CN113534246B - Pre-stack AVO inversion method based on bee colony optimization algorithm - Google Patents
Pre-stack AVO inversion method based on bee colony optimization algorithm Download PDFInfo
- Publication number
- CN113534246B CN113534246B CN202010302615.5A CN202010302615A CN113534246B CN 113534246 B CN113534246 B CN 113534246B CN 202010302615 A CN202010302615 A CN 202010302615A CN 113534246 B CN113534246 B CN 113534246B
- Authority
- CN
- China
- Prior art keywords
- inversion
- avo
- stack
- honey source
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000005457 optimization Methods 0.000 title claims abstract description 36
- 235000012907 honey Nutrition 0.000 claims abstract description 58
- 241000257303 Hymenoptera Species 0.000 claims description 30
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 6
- 230000002068 genetic effect Effects 0.000 claims description 5
- 239000002245 particle Substances 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000003384 imaging method Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 12
- 238000000605 extraction Methods 0.000 abstract description 5
- 230000000704 physical effect Effects 0.000 abstract description 3
- 241000256844 Apis mellifera Species 0.000 description 49
- 230000006870 function Effects 0.000 description 36
- 230000006399 behavior Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 9
- 238000011161 development Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 238000011160 research Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000009395 breeding Methods 0.000 description 3
- 230000001488 breeding effect Effects 0.000 description 3
- 230000013011 mating Effects 0.000 description 3
- 230000001737 promoting effect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000010845 search algorithm Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 244000207740 Lemna minor Species 0.000 description 1
- 235000006439 Lemna minor Nutrition 0.000 description 1
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 235000001855 Portulaca oleracea Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000010429 evolutionary process Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. analysis, for interpretation, for correction
- G01V1/30—Analysis
- G01V1/306—Analysis for determining physical properties of the subsurface, e.g. impedance, porosity or attenuation profiles
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. analysis, for interpretation, for correction
- G01V1/30—Analysis
- G01V1/303—Analysis for determining velocity profiles or travel times
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/60—Analysis
- G01V2210/62—Physical property of subsurface
- G01V2210/622—Velocity, density or impedance
- G01V2210/6222—Velocity; travel time
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/60—Analysis
- G01V2210/62—Physical property of subsurface
- G01V2210/622—Velocity, density or impedance
- G01V2210/6224—Density
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V2210/00—Details of seismic processing or analysis
- G01V2210/60—Analysis
- G01V2210/62—Physical property of subsurface
- G01V2210/622—Velocity, density or impedance
- G01V2210/6226—Impedance
Abstract
The invention provides a pre-stack AVO inversion method based on a bee colony optimization algorithm, which comprises the following steps: step 1, inputting a three-dimensional pre-stack seismic data volume and a seismic wavelet; step 2, setting an initial honey source, namely constructing an inversion initial model; step 3, calculating an objective function value according to the constructed AVO inversion equation based on the Bayesian theory; step 4, calculating a honey source fitness value according to the relation between the AVO inversion objective function and the bee colony algorithm fitness function; step 5, applying an improved bee colony algorithm to search the neighborhood to find the optimal solution; and step 6, obtaining an inversion optimal solution when the maximum search times are reached or the iteration termination condition is met, and realizing reservoir physical property parameter extraction. The pre-stack AVO inversion method based on the bee colony optimization algorithm fuses the improved artificial bee colony algorithm and AVO inversion based on the Bayes theory and the accurate Zoeppritz equation, and provides an effective and reliable reservoir physical parameter pre-stack AVO nonlinear inversion technology.
Description
Technical Field
The invention relates to the technical field of oilfield development, in particular to a pre-stack AVO inversion method based on a bee colony optimization algorithm.
Background
The research of the reservoir geophysical inversion can be generally divided into linear inversion and nonlinear inversion, wherein the linear inversion has long development time, is the inversion method with the most complete theory, the most extensive application and the most mature technology in the geophysical inversion, and the nonlinear inversion starts later, and the development in the geophysical inversion is behind the linear inversion. Most of the problems faced in geophysical inversion are nonlinear inversion problems, however, since nonlinear inversion is much more complex and difficult than linear inversion, nonlinear inversion problems are typically linearized or pseudo-linearized in solving an objective function and then linear inversion is performed. However, the linear inversion method has the problems of monotonous algorithm, more local search, strong dependence on an initial model and the like, and if the initial model is selected improperly, the search is often trapped in local minima or even is not converged. Therefore, the inversion problem is solved by using a conventional linear inversion method, so that a local optimal solution is often trapped, and the inversion result is unreliable. Meanwhile, various linear inversion methods are difficult to surmount against the obstacles and difficulties encountered when facing the multi-parameter and multi-polar geophysical nonlinear optimization inversion problem. In order to improve reliability of reservoir physical property parameter pre-stack AVO inversion, in recent years, a plurality of nonlinear inversion methods based on intelligent optimization algorithms, such as simulated annealing, genetic algorithms, particle swarm optimization, ant colony algorithm and the like, appear at home and abroad, and the intelligent optimization algorithms have the advantages of being independent or weakly dependent on an initial model, not needing to solve complex Jacobian matrix, being simpler in algorithm implementation and the like, and have been widely applied to geophysical inversion research.
The pre-stack AVO inversion is an effective means for researching formation lithology and oil-gas content by utilizing the law of the variation of seismic reflection amplitude along with offset, has the obvious advantages of stable result, high resolution, strong controllability and the like, is widely applied to oil-gas exploration, but the technology is subject to the problems of low inversion efficiency, low precision and the like due to the influence of various noises and uncertain factors in the acquisition process, and the key for promoting the development of the pre-stack AVO inversion technology to obtain accurate reservoir physical parameters is to explore an effective pre-stack AVO inversion technical process and an efficient inversion algorithm.
The development of pre-stack AVO inversion has been carried out in more than ten years, a plurality of achievements are obtained in the aspect of petroleum geophysical exploration, and a plurality of classical method technologies of the aspect appear, for example, ma (2002) proposes to integrate AVO attribute extraction and wave impedance inversion into one step, which means that pre-stack P wave data can be directly converted into rock attribute, and the step of estimating P wave reflection coefficient and S wave reflection coefficient through AVO analysis is skipped; buland et al (2003) based on Bayesian theory, developed a series of studies using prestack seismic data to derive inversion equations for longitudinal and transverse wave velocities and densities from joint distribution of model parameters and observed data; mogense (2001) performs pre-stack AVO inversion by using a neural network technology, directly estimates longitudinal wave and transverse wave speeds from pre-stack seismic data, calculates poisson ratio, and has good agreement between inversion results and true values; based on former research, russell (2005) establishes a new method, namely, a prestack AVO synchronous inversion formula is given by combining a post-stack wave impedance formula derivation process, longitudinal wave impedance, transverse wave impedance and density are directly inverted based on the formula, a good inversion effect is obtained, and a part of Strata business software about prestack AVO synchronous inversion is based on the research thought. In the aspect of nonlinear research of pre-stack AVO, wei Chaodeng (2011) introduces a quantum Monte Carlo method into nonlinear three-parameter inversion of pre-stack AVO, the random search algorithm can better find a global optimal solution, the stability is better, the precision of pre-stack inversion is effectively improved, and a better application effect is obtained; Duckweed et al (2011) provides a novel AVO nonlinear inversion method based on a support vector machine, and the method has the advantages of high inversion speed, good stability and the like. Although pre-stack AVO inversion is an emerging technology in the field of seismic exploration and is widely applied to oil and gas exploration, the technology still faces the problems of low inversion efficiency, low inversion precision, particularly insufficient precision of density inversion results and the like at present, so that the exploration of an effective pre-stack AVO inversion technical process and an efficient inversion algorithm is a direction for promoting the development of the pre-stack AVO technology.
The bee colony algorithm is an optimization method provided by simulating bee behaviors, is a specific application of the intelligent cluster thought, and is mainly characterized in that special information of problems is not needed to be known, only the problems are compared in quality, and the global optimal value is finally highlighted in a colony through the local optimizing behaviors of each artificial bee individual, so that the method has a higher convergence rate. According to the different behaviors of the inspired bees, the bee colony algorithm is mainly divided into two major categories, namely a breeding behavior and a honey collecting behavior. The bee colony algorithm based on the bee propagation behavior divides the bee colony into three groups of queen bees, drones and worker bees, the generation of new queen bees is similar to an optimization process in evolutionary computation, the queen bees are the optimal solutions of problems to be solved in the optimization process, and the bee colony propagation evolutionary process is a process for continuously updating the queen bees. The most common bee colony algorithm based on bee breeding behavior is the bee mating optimization algorithm proposed by abmass in 2001 (honey bee mating optimization, HBMO for short), which simulates the entire process of proceeding from one queen bee to a colony comprising one or more queen bees. Amiri et al (2007) combine self-organizing map neural networks with HBMOs for subdividing online bookstore markets based on K-means clustering algorithms; niknam (2011) combines HBMO with a chaotic local search and fuzzy clustering technology respectively to solve the reconstruction problem of a multi-target power distribution network; yang et al (2009) optimized military deployment of weapon networking systems with improved mating HBMO; marinakis et al (2010) combine HBMO with a multi-level neighborhood search algorithm to solve the vehicle path problem; horng (2009, 2010, 2011) combines HBMO with maximum entropy techniques, multi-level minimum cross entropy methods, linde-Buzo-Gray algorithms, respectively, for finding image boundaries and constructing a codebook for vector quantization. Besides HBMO, bee colony algorithms based on bee breeding behaviors include bee evolution type genetic algorithms, queen bee algorithms and the like. The most common bee colony algorithm based on honey taking action of bees is the artificial bee colony algorithm (artificial bee colony algorithm, ABC for short) proposed by Karaboga in 2005. Karaboga (2005) trains neural networks with ABC, cluster analysis and solves constraint optimization problems; kang et al (2009) combined ABC with the Nelder-Mead simplex method to solve the inverse analysis problem; pulikanti et al (2009) combine ABC with greedy heuristics and local search to solve the quadratic taper problem; rebeyend et al (2010) combine ABC with the greedy approach to optimize multiprocessor scheduling; omkar et al (2011) optimized multi-objective composite designs with parallel vector estimation ABC; jeya et al (2010) tested localization with ABC optimization software for parallel search of three bees. The characteristics of honey collection behavior, learning, memory and information sharing of bees become one of the research hotspots of group intelligence, and the application prospect is quite broad.
Therefore, the invention provides a novel pre-stack AVO inversion method based on a bee colony optimization algorithm, and solves the technical problems.
Disclosure of Invention
The invention aims to provide a pre-stack AVO inversion method based on a bee colony optimization algorithm for an effective and reliable reservoir physical parameter pre-stack AVO nonlinear inversion technology.
The aim of the invention can be achieved by the following technical measures: the pre-stack AVO inversion method based on the swarm optimization algorithm comprises the following steps: step 1, inputting a three-dimensional pre-stack seismic data volume and a seismic wavelet; step 2, setting an initial honey source, namely constructing an inversion initial model; step 3, calculating an objective function value according to the constructed AVO inversion equation based on the Bayesian theory; step 4, calculating a honey source fitness value according to the relation between the AVO inversion objective function and the bee colony algorithm fitness function; step 5, applying an improved bee colony algorithm to search the neighborhood to find the optimal solution; and step 6, obtaining an inversion optimal solution when the maximum search times are reached or the iteration termination condition is met, and realizing reservoir physical property parameter extraction.
The aim of the invention can be achieved by the following technical measures:
in step 1, the input pre-stack channel set is used as a standard SEGY format file which is subjected to optimization processing, and the input seismic wavelets are theoretical wavelets or extracted seismic wavelets.
In the step 1, the seismic data acquired by the industry are shot records, a CMP gather is obtained after processing, the CMP gather is converted into an angle domain common imaging point gather, and the angle gather is extracted by adopting a method based on a ray theory; preprocessing logging data, carrying out time-depth conversion, calibrating the logging data on the seismic data, and further extracting seismic wavelets.
In the step 2, setting a physical parameter initial model under the condition of conforming to a certain probability distribution, namely setting initial honey sources, wherein the number of the honey sources is represented by SN.
In step 2, algorithm parameters are set, wherein the parameters comprise the size of the bee colony, the maximum searching times of bees, the upper and lower boundaries of the bee searching space and iteration termination conditions.
In step 3, using the input prestack gather seismic record d and model m, according to the constructed inversion targets
Calculating objective function values, wherein parameters including weight coefficient epsilon of model constraint item need to be set 2 Low and lowFrequency model m r Weighting matrix W m The method comprises the steps of carrying out a first treatment on the surface of the The first term in the formula represents a data error term and controls inversion accuracy; the second term represents the prior distribution constraint term of inversion parameters, controls the sparsity of inversion, sigma n 2 The larger the inversion result is, the higher the sparsity is, otherwise, the inversion result has obvious band-limited characteristics; the third term represents a model constraint term, low-frequency information is compensated for inversion, and stability and accuracy of inversion are further improved.
In step 4, AVO inversion is combined with the swarm algorithm, and the correspondence shown in the following formula is used according to the relationship between the fitness function and the objective function, wherein fit i Is the fitness value of the ith honey source,i=1, 2,3, SN for the objective function value corresponding to the i-th honey source; the corresponding fitness value is calculated for the objective function using the equation,
in step 5, the neighborhood search is performed according to the improved swarm algorithm combining the particle swarm and the genetic algorithm to update the honey source position, and various parameters are set.
In step 6, when the neighborhood maximum search number or iteration termination condition is not reached, a new honey source position is generated and the inversion objective function value is recalculated in step 3.
The invention discloses a pre-stack AVO inversion method based on a bee colony optimization algorithm, which aims at solving the problems that the existing pre-stack AVO inversion method is low in efficiency and low in precision and is difficult to accurately predict effective reservoir distribution, and integrates an improved artificial bee colony algorithm, AVO inversion based on a Bayes theory and an accurate Zoeppritz equation, so that an effective and reliable reservoir physical parameter pre-stack AVO nonlinear inversion technology is provided.
Drawings
FIG. 1 is a flow chart of a specific embodiment of a pre-stack AVO inversion method based on a swarm optimization algorithm of the present invention;
FIG. 2 is a graph showing the relationship between fitness function and objective function according to an embodiment of the present invention;
FIG. 3 is a graph showing the comparison of the inversion results when the weight coefficients of the Cauchy constraint terms in the AVO inversion of the present invention are different;
FIG. 4 is a graph comparing AVO inversion result synthetic seismic records of the present invention with actual seismic records;
FIG. 5 is a graph comparing the inversion results of the present invention with actual well data.
Detailed Description
The foregoing and other objects, features and advantages of the invention will be apparent from the following more particular description of preferred embodiments, as illustrated in the accompanying drawings.
As shown in fig. 1, fig. 1 is a flowchart of a pre-stack AVO inversion method based on a swarm optimization algorithm according to the present invention.
The method comprises the following specific steps:
1. (step 101) inputting a record of the seismic angle gather and a seismic wavelet. The seismic data acquired in the industry are shot records, CMP gathers are obtained after processing, in order to meet the data requirement of AVO inversion, the CMP gathers are converted into angle domain common imaging point gathers, and the angle gathers are extracted by adopting a method based on a ray theory; preprocessing logging data, carrying out time-depth conversion, calibrating the logging data on the seismic data, and further extracting seismic wavelets.
2. (step 102) initializing the honey source location, i.e., establishing an initial model of the inversion. The algorithm parameters are set, and the parameters mainly comprise the size of the bee colony, the maximum searching times of bees, the upper and lower boundaries of the bee searching space, iteration termination conditions and the like. And combining the extraction requirement of physical parameters, and formulating each parameter according to the following rule.
(1) Size of bee colony
The size of the bee colony generally defines that the number of the collected bees and the observed bees respectively account for half, and the number of the collected bees is equal to that of honey sources. The size of the bee colony is an integer parameter, and if the value is too small, the algorithm is trapped in local preference with high probability; if the value is too large, the optimization capacity of the algorithm is good, but the convergence speed is low; when the colony size increases to a certain level, it will no longer have a significant effect to continue to increase the colony number. The size of the bee colony is selected to be a proper value according to the volume of the AVO inverted data.
(2) Maximum number of searches for bees
The effect of the maximum number of bees is to discard the honey source if the number of bees search exceeds the limit value and no honey source with higher nectar content is found, and then to convert the honey source into a detection bee, and randomly generating a new honey source in a global range. The mechanism can lead the algorithm to better jump out of an unsuitable searching range and avoid sinking into local optimum.
(3) Upper and lower bounds of bee search space
The upper and lower bounds of the bee search space are used for mainly restricting the bee search range, and after the position of the honey source is updated, whether the position exceeding the bee search space range exists or not is judged, and if the position exists, the position is replaced by the position of the search boundary. Generally, the more accurate the range of the bee search space is obtained, the better the stability of the inversion, and the more accurate the result.
(4) Iteration termination condition
The iteration termination condition is generally that a maximum iteration number and a minimum precision are set, when the bee search reaches the maximum iteration number or the increment of the fitness value meets the minimum precision error, the iteration is stopped, and the current position of the optimal honey source is the optimal solution.
3. (step 103) calculating the AVO inversion objective function value at each honey source. Using d to represent the actual observed gather seismic record and m to represent the model parameters, the bayesian theory can be expressed as:
wherein P (m|d) is the posterior probability distribution of the model parameter m; p (m) is the prior probability distribution of the model parameter m; p (d|m) is a likelihood function representing the degree of matching between the model parameter m and the observed data d; p (d) is the edge probability distribution of the observed data d, a normalization factor in the case of known observed data, and can be regarded as a constant. The likelihood function is assumed to obey a gaussian distribution of zero mean, expressed as:
wherein m is the number of incidence angles, n is the number of sampling points, g is a nonlinear positive operator, C n Is a covariance matrix, which can be expressed as C n ＝σ n 2 I，σ n I is the standard deviation of noise, and I is the identity matrix of mn×mn. Assuming that the relative changes in longitudinal wave velocity, transverse wave velocity and density conform to a three-variable cauchy prior distribution with a mean value of zero, the following is true:
wherein phi is i ＝D i T C m -1 D i ；C m A 3 x 3 covariance matrix comprising statistical correlations between three AVO parameters; d (D) i A matrix of 3 x 3 n; m is m c Is a vector containing the relative changes of longitudinal wave speed, transverse wave speed and density, and is expressed as The symbol-represents taking the mean of the upper and lower medium parameters at the current sampling point. Substituting the formulas (3) and (2) into the formula (1) according to a Bayesian formula to obtain posterior probability distribution of model parameters as follows:
in order to maximize the posterior probability P (m|d) of the model parameters, which corresponds to minimizing the negative of its exponential terms, an objective function can be established that is minimized as follows:
to compensate for the low frequency information, model constraint terms are added to the objective function, where the final expression of the objective function is:
wherein ε is 2 Weight coefficient representing model constraint term, m r Representing a low frequency model, W m Representing the weighting matrix. The objective function consists of three items, wherein the first item represents a data error item and mainly controls inversion precision; the second term represents the prior distribution constraint term of inversion parameters, and mainly controls the sparsity and sigma of inversion n 2 The larger the inversion result is, the higher the sparsity is, otherwise, the inversion result has obvious band-limited characteristics; the third term represents a model constraint term, mainly inversion compensation low-frequency information, and can further improve stability and accuracy of inversion.
The objective function value at the corresponding honey source is calculated according to equation (6).
4. And (step 104) evaluating the honey sources according to the fitness values of the honey sources, wherein the honey source position with the minimum objective function value and the larger fitness value is the current optimal honey source position. The artificial bee colony algorithm is a maximum optimization process and is embodied by a fitness function; AVO inversion is a minimum optimization problem, which is embodied by the objective function. The key point of combining AVO inversion with a swarm algorithm is to obtain the relation between the fitness function and the objective function. The invention uses the corresponding relation shown in the following formula (7), wherein fit i Is the fitness value of the ith honey source,i=1, 2,3, SN for the objective function value corresponding to the i-th honey source. And (3) using the formula (7), and calculating according to the objective function to obtain a corresponding fitness value.
The relationship between the fitness function and the objective function is schematically shown in fig. 2, and it can be seen that the fitness function and the objective function obviously have a negative correlation, that is, the fitness value decreases with the increase of the objective function value. In a general inversion problem, the ideal value of the objective function, i.e. the minimum value is 0, and the fitness function takes the maximum value of 1.
5. (step 105) evaluating the objective function value and fitness value at the location of the honey source, and then updating the honey source using a modified swarm algorithm comprising a swarm of particles and a genetic algorithm. If the new honey source position is better than the original honey source position, replacing the original honey source position with the new honey source position, otherwise, retaining the original honey source position. Taking the update of the weight coefficient of the inversion equation cauchy prior constraint term in step 103 as an example, in cauchy distribution, the scale matrix contains the related information among model parameters, so that the stability of inversion can be improved; the cauchy distribution also has long tail characteristics, and can provide sparse constraint for inversion to obtain a high-resolution solution. The parameters to be considered in the Cauchy constraint term are the weight coefficient sigma n 2 Parameter optimization is performed by testing the influence of the weight coefficients on the inversion effect. The inversion results of the cauchy constraint weight coefficients at too small, too large and moderate values are shown in fig. 3, wherein graphs (a), (b) and (c) respectively represent longitudinal wave velocity, transverse wave velocity and density. Solid lines in the figure represent logging results; the dotted line shows the inversion result when the weight coefficient is excessively small, and the inversion errors of the longitudinal wave speed, the transverse wave speed and the density are respectively 4.56%, 9.21% and 1.41%; the stippled line shows the inversion result when the weight coefficient is excessively large, and the inversion errors of the longitudinal wave speed, the transverse wave speed and the density are respectively 4.70%, 9.24% and 1.45%; the short horizontal line represents the inversion result when the weight coefficient is proper in value, and the longitudinal wave speed, the transverse wave speed and the density areThe inversion errors of the degrees were 4.51%, 9.05% and 1.39%, respectively. The comparison between the inversion result synthetic seismic record and the actual seismic record is shown in fig. 4, wherein the graph (a) shows that the weight coefficient is too small in value, the graph (b) shows that the weight coefficient is too large in value, and the graph (c) shows that the weight coefficient is moderate in value. It can be seen that when the weight coefficient is excessively large, the inversion result error is maximum; when the weight coefficient is too small, the inversion result is relatively close to the inversion result when the weight coefficient is moderate, and mainly because the low-frequency model constraint in the objective function can also improve the stability of inversion, the cauchy constraint item weight coefficient is too small and cannot have too great influence on inversion; and when the weight coefficient is moderate, the error of the inversion result is minimum.
6. (step 106) calculating the probability that the observed bees select a honey source in a roulette manner, if the observed bees select the honey source, converting the observed bees into bees, carrying out neighborhood search according to the flow shown in steps 103-105 and updating the honey source according to an improved artificial bee colony algorithm;
7. (step 107) judging whether the neighborhood searching times exceeds the preset limit value (namely the maximum searching times of bees), if so, discarding the current honey source, converting the roles of the collected bees and the observed bees into a scout bee, randomly generating a new honey source according to the step (1), otherwise, returning to the step 103 to continuously update the honey source position;
8. and (108) judging whether an iteration termination condition is reached, namely, the set maximum iteration times are reached or the optimization error is smaller than a limit value, if so, the optimal honey source position at the moment is the optimal solution (final inversion result) obtained by inversion, otherwise, returning to the step 103 to continuously update the honey source position. In the embodiment, the optimization processing of the gather, the establishment of an initial model, the extraction of wavelets corresponding to 6 incidence angles and the use of related parameters in a preferred artificial bee colony algorithm are combined, AVO inversion based on an improved artificial bee colony algorithm is respectively carried out on the well 2D survey line and the optimized angle gather data body in a three-dimensional work area, and the inversion result can be found to be substantially consistent with the logging curve and the horizon information after low-pass filtering through comparison, so that the accuracy of the inversion result is proved. Fig. 5 is a graph comparing AVO inversion results with logging results, and it can be seen that inversion results are consistent with the transformation trend of logging results, correlation coefficients between inverted longitudinal wave velocity curves and logging velocity curves are higher, and average relative errors are smaller. The result shows that the AVO inversion effect based on the improved artificial bee colony algorithm is good and consistent with the logging and geological analysis results.
Claims (6)
1. The pre-stack AVO inversion method based on the swarm optimization algorithm is characterized by comprising the following steps of:
step 1, inputting a three-dimensional pre-stack seismic data volume and a seismic wavelet;
step 2, setting an initial honey source, namely constructing an inversion initial model;
step 3, calculating an objective function value according to the constructed AVO inversion equation based on the Bayesian theory;
step 4, calculating a honey source fitness value according to the relation between the AVO inversion objective function and the bee colony algorithm fitness function;
combining AVO inversion with bee colony algorithm, and using the following corresponding relationship according to the relationship between fitness function and objective function, wherein fit i Is the fitness value of the ith honey source, phi i I=1, 2,3, SN for the objective function value corresponding to the i-th honey source; the corresponding fitness value is calculated for the objective function using the equation,
step 5, applying an improved bee colony algorithm to search the neighborhood to find the optimal solution;
carrying out neighborhood search to update the honey source position according to an improved bee colony algorithm combining a particle swarm and a genetic algorithm, and setting various parameters;
step 6, judging whether the neighborhood searching times exceeds a set limit value, namely, the maximum searching times of bees, if so, giving up the current honey source, converting the roles of collecting bees and observing bees into scout bees, and randomly generating a new honey source; otherwise, returning to the step 3 to continuously update the honey source position;
and 7, judging whether an iteration termination condition is reached, namely, the set maximum iteration times are reached or the optimization error is smaller than a limit value, if so, the optimal honey source position at the moment is the optimal solution obtained by inversion, otherwise, returning to the step 3, and continuing updating the honey source position.
2. The method of inversion of pre-stack AVO based on a swarm optimization algorithm according to claim 1, wherein in step 1, the input pre-stack gather is a standard SEGY format file subjected to optimization processing, and the input seismic wavelet is a theoretical wavelet or an extracted seismic wavelet.
3. The method for inversion of pre-stack AVO based on a swarm optimization algorithm according to claim 2, wherein in step 1, the seismic data collected by industry are shot records, a CMP gather is obtained after processing, the CMP gather is converted into an angle domain common imaging point gather, and the angle gather is extracted by adopting a method based on a ray theory; preprocessing logging data, carrying out time-depth conversion, calibrating the logging data on the seismic data, and further extracting seismic wavelets.
4. The pre-stack AVO inversion method based on the swarm optimization algorithm according to claim 1, wherein in step 2, an initial model of physical parameters is set under the condition of conforming to a certain probability distribution, namely an initial honey source is set, and the number of the honey sources is represented by SN.
5. The method of inversion of pre-stack AVO based on bee colony optimization algorithm according to claim 4, wherein in step 2, algorithm parameters are set, the parameters including bee colony size, maximum number of searches of bees, upper and lower bounds of bee search space, and iteration termination conditions.
6. The method of inversion of pre-stack AVO based on swarm optimization according to claim 1, wherein in step 3, the seismic record d and the model m are collected using the input pre-stack trace, according to the inversion target constructed
Calculating objective function value, wherein each parameter is required to be set, phi (m) is the objective function value at honey source, g (m) is nonlinear positive algorithm, sigma n 2 Is the square of the standard deviation of noise, m c To include the relative change vector of longitudinal and transverse wave speed and density i For the objective function value corresponding to the ith honey source epsilon 2 Weight coefficient for model constraint term, m r Is a low-frequency model, W m Is a weighting matrix; the first term in the formula represents a data error term and controls inversion accuracy; the second term represents the prior distribution constraint term of inversion parameters, controls the sparsity of inversion, sigma n 2 The larger the inversion result is, the higher the sparsity is, otherwise, the inversion result has obvious band-limited characteristics; the third term represents a model constraint term, low-frequency information is compensated for inversion, and stability and accuracy of inversion are further improved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010302615.5A CN113534246B (en) | 2020-04-17 | 2020-04-17 | Pre-stack AVO inversion method based on bee colony optimization algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010302615.5A CN113534246B (en) | 2020-04-17 | 2020-04-17 | Pre-stack AVO inversion method based on bee colony optimization algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113534246A CN113534246A (en) | 2021-10-22 |
| CN113534246B true CN113534246B (en) | 2023-11-24 |
Family
ID=78088498
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010302615.5A Active CN113534246B (en) | 2020-04-17 | 2020-04-17 | Pre-stack AVO inversion method based on bee colony optimization algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113534246B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104570101A (en) * | 2013-10-09 | 2015-04-29 | 中国石油化工股份有限公司 | AVO (amplitude versus offset) three-parameter inversion method based on particle swarm optimization |
| DE102014000234A1 (en) * | 2014-01-09 | 2015-07-09 | Angelo Piasentin | Deterministic and stochastic method for the computation, dissemination and interpretation of petrophysical micro-properties and macro-properties as petrophysical-seismic attributes in the "3D Seismic Volume" for interpretation purposes |
| CN107024717A (en) * | 2017-05-27 | 2017-08-08 | 伍庆华 | A kind of improved adaptive GA-IAGA for earthquake data before superposition parametric inversion |
| CN107193044A (en) * | 2017-05-31 | 2017-09-22 | 河海大学 | A kind of pre-stack seismic Multi-parameters conversion method of hybrid global optimization |
| WO2018107904A1 (en) * | 2016-12-12 | 2018-06-21 | 中国石油大学(华东) | Method for precisely inverting young's modulus and poisson's ratio |
| CN108445537A (en) * | 2018-02-07 | 2018-08-24 | 中国地质大学（武汉） | Earthquake data before superposition AVO elastic parameter inversion methods based on Spark and system |
| CN109143355A (en) * | 2018-08-23 | 2019-01-04 | 电子科技大学 | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM |
| CN110795882A (en) * | 2019-10-31 | 2020-02-14 | 北京化工大学 | Wave field simulation method based on improved quantum particle swarm algorithm |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7072767B2 (en) * | 2003-04-01 | 2006-07-04 | Conocophillips Company | Simultaneous inversion for source wavelet and AVO parameters from prestack seismic data |
-
2020
- 2020-04-17 CN CN202010302615.5A patent/CN113534246B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104570101A (en) * | 2013-10-09 | 2015-04-29 | 中国石油化工股份有限公司 | AVO (amplitude versus offset) three-parameter inversion method based on particle swarm optimization |
| DE102014000234A1 (en) * | 2014-01-09 | 2015-07-09 | Angelo Piasentin | Deterministic and stochastic method for the computation, dissemination and interpretation of petrophysical micro-properties and macro-properties as petrophysical-seismic attributes in the "3D Seismic Volume" for interpretation purposes |
| WO2018107904A1 (en) * | 2016-12-12 | 2018-06-21 | 中国石油大学(华东) | Method for precisely inverting young's modulus and poisson's ratio |
| CN107024717A (en) * | 2017-05-27 | 2017-08-08 | 伍庆华 | A kind of improved adaptive GA-IAGA for earthquake data before superposition parametric inversion |
| CN107193044A (en) * | 2017-05-31 | 2017-09-22 | 河海大学 | A kind of pre-stack seismic Multi-parameters conversion method of hybrid global optimization |
| CN108445537A (en) * | 2018-02-07 | 2018-08-24 | 中国地质大学（武汉） | Earthquake data before superposition AVO elastic parameter inversion methods based on Spark and system |
| CN109143355A (en) * | 2018-08-23 | 2019-01-04 | 电子科技大学 | Semi-supervised global optimization seismic facies quantitative analysis method based on SOM |
| CN110795882A (en) * | 2019-10-31 | 2020-02-14 | 北京化工大学 | Wave field simulation method based on improved quantum particle swarm algorithm |
Non-Patent Citations (6)
| Title |
|---|
| High-resolution three-term AVO inversion by means of a Trivariate Cauchy probability distribution;Wubshet Alemie, Mauricio D. Sacchi;《GEOPHYSICS》;第76卷(第03期);第43-55页 * |
| Prestack AVO inversion using the improved artificial bee colony algorithm based on exact Zoeppritz equations;Hongmei Luo等;《SEG Technical Program Expanded Abstracts 2020》;第3887页 * |
| 基于混合智能优化算法的非线性AVO反演;方中于;王丽萍;杜家元;梁立锋;《石油地球物理勘探》;第52卷(第04期);第797-804页 * |
| 基于精确Zoeppritz方程和改进人工蜂群算法的AVO反演方法研究;罗红梅等;《石油物探》;第61卷(第04期);第659-672页 * |
| 基于蚁群算法的非线性AVO反演;严哲;顾汉明;赵小鹏;《石油地球物理勘探》;第44卷(第06期);第700-702页 * |
| 智能优化算法叠前AVO非线性反演研究;王丽萍;《CNKI博士学位论文全文库》;第1-124页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113534246A (en) | 2021-10-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109709603B (en) | Seismic horizon identification and tracking method and system | |
| Mohamed et al. | Comparison of stochastic sampling algorithms for uncertainty quantification | |
| CN111596978A (en) | Web page display method, module and system for lithofacies classification by artificial intelligence | |
| Jiang et al. | Nonlinear inversion of electrical resistivity imaging using pruning Bayesian neural networks | |
| CN116665067B (en) | Ore finding target area optimization system and method based on graph neural network | |
| Liu et al. | Artificial intelligence method for shear wave travel time prediction considering reservoir geological continuity | |
| CN114723095A (en) | Missing well logging curve prediction method and device | |
| CN110568494B (en) | Pre-stack non-Gaussian AVO inversion method based on generalized extremum distribution | |
| Wu et al. | An improved particle swarm optimization algorithm for AVO elastic parameter inversion problem | |
| Shirazy et al. | K-means clustering and general regression neural network methods for copper mineralization probability in Chahar-Farsakh, Iran | |
| Brown et al. | Machine learning on Crays to optimize petrophysical workflows in oil and gas exploration | |
| CN111894563A (en) | Method and system for determining classification of fractured reservoir in buried hill section | |
| CN113534246B (en) | Pre-stack AVO inversion method based on bee colony optimization algorithm | |
| Colombo et al. | Machine-learning inversion via adaptive learning and statistical sampling: Application to airborne micro-TEM for seismic sand corrections | |
| CN116776949A (en) | Machine learning chemical exploration data processing method and system based on mineral control element restriction | |
| CN115964667A (en) | River-lake lithofacies well logging identification method based on deep learning and resampling | |
| GB2583906A (en) | Method of estimating a mineral content of a geological structure | |
| Malvić et al. | Neural networks in petroleum geology as interpretation tools | |
| Dixit et al. | Genetic-evolutionary adaptive moment estimation-based semisupervised deep sequential convolution network for seismic impedance inversion: Application and uncertainty analysis | |
| CN112749807A (en) | Quantum state chromatography method based on generative model | |
| Wang et al. | Auto recognition of carbonate microfacies based on an improved back propagation neural network | |
| Zheng et al. | Lithofacies logging identification for strongly heterogeneous deep-buried reservoirs based on improved Bayesian inversion: The Lower Jurassic sandstone, Central Junggar Basin, China | |
| Luo et al. | Prestack AVO inversion using the improved artificial bee colony algorithm based on exact Zoeppritz equations | |
| Wu et al. | A spatially constrained divisive hierarchical k-means clustering to capture prior features from migration velocity model to build training model set for deep-learning LSRTM | |
| Xu et al. | Subsurface lithofacies identification with meta learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |