Sweet spot identification method and device, equipment and medium

Abstract

The embodiment of the invention discloses a dessert identification method and device, equipment and a medium, and relates to the technical field of geophysical exploration signal processing. The method comprises the following steps: performing pre-stack forward modeling on a plurality of pre-established dessert mechanism models to obtain simulated seismic data; the dessert mechanism model is established according to a dessert type of a target area, rock physical parameters and a contact relationship between rocks; carrying out attribute extraction on the simulated seismic data to obtain initial attributes, and determining sensitive attributes for model training from the initial attributes; determining each sensitive attribute value as a training sample, and training a fuzzy self-organizing neural network for sweet spot recognition through the training sample to obtain a trained sweet spot recognition model; the trained dessert identification model is used for identifying the dessert type of the target area. According to the technical scheme, sweet spot type identification is carried out through the sensitive attributes, sweet spot distribution can be accurately identified, and technical support is provided for exploration and development of oil and gas.

Description

Technical Field

[0001] The present invention relates to the technical field of geophysical exploration signal processing, and in particular to a dessert recognition method, device, equipment and medium. Background Art

[0002] The sweet spot refers to the high-porosity and high-quality reservoir section in the formation. Accurate identification of the sweet spot is a very important part of oil and gas development.

[0003] When identifying sweet spots, traditional geophysical attribute analysis and single parameter threshold methods are difficult to accurately identify the distribution of sweet spots, thus affecting oil and gas development. Summary of the Invention

[0004] The present invention provides a sweet spot identification method, device, equipment and medium, which can accurately identify the distribution of sweet spots and provide technical support for oil and gas exploration and development.

[0005] According to one aspect of the present invention, a method for identifying a dessert is provided, the method comprising:

[0006] Perform pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism models are established based on the sweet spot type, rock physical parameters, and contact relationships between rocks in the target area;

[0007] Extracting attributes from the simulated seismic data to obtain initial attributes, and determining sensitive attributes for model training from the initial attributes;

[0008] Each sensitive attribute value is determined as a training sample, and a fuzzy self-organizing neural network for dessert recognition is trained using the training sample to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the dessert type in the target area.

[0009] According to another aspect of the present invention, a dessert identification device is provided, comprising:

[0010] The pre-stack forward modeling module is used to perform pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism model is established based on the sweet spot type, rock physical parameters and contact relationship between rocks in the target area;

[0011] A sensitive attribute determination module is used to extract attributes from the simulated seismic data to obtain initial attributes, and to determine sensitive attributes for model training from the initial attributes;

[0012] The neural network training module is used to determine each sensitive attribute value as a training sample, and train the fuzzy self-organizing neural network for dessert recognition through the training sample to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the dessert type in the target area.

[0013] According to another aspect of the present invention, an electronic device is provided, comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the dessert identification method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the dessert identification method according to any embodiment of the present invention when executed.

[0018] The technical solution of the embodiment of the present application includes: performing pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism model is established based on the sweet spot type, rock physical parameters, and contact relationships between rocks in the target area; performing attribute extraction on the simulated seismic data to obtain initial attributes, and determining sensitive attributes for model training from the initial attributes; determining each sensitive attribute value as a training sample, and using the training samples to train a fuzzy self-organizing neural network for sweet spot identification to obtain a trained sweet spot identification model; the trained sweet spot identification model is used to identify the sweet spot type in the target area. This technical solution uses sensitive attributes to identify the sweet spot type, can accurately identify the distribution of sweet spots, and provides technical support for oil and gas exploration and development.

