A kind of combination system of selection of seismic properties set
Technical field
The present invention relates to a kind of seismic attributes analysis method in oil seismic exploration, particularly about a kind of combination system of selection of seismic properties set.
Background technology
In the exploration stage of hydrocarbon resources, generally all utilize 3D seismic data describing reservoir information.The full detail that 3D seismic data obtains through direct computing or logical operation or experience inference is called seismic properties.Since the eighties in 20th century, a large amount of successful case research shows: integrated use seismic properties and log data become the most effective layer description method and means.Since the people such as Tanner in 1979 clearly propose three wink seismic properties, for the seismic properties quantity of layer description reached hundreds of more than.But too much seismic properties causes a large amount of spurious correlations on the contrary in interpretation process, and then cause falling through of a large amount of prospect pit.Therefore, select optimum seismic properties, thus improve the precision of reservoir prediction, increase the success ratio of prospect pit, there is significant economic benefit and social benefit.
Seismic properties preferably can be defined as the experience or mathematical method that utilize people, optimize the most responsive to institute's Solve problems (as the oil and gas prediction of subsurface reservoir or the sign etc. of subsurface reservoir parameter), the most effectively, the combination of the seismic properties that the most representative and number is minimum.Existing seismic properties method for optimizing can be divided into two large classes: a class is the artificial selection based on expert's priori, another kind of be based on mathematical algorithm automatically preferably.Wherein, artificial selection seismic properties combination based on expert's priori is that expert is by rule of thumb on the seismic section of concrete study area and the dropping cut slice of 3D seismic data, the seismic properties of paropsia is chosen, determines more effective several seismic properties.But once reservoir situation too complex, seismic properties very many (hundreds of kind), the time cost that expert selects one by one is too high, can only propose several preferably seismic properties simply; Automatically preferably earthquake combinations of attributes based on mathematical algorithm be in conjunction with computing machine mathematical algorithm automatically preferably, mainly solve local or the global optimizing problem of multiple seismic properties, searching process and optimizing result also need to control in conjunction with expertise, analysis and inspection.At present, mathematical algorithm comprises seismic properties dimensionality reduction method for optimizing and seismic properties optimization algorithm.Seismic properties dimensionality reduction method for optimizing does not need the constraint of logging trace, and it mainly comprises principal component analysis (Karhunen-Loeve transform, Karhunen-Loeve transformation), independent component analysis and local linearly embedding method etc.Seismic properties optimization algorithm needs the constraint of logging trace, it mainly comprises intersection analytic approach, contribution amount analytic approach, search procedure, GA-BP (Genetic Algorithm-Back Propagation NeuralNetwroks, Genetic Algorithm-BP Neural Network) method, rough set method, discriminatory analysis, forward modeling model analysis etc.But above-mentioned seismic properties method for optimizing, be all often the application of a kind of single method, multi-solution is strong, stability is inadequate.
Summary of the invention
For the problems referred to above, the object of this invention is to provide a kind of precision that can improve reservoir prediction, increase the combination system of selection of the seismic properties set of the success ratio of prospect pit.
For achieving the above object, the present invention takes following technical scheme: a kind of combination system of selection of seismic properties set, comprise the following steps: 1) gather and shake data two-dimensional or three-dimensional, in the geological data collected, the destination layer position that will study is calibrated by drilling data, well-log information and geologic information, and window scope when determining the seismic reflection of destination layer position; 2) from the geological data within the scope of window during the seismic reflection of destination layer position, some seismic properties are extracted; 3) each seismic properties extracted is carried out histogram normalization respectively, the numerical value that each seismic properties is comprised all is distributed in [0,1] interval, and the seismic properties after normalization is carried out pre-service, make the data Gaussian distributed comprised in each seismic properties; 4) according to the redundancy feature between seismic properties, by step 3) in all seismic properties of obtaining carry out de-redundancy process, obtain nonredundancy seismic properties; 5) by step 4) in the nonredundancy seismic properties that obtains adopt eigenwert transform method to carry out independence analysis, obtain some separate seismic properties, i.e. step 4) in the principal component of nonredundancy seismic properties; 6) by step 5) in the principal component of seismic properties that obtains adopt sparse transformation method to carry out sparse distribution analysis, obtain the seismic properties combination meeting sparse distribution, i.e. optimum earthquake community set.
Described step 3) middle pre-service employing outlier cutting method.
Described step 4) adopt the compatible decision table of rough set method when carrying out redundancy process.
Described step 5) in eigenwert mapping algorithm adopt Karhunen-Loeve transformation method.
Described step 6) in sparse transformation method adopt anatomic element analytical approach.
The present invention is owing to taking above technical scheme, and it has the following advantages: 1, the present invention is due to the redundancy by removing in the many community sets of earthquake between each attribute, can improve the stability of seismic multi-attribute pattern-recognition.2, the present invention is due to the independence by keeping in the many community sets of earthquake between each attribute, can reduce the multi-solution of seismic multi-attribute pattern-recognition to greatest extent.3, the present invention is due to openness by what keep in earthquake many community sets between each attribute, can improve counting yield and the convergence capabilities of seismic multi-attribute pattern-recognition further.4, the present invention is owing to performing removal redundancy according to selected seismic properties according to certain sequential combination, independent sum is kept to give prominence to sparse single seismic properties method for optimizing, to obtain accurate and stable preferred result, greatly reduce preferred cost of labor, for follow-up reservoir prediction provides high-quality input attributes set, by the identification to optimum earthquake community set, complete the quantitative forecast of subsurface reservoir parameter, thus improve the precision of reservoir prediction, increase the success ratio of prospect pit, there is significant economic benefit and social benefit.Therefore the present invention can be widely used in oil exploration and exploitation field.
Accompanying drawing explanation
Fig. 1 is method flow schematic diagram of the present invention.
Embodiment
As shown in Figure 1, the combination system of selection of seismic properties set of the present invention, be based on some seismic properties between redundancy, independence and openness feature, removal redundancy is performed according to certain sequential combination according to selected seismic properties, keep independent sum to give prominence to sparse single seismic properties method for optimizing, it comprises the following steps:
1) collection shakes data two-dimensional or three-dimensional, and in the geological data collected, the destination layer position (data namely in geological data within the scope of certain hour that will study are calibrated by existing drilling data, well-log information and geologic information, data also referred to as within the scope of window when one), and window scope when determining the seismic reflection of destination layer position;
2) from the geological data within the scope of window during the seismic reflection of destination layer position, extract some seismic properties, seismic properties comprises amplitude generic attribute, instantaneous generic attribute, frequency generic attribute, sequence generic attribute and non-linear generic attribute etc.;
3) each seismic properties extracted is carried out histogram normalization respectively, the data that each seismic properties is comprised all are distributed in [0,1] interval, and adopt existing (excessive or too small) outlier cutting method to carry out pre-service the seismic properties after normalization, make the data Gaussian distributed comprised in each seismic properties;
4) according to the redundancy feature between seismic properties, by step 3) in all seismic properties of obtaining adopt rough set method to carry out de-redundancy process, detailed process is: utilize the compatible decision table based on rough set method to carry out de-redundancy process to all seismic properties, obtain compatible seismic properties, i.e. nonredundancy seismic properties;
5) by step 4) in the nonredundancy seismic properties that obtains adopt eigenwert transform method to carry out independence analysis, obtain some separate seismic properties, i.e. step 4) in the principal component of nonredundancy seismic properties, wherein, eigenwert mapping algorithm can adopt Karhunen-Loeve transformation method;
6) by step 5) in the principal component of seismic properties that obtains adopt sparse transformation method to carry out sparse distribution analysis, obtain the seismic properties combination meeting sparse distribution, i.e. optimum earthquake community set; Wherein, sparse transformation method can adopt anatomic element analytical approach;
7) by step 6) the optimum earthquake community set that obtains for next step reservoir parameter pattern-recognition, complete the quantitative forecast of subsurface reservoir parameter, improve the precision of reservoir prediction, increase the success ratio of prospect pit.
Be described in detail below in conjunction with the combination system of selection of specific embodiment to seismic properties set of the present invention.
1) geological data when choosing the seismic reflection of the top bottom boundary of destination layer position within the scope of window;
2) 60 interlayer seismic properties are extracted and from the geological data chosen;
3) one by one histogram normalization is carried out to selected 60 interlayer seismic properties, the data that each seismic properties is comprised all are distributed in [0,1] interval, adopt existing (excessive or too small) outlier cutting method to carry out pre-service to the seismic properties after normalization, make each interlayer seismic properties Gaussian distributed;
4) according to the redundancy of seismic properties, select the compatible decision table of rough set method to remove the redundant attributes that in 60 seismic properties, correlativity is strong, obtain 18 nonredundant seismic properties;
5) by step 4) in 18 seismic properties obtaining adopt Karhunen-Loeve transformation methods to carry out independence analysis, obtain the principal component of 11 seismic properties;
6) by step 5) in the principal component of 11 seismic properties adopt anatomic element analysis structure dictionary to obtain 6 form seismic properties, 6 form seismic properties are now the optimal set of seismic properties;
7) being used for next step reservoir parameter pattern-recognition by finally obtaining 6 attributes, completing the quantitative forecast of subsurface reservoir parameter.
The various embodiments described above are only for illustration of the present invention, and wherein each step etc. all can change to some extent, and every equivalents of carrying out on the basis of technical solution of the present invention and improvement, all should not get rid of outside protection scope of the present invention.