved significantly for several consecutive generations.
[0090] Based on this, the present invention provides a method for optimizing the residual static correction amount, which globally optimizes the residual static correction amount of the reflected wave based on the genetic algorithm. The embodiments of the method are as follows:
[0091] [Embodiment 1]
[0092] As Figure 1 shown, the specific steps of the method include:
[0093] Step 1, generate an initial population;
[0094] Step 2, calculate the fitness value of each individual;
[0095] Step 3, sort all individuals in descending order according to their fitness values;
[0096] Step 4, obtain the grid index of each shot point and geophone point according to the geodetic coordinates X and Y of the shot point and geophone point;
[0097] Step 5, gene crossover to obtain the individuals after gene crossover;
[0098] Step 6, calculate the fitness value of the individuals after gene crossover and sort the individuals after gene crossover in descending order according to their fitness values;
[0099] Step 7, compare the average value en2 of all fitness values in Step 6 with the average value en1 of all fitness values in Step 2. If (en2 - en1) < en1 * 0.01, enter Step 9; otherwise, enter Step 8;
[0100] Step 8: Determine if the current generation is greater than the set maximum generation N. If yes, proceed to step 9. Otherwise, replace the genes of all individuals in the RstMax / SampInt-2 group with the genes of the individual with the highest fitness value in the initial population and return to step 2.
[0101] Step 9: Output the individual with the largest static correction value from step 6, which is the optimal result.
[0102] The residual static correction problem is essentially a nonlinear global optimization problem with multiple parameters and extrema, and should therefore be solved using stochastic global optimization (nonlinear optimization) algorithms. Nonlinear optimization methods, when searching for the optimal solution, can automatically discard local optima that may be caused by periodic jumps, converging to the true global optimum. These characteristics of nonlinear algorithms demonstrate great application potential in complex near-surface residual static correction problems, representing a significant update and development of traditional linear reflected wave residual static correction techniques.
[0103] Compared with the traditional linear residual static correction method for reflected waves, the nonlinear optimization method for residual static correction of reflected waves in this invention has the following characteristics: 1) The calculated static correction can exceed 1 / 2 period of the wavelet; 2) Due to its strong noise resistance, it performs better than the linear algorithm on low signal-to-noise ratio and large static correction data.
[0104] This invention overcomes the shortcomings of traditional linear residual static correction methods and improves the imaging quality of seismic data.
[0105]
Example 2
[0106] The first step is to generate the initial population, which includes the following steps:
[0107] (1) Assume that the maximum time shift for calculating the remaining static correction in the work area is RstMax, and the sampling interval is SampInt;
[0108] (2) The correction values of the shot points and receiver points in the entire work area are taken together as a single residual correction value individual, where each shot point and receiver point correction value is a single gene in the individual;
[0109] (3) For each shot point and receiver point, the random number generator rand is used to obtain an array of RstArr, which consists of RstMax / SampInt random numbers between [0,RstMax].
[0110] (4) Take out the i-th value from all random number arrays RstArr and combine them to form the i-th correction value individual; there are a total of RstMax / SampInt individuals, forming the initial population.
[0111]
Example 3
[0112] The second step is to calculate the fitness value for each individual. The specific operations include:
[0113] Each individual correction value is applied to the pre-stack CMP gather, and the absolute values of all trace amplitude values are summed to obtain the relevant stacking energy value of the gather, which is used as the fitness value of that individual.
[0114]
Example 4
[0115] The fourth step is to calculate the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y). Specific operations include:
[0116] The grid indexes are iline and icdp, where,
[0117] iline = (Y - Y0) / gridY, where iline takes values in the range [0, gridNLine].
[0118] icdp = (X - X0) / gridX, where the value range of icdp is [0, gridNTrace].
[0119] Where X0 and Y0 are the coordinates of the origin of the observation system grid, gridX is the grid spacing in the east-west direction, and gridY is the grid spacing in the north-south direction.
[0120]
Example 5
[0121] The fifth step is gene crossover, which yields individuals with the resulting gene crossover. Specific operations include:
[0122] For the i-th individual, a random number generator is used to generate a random number kcdpi within the range [0, gridNTrace] and a random number klinei within the range [0, gridNLine]. The random number seed is the current iteration number. After sorting, the genes (residual corrections) between adjacent individuals are exchanged according to the range defined by klinei, gridNLine, kcdpi, and gridNTrace. That is, the i-th individual and the (i+1)-th individual are exchanged within the ranges [klinei, gridNLine] and [kcdpi, gridNTrace] to obtain the gene-crossed individual.
[0123]
Example 6
[0124] Pre-stack common midpoint seismic data after one static correction, linear reflection residual static correction, and dynamic correction, such as... Figure 2 As shown, the residual static correction was optimized using the method of this invention on the pre-stack common center point seismic data. The optimized residual static correction was then applied to the common center point data to verify the stacking effect. The results are as follows. Figure 3 As shown. From Figure 2 and Figure 3 As can be seen, compared with the application of this invention, the continuity of the phase axis of the reflected wave in the CMP superimposed profile after the application of this invention is significantly enhanced, the energy focusing is better, and the characterization of the underground structure is clearer.
[0125]
Example 7
[0126] This invention provides a nonlinear optimization device for residual static correction, comprising:
[0127] The initial population generation module is used to generate the initial population. Specific operations include:
[0128] (1) Assume that the maximum time shift for calculating the remaining static correction in the work area is RstMax, and the sampling interval is SampInt;
[0129] (2) The correction values of the shot points and receiver points in the entire work area are taken together as a single residual correction value individual, where each shot point and receiver point correction value is a single gene in the individual;
[0130] (3) For each shot point and receiver point, the random number generator rand is used to obtain an array of RstArr, which consists of RstMax / SampInt random numbers between [0,RstMax].
[0131] (4) Take out the i-th value from all random number arrays RstArr and combine them to form the i-th correction value individual; there are a total of RstMax / SampInt individuals, forming the initial population;
[0132] The calculation module is used to calculate the fitness value of each individual. Specific operations include:
[0133] Each individual correction value is applied to the pre-stack CMP gather, and the absolute values of all trace amplitude values are summed to obtain the relevant stacking energy value of the gather, which is used as the fitness value of that individual.
[0134] The sorting module is used to sort all individuals in descending order of their fitness values;
[0135] The grid index retrieval module is used to derive the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y). Specific operations include:
[0136] The grid indexes are iline and icdp, where,
[0137] iline = (Y - Y0) / gridY, where iline takes values in the range [0, gridNLine].
[0138] icdp = (X - X0) / gridX, where the value range of icdp is [0, gridNTrace].
[0139] Where X0 and Y0 are the coordinates of the origin of the observation system grid, gridX is the grid spacing in the east-west direction, and gridY is the grid spacing in the north-south direction.
[0140] The gene crossover module is used for gene crossover to obtain crossover individuals. For the i-th individual, a random number generator is used to generate a random number kcdpi in the range [0, gridNTrace] and a random number klinei in the range [0, gridNLine]. The random number seed is the current iteration number. After sorting, the genes of adjacent individuals are exchanged according to the range defined by klinei, gridNLine, kcdpi, and gridNTrace. That is, the i-th individual and the (i+1)-th individual are exchanged within the ranges [klinei, gridNLine] and [kcdpi, gridNTrace] to obtain crossover individuals.
[0141] The sorting module is used to calculate the fitness value of individuals after gene crossover and sort the individuals in descending order of fitness value.
[0142] The first judgment module is used to compare the average value en2 of all fitness values obtained in the calculation and sorting module with the average value en1 of all fitness values obtained in the calculation module, and to determine whether (en2-en1) is less than en1*0.01;
[0143] The second judgment module is used to determine whether the current generation number is greater than the set maximum generation number N.
[0144] The output module outputs the individual with the largest static correction value from step six, which is the optimal result.
[0145]
Example 8
[0146] This invention provides a computer-readable storage medium storing at least one computer-executable program. When executed by the computer, the at least one program causes the computer to perform steps in the nonlinear optimization method for residual static correction, the steps including:
[0147] The first step is to generate an initial population;
[0148] The second step is to calculate the fitness value for each individual;
[0149] The third step is to sort all individuals in descending order of their fitness values;
[0150] The fourth step is to obtain the grid index of each shot point and geophone point according to the geodetic coordinates X and Y of the shot point and geophone point;
[0151] The fifth step is gene crossover to obtain the individuals after gene crossover;
[0152] The sixth step is to calculate the fitness value of the individuals after gene crossover and sort the individuals after gene crossover in descending order of fitness value;
[0153] The seventh step is to compare the average value en2 of all fitness values in the sixth step with the average value en1 of all fitness values in the second step. If (en2 - en1) < en1 * 0.01, go to the ninth step; otherwise, go to the eighth step;
[0154] The eighth step is to determine whether the current genetic generation is greater than the set maximum genetic generation N. If so, go to the ninth step; otherwise, replace the genes of the (RstMax / SampInt - 2)-th group of individuals with the genes of the individual with the largest fitness value in the initial population, start the next generation of inheritance, and return to the second step;
[0155] The ninth step is to output the individual with the largest static correction amount of fitness value in the sixth step, which is the optimal result.
[0156]
Example 9
[0157] An embodiment of the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the remaining static correction amount non-linear optimization method are implemented. The steps include:
[0158] The first step is to generate an initial population;
[0159] The second step is to calculate the fitness value of each individual;
[0160] The third step is to sort all individuals in descending order of fitness value;
[0161] The fourth step is to obtain the grid index of each shot point and geophone point according to the geodetic coordinates X and Y of the shot point and geophone point;
[0162] The fifth step is gene crossover to obtain the individuals after gene crossover;
[0163] The sixth step is to calculate the fitness value of the individuals after gene crossover and sort the individuals after gene crossover in descending order of fitness value;
[0164] Step 7: Compare the average value en2 of all fitness values in Step 6 with the average value en1 of all fitness values in Step 2. If (en2 - en1) < en1 * 0.01, proceed to Step 9; otherwise, proceed to Step 8.
[0165] Step 8: Determine whether the current genetic generation is greater than the set maximum genetic generation N. If so, proceed to Step 9; otherwise, globally replace the genes of the (RstMax / SampInt - 2)-th group of individuals with the genes of the individual with the maximum fitness value in the initial population, start the next generation of inheritance, and return to Step 2.
[0166] Step 9: Output the static correction amount individual with the maximum fitness value in Step 6, which is the optimal result.
[0167] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0168] The above technical solution is only one implementation manner of the present invention. For those skilled in the art, based on the disclosed principles of the present invention, it is easy to make various types of improvements or modifications, not limited to the technical solutions described in the above specific embodiments of the present invention. Therefore, the foregoing description is only preferred and does not have a restrictive meaning.
Claims
1. A nonlinear optimization method for residual static correction, characterized in that, Specifically, it includes the following steps: The first step is to generate an initial population; The second step is to calculate the fitness value of each individual; The third step is to sort all individuals in descending order according to their fitness values; The fourth step is to obtain the grid index of each shot point and geophone point based on the geodetic coordinates X and Y of the shot points and geophone points; The fifth step is gene crossover to obtain individuals after gene crossover; The sixth step is to calculate the fitness value of the individuals after gene crossover and sort the individuals after gene crossover in descending order according to their fitness values; The seventh step is to compare the average value en2 of all fitness values in the sixth step with the average value en1 of all fitness values in the second step. If (en2 - en1) < en1 * 0.01, then go to the ninth step; otherwise, go to the eighth step; The eighth step is to determine whether the current genetic generation is greater than the set maximum genetic generation N. If so, go to the ninth step; otherwise, replace the genes of the (RstMax / SampInt - 2) -th group of individuals with the genes of the individual with the maximum fitness value in the initial population and return to the second step; The ninth step is to output the static correction amount individual with the maximum fitness value in the sixth step as the optimal result.
2. The method according to claim 1, characterized in that, The first step is to generate an initial population. The specific operations include: (1) Assume that the maximum time shift for calculating the remaining static correction in the work area is RstMax and the sampling interval is SampInt; (2) Take the correction amounts of all shot points and geophone points in the entire work area as a single remaining correction amount individual, where the correction amount of each shot point and geophone point is a single gene in the individual; (3) For each shot point and geophone point, use the random number generator rand to obtain an array of random numbers RstArr with RstMax / SampInt numbers between [0, RstMax]; (4) Take out the i -th value in all the random number arrays RstArr and combine them into the i -th correction amount individual; there are RstMax / SampInt individuals in total, forming the initial population.
3. The method according to claim 2, characterized in that, The second step is to calculate the fitness value of each individual. The specific operations include: Apply each correction amount individual to the prestack CMP gather, and perform an accumulative summation calculation on the absolute values of all trace amplitudes to obtain the related stacking energy value of the gather as the fitness value of this individual.
4. The method according to claim 3, characterized in that, The fourth step is to obtain the grid index of each shot point and geophone point based on the geodetic coordinates X and Y of the shot points and geophone points. The specific operations include: The grid indices are iline and icdp, where, iline = (Y - Y0) / gridY, and the value range of iline is [0, gridNLine], icdp = (X - X0) / gridX, and the value range of icdp is [0, gridNTrace], where X0 and Y0 are the coordinates of the grid origin of the acquisition system, gridX is the interval of the grid in the east - west direction, and gridY is the interval of the grid in the north - south direction.
5. The method according to claim 4, characterized in that, The fifth step is gene crossover to obtain individuals after gene crossover. The specific operations include: For the i-th individual, a random number generator is used to generate a random number kcdpi within the range [0, gridNTrace] and a random number klinei within the range [0, gridNLine]. The random number seed is the current iteration number. After sorting, the individuals are subjected to gene exchange between adjacent individuals within the range defined by klinei, gridNLine, kcdpi, and gridNTrace. That is, the i-th individual and the (i+1)-th individual are exchanged within the ranges [klinei, gridNLine] and [kcdpi, gridNTrace] to obtain the gene-crossovered individual.
6. A nonlinear optimization device for residual static correction, characterized in that, include: The initial population generation module is used to generate the initial population; The computation module is used to calculate the fitness value for each individual. The sorting module is used to sort all individuals in descending order of their fitness values; The grid index retrieval module is used to retrieve the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y). The gene crossover module is used for gene crossover to obtain individuals with the resulting gene crossover. The sorting module is used to calculate the fitness value of individuals after gene crossover and sort the individuals in descending order of fitness value. The first judgment module is used to compare the average value en2 of all fitness values obtained in the calculation and sorting module with the average value en1 of all fitness values obtained in the calculation module, and to determine whether (en2-en1) is less than en1*0.01; The second judgment module is used to determine whether the current generation number is greater than the set maximum generation number N. The output module outputs the individual with the largest static correction value from step six, which is the optimal result.
7. The apparatus according to claim 6, characterized in that, The grid index retrieval module is used to retrieve the grid index of each shot point and receiver point based on their geodetic coordinates (X, Y). Specific operations include: The grid indexes are iline and icdp, where, iline = (Y - Y0) / gridY, where iline takes values in the range [0, gridNLine]. icdp = (X - X0) / gridX, where the value range of icdp is [0, gridNTrace]. Where X0 and Y0 are the coordinates of the origin of the observation system grid, gridX is the grid spacing in the east-west direction, and gridY is the grid spacing in the north-south direction.
8. The apparatus according to claim 6, characterized in that, The gene crossover module is used for gene crossover to obtain individuals with the resulting genes. Specific operations include: For the i-th individual, a random number generator is used to generate a random number kcdpi within the range [0, gridNTrace] and a random number klinei within the range [0, gridNLine]. The random number seed is the current iteration number. After sorting, the individuals are subjected to gene exchange between adjacent individuals within the range defined by klinei, gridNLine, kcdpi, and gridNTrace. That is, the i-th individual and the (i+1)-th individual are exchanged within the ranges [klinei, gridNLine] and [kcdpi, gridNTrace] to obtain the gene-crossovered individual.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer-executable program, which, when executed by the computer, causes the computer to perform the steps in the nonlinear optimization method for residual static correction as described in any one of claims 1-5.
10. A computer device, characterized in that, The device includes a memory and a processor, the memory storing a computer program, characterized in that the processor executes the computer program to implement the steps of the nonlinear optimization method for residual static correction as described in any one of claims 1-5.