[0019] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flowchart of a dessert identification method provided according to Example 1 of the present application;

[0022] Figure 2 is a flowchart of a dessert identification method provided according to Example 2 of the present application;

[0023] Figure 3 This is a schematic diagram of a dessert mechanism model provided according to Example 2 of the present application;

[0024] Figure 4 This is a schematic diagram of a change trend of an initial attribute provided according to Example 2 of the present application;

[0025] Figure 5 A two-dimensional attribute graph for verifying the correlation between actual well logging data and mechanism model attributes is provided according to the second embodiment of the present application;

[0026] Figure 6 A fuzzy self-organizing neural network topology diagram is provided according to the second embodiment of the present application;

[0027] Figure 7 is a schematic structural diagram of a dessert identification device provided according to the third embodiment of the present application;

[0028] Figure 8 It is a structural diagram of an electronic device for implementing a dessert identification method according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", "target", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Example 1

[0032] Figure 1 A flowchart of a dessert recognition method is provided for the first embodiment of the present application. The embodiment of the present application is applicable to the case of identifying desserts in a target area. The method can be performed by a dessert recognition device. The dessert recognition device can be implemented in the form of hardware and / or software. The dessert recognition device can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0033] S110, performing pre-stack forward modeling on a plurality of pre-established sweet spot mechanism models to obtain simulated seismic data.

[0034] Sweet spots are areas with good reservoir properties, high oil and gas content, and high production value. This application embodiment can identify sweet spots within a target area and determine their spatial distribution. This target area can be a typical tight sandstone gas reservoir characterized by low to ultra-low permeability, complex pore structure, and strong heterogeneity. The distribution of sweet spots in such reservoirs is controlled by the coupling of sedimentary microfacies, diagenetic evolution, and structural fractures. Their identification and prediction are key technical bottlenecks for efficient development.

[0035] The sweet spot mechanism model is established based on the sweet spot type, rock physical parameters and contact relationship between rocks in the target area; for example, the sweet spot type, typical rock physical parameters and contact relationship of mudstone are obtained by analyzing logging data and seismic data, and the sweet spot mechanism model is established based on this.

[0036] Specifically, pre-stack forward modeling is performed on a variety of pre-established sweet spot mechanism models. During the pre-stack forward modeling, the Aki-Richards equation (a key approximate formula used in the field of seismic exploration to describe the reflection and transmission behavior of plane elastic waves at the interface of the formation) is used to obtain simulated seismic data.

[0037] S120 , extracting attributes from the simulated seismic data to obtain initial attributes, and determining sensitive attributes for model training from the initial attributes.

[0038] Specifically, attributes are extracted from the simulated seismic data to obtain initial attributes, which include but are not limited to: intercept attribute P, gradient attribute G, G*sin(P), P*sin(G), polarization angle difference, polarization coefficient square, polarization intensity, Polarization Product, Product(P*G) (product of intercept and gradient), Poisson's ratio, shear wave reflection coefficient and Weighted Polarization Product, etc. After obtaining the initial attributes, it is necessary to determine the sensitive attributes used for model training. Sensitive attributes are attributes associated with sweet spots.

[0039] Furthermore, determining the sensitive attribute in the initial attributes includes: determining a change trend of the initial attribute; if the change trend is associated with a thickness change trend of the sweet spot, or is associated with a change trend of the attribute in the actual formation, then determining the initial attribute as a sensitive attribute.

[0040] Exemplarily, the sensitive attributes include but are not limited to: P*G, G*sin(P), polarization intensity, P*sin(22°)+G*cos(22°); wherein P is the intercept and G is the gradient.

[0041] S130 , determining each sensitive attribute value as a training sample, and training a fuzzy self-organizing neural network for dessert recognition using the training samples to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the type of dessert in the target area.

[0042] Specifically, after determining the sensitive attribute, on the one hand, the specific sensitive attribute value of the sensitive attribute can be determined from the sweet spot mechanism model, and on the other hand, the sensitive attribute value can be determined based on the logging data. Then, the sensitive attribute value is used as a training sample to train the sweet spot recognition model to obtain the trained sweet spot recognition model; the sweet spot type in the target area can be identified through the trained sweet spot recognition model.

[0043] The technical solution of the embodiment of the present application includes: performing pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism model is established based on the sweet spot type, rock physical parameters, and contact relationships between rocks in the target area; performing attribute extraction on the simulated seismic data to obtain initial attributes, and determining sensitive attributes for model training from the initial attributes; determining each sensitive attribute value as a training sample, and using the training samples to train a fuzzy self-organizing neural network for sweet spot identification to obtain a trained sweet spot identification model; the trained sweet spot identification model is used to identify the sweet spot type in the target area. This technical solution uses sensitive attributes to identify the sweet spot type, can accurately identify the distribution of sweet spots, and provides technical support for oil and gas exploration and development.

[0044] Example 2

[0045] Figure 2 This is a flow chart of a dessert identification method provided in Example 2 of the present application. This embodiment of the present application is optimized based on the above embodiment.

[0046] like Figure 2 As shown, the method of the embodiment of the present application specifically includes the following steps:

[0047] S210, performing pre-stack forward modeling on a plurality of pre-established sweet spot mechanism models to obtain simulated seismic data.

[0048] Wherein, the sweet spot mechanism model is established based on the sweet spot type, rock physical parameters and contact relationship between rocks in the target area. Exemplarily, the logging data and seismic data are analyzed to obtain a model based on the sweet spot type, typical rock physical parameters of mudstone and contact relationship. In an embodiment of the present application, optionally, the multiple sweet spot mechanism models include at least two of the following: a single-layer sweet spot thickness change model, a single-layer sweet spot position change model, a double-layer sweet spot double-layer position change model, a multi-layer sweet spot position change model, a sweet spot type change model and a sweet spot cumulative thickness constant layer number change model. This scheme is set up in this way so that the sweet spot mechanism model can reflect the distribution of sweet spots as much as possible.

[0049] Figure 3 This is a schematic diagram of the dessert mechanism model, where orange represents type I dessert, yellow represents type II_1 dessert, light pink represents type II_2 dessert, light gray represents type III dessert, and dark gray represents interlayer mud. There are a total of 6 mechanism models for comparison:

[0050] ① Single-layer dessert thickness variation model: With a Class III dessert with a thickness of 6m as the background, Class I, Class II_1, Class II_2 desserts or interlayer mud are placed at the center of the background with thicknesses of 1m, 2m, and 3m, respectively; with a Class III dessert with a thickness of 12m as the background, Class I, Class II_1, Class II_2 desserts or interlayer mud are placed at the center of the background with thicknesses of 2m, 3m, and 6m, respectively.

[0051] ② Single-layer dessert position change model: With the Class III dessert of the same thickness as the background, the Class I, Class II_1, Class II_2 desserts and interlayer mud of the same thickness sink in sequence at the same intervals.

[0052] ③ Double-layer dessert double-layer position change model: The first group of models uses the same thickness of Class III dessert as the background, and divides the 3m cumulative thickness of Class II_1 dessert into two 1.5m thicknesses. In the first three figures, the double-layer desserts are arranged at the same interval and sink in the background position in sequence. In the last figure, the interval between the double-layer desserts is changed; the second group of models uses the same thickness of Class III dessert as the background, and divides the 3m cumulative thickness of Class II_1 dessert into 1m and 2m thicknesses. In the first two figures and the last two figures, the relative positions of the 1m thick dessert and the 2m thick dessert are changed at different intervals.

[0053] ④ Multi-layer dessert position change model: The first group of models uses Class III desserts of the same thickness as the background, Class II_1 desserts of the same cumulative thickness, divided into three layers of 1m thickness, and centered overall, and the interval thickness of Class II_1 desserts is changed; the second group of models uses Class III desserts of the same thickness as the background, Class II_1 desserts of the same cumulative thickness, divided into two layers of 1m and one layer of 2m thickness, and centered overall, and Class I desserts of 2m thickness are placed in high, middle, and low positions; the third group of models replaces Class II_1 desserts in the second group of models with Class II_2 desserts; the fourth group of models replaces Class II_1 desserts in the first group of models with Class II_2 desserts, and Figure 3 Class II_2 desserts have two layers of 1m thick interlayer mud placed at moderate intervals.

[0054] ⑤ Dessert type variation model: The first group of models uses a Class III dessert of the same thickness as the background, and places a 3m thick layer of interlayer mud, Class II_2, Class II_1, and Class I desserts at the same position in the middle of the background; the second group of models uses a Class III dessert of the same thickness as the background, and places two 1.5m thick desserts at the same position in the middle of the background. Figure 1 It is a two-layer II_2 dessert. Figure 2 It is a two-layer II_1 dessert. Figure 3 and Figure 4 Add the background of Class II_1 dessert on the basis of Class III dessert, Figure 3 The upper layer is the II_2 dessert, and the lower layer is the I dessert. Figure 4The third group of models uses a Class III dessert of the same thickness as the background, and places a 1m thick layer and a 2m thick layer of dessert from top to bottom at the same position in the middle of the background. Figure 1 For desserts of type II_2, Figure 2 It is a dessert of type II_1. Figure 3 and Figure 4 Add the background of Class II_1 dessert on the basis of Class III dessert, Figure 3 The upper layer is a 1m thick Class II_2 dessert, and the lower layer is a 2m thick Class I dessert. Figure 4 It is a Class I dessert.

[0055] ⑥ Dessert cumulative thickness unchanged layer number variation model: The first group of models uses a Class III dessert of the same thickness as the background, and places a Class II_1 dessert with a cumulative thickness of 3m in the middle of the background. Figure 1 It is a single-layer 3m thick II_1 dessert. Figure 2 It is a double-layer 1.5m thick II_1 dessert. Figure 3 The second group of models replaces the Class II_1 dessert in the first group of models with the Class II_2 dessert. The third group of models uses the Class III dessert of the same thickness as the background, and places a 4m thick interlayer mud in the middle of the background, and a 1m thick interlayer mud in the same position below the background. Figure 1 Place a 3m thick interlayer mud layer above the background. Figure 2 In the background above there is a 2m thick layer of mud and a 1m thick layer of mud between the layers. Figure 2 Three layers of 1m thick interlayer mud are placed above in the background.

[0056] For example, after establishing the sweet spot mechanism model, the reflection coefficient sequence and wavelet corresponding to each sweet spot mechanism model are determined based on the sweet spot mechanism model and logging parameters, and convolution operation is performed on them to generate a pre-stack synthetic seismic trace set to obtain simulated seismic data; this technical solution can use the Aki-Richards equation for pre-stack forward modeling when performing pre-stack forward modeling.

[0057] S220: Extract attributes from the simulated seismic data to obtain initial attributes.

[0058] For example, after the simulated seismic data is determined, the attributes of the simulated seismic data are extracted, specifically: the intercept attribute P, gradient attribute G, G*sin(P), P*sin(G), polarization angle difference, polarization coefficient square, polarization intensity, Polarization Product (polarization product), Product (P*G) (product of intercept and gradient), Poisson's ratio, shear wave reflection coefficient and Weighted Polarization Product (weighted polarization product) and other twelve attributes in the pre-stack AVO attributes are calculated to obtain the initial attributes.

[0059] S230, obtaining the change trend of each initial attribute, determining the initial attribute with a monotonically increasing or monotonically decreasing change trend as a sensitive attribute; and / or determining the first change trend of the same initial attribute on the well and the second change trend on the sweet spot mechanism model respectively; if the first change trend and the second change trend meet the correlation condition, then the initial attribute is determined as a sensitive attribute.

[0060] Specifically, after determining the initial attribute, the specific values ​​of the initial attribute can be arranged based on the change of a certain parameter in the dessert mechanism model, or a two-dimensional chart of the initial attribute and the parameter can be established to determine the change trend of the initial attribute. For example, the initial attribute values ​​can be arranged from thin to thick based on the thickness of the dessert. If the initial attribute value is monotonically increasing or monotonically decreasing, it is determined that the initial attribute is associated with the thickness of the dessert, and the initial attribute can be a sensitive attribute. It should be noted that the initial attribute value in the embodiment of the present application is the specific value of the initial attribute. For example, if the initial attribute is an intercept, the specific value of the intercept is the initial attribute value.

[0061] For example, the AVO properties of the six mechanism models are compared and analyzed to determine the changing trends of the twelve initial properties, and the initial property values ​​are displayed in a chart. Figure 4 This is a diagram showing the changing trend of an initial attribute. Blue represents the first dessert, orange represents the second dessert, and yellow represents the third dessert or interlayer mud. When the changing trend in the chart shows a monotonically increasing or decreasing trend, the attribute is identified as a sensitive attribute; when the changing trend in the chart shows a nonlinear change, the attribute is identified as a non-sensitive attribute.

[0062] When determining sensitive attributes in the initial attributes, it is also possible to determine whether the initial attribute is a sensitive attribute based on the correlation between the initial attribute value obtained from the actual well logging data and the initial attribute value obtained from the sweet spot mechanism model. The correlation is determined by comparing the change trend of the attribute value on the well with the change trend of the sensitive attribute selected from the mechanism model. Figure 5To verify the correlation between actual logging data and the attributes of the mechanism model, a two-dimensional attribute diagram is shown. The top is a two-dimensional display of the AVO attribute, with blue to red representing a change from weak to strong amplitude. The middle shows the change trend of the AVO attribute at the actual well point, and the bottom shows the change trend of the AVO attribute in the mechanism model. When the attribute change trends of the two are consistent, the attribute is considered a sensitive attribute. Through correlation confirmation, four AVO concentrated attributes (P*G, G*sin(P), polarization intensity, and P*sin(22°)+G*cos(22°)) are identified as sensitive attributes. P*G represents the product attribute of intercept and gradient, G*sin(P) represents the gradient-intercept sine product attribute, polarization intensity represents the change intensity of the seismic wave polarization direction, and P*sin(22°)+G*cos(22°) represents the intercept-gradient angle combination attribute.

[0063] This solution is configured such that attributes associated with dessert types or actual distributions can be determined as sensitive attributes in the initial attributes, and then a dessert recognition model trained based on the sensitive attribute values ​​can more accurately recognize desserts.

[0064] S240 , determining each sensitive attribute value as a training sample, and training a fuzzy self-organizing neural network for dessert recognition using the training samples to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the type of dessert in the target area.

[0065] In an embodiment of the present application, optionally, each sensitive attribute value is determined as a training sample, and after the fuzzy self-organizing neural network for dessert recognition is trained with the training sample, the method further includes: if the fuzzy self-organizing neural network for dessert recognition does not meet the loss function convergence condition during the iteration process, then executing the step of determining the sensitive attributes for model training from the initial attributes, and training the fuzzy self-organizing neural network for dessert recognition again based on the new sensitive attribute values.

[0066] Specifically, if the fuzzy self-organizing neural network for dessert recognition fails to meet the loss function convergence criteria during the iteration process, meaning that the fuzzy self-organizing neural network cannot accurately classify desserts, this may indicate a problem with the selection of sensitive attributes. In this case, the sensitive attributes can be re-determined. Specifically, the step of determining the sensitive attributes for model training from the initial attributes is repeated, and the fuzzy self-organizing neural network for dessert recognition is retrained based on the new sensitive attribute values.

[0067] For example, the selected sensitive seismic attributes are fed into a fuzzy self-organizing neural network system. If the system is unable to clearly distinguish the sweet spot types and their distribution after this process, it indicates that the sensitive attributes are not ideal and require further optimization. Conversely, if they can successfully distinguish, it indicates that the sensitive attributes are sufficiently accurate to accurately distinguish the sweet spot types and their distribution, and can then be applied to actual data analysis.

[0068] For example, after the sensitive attributes are determined, the sensitive attribute values ​​are preprocessed, specifically by cleaning the data and processing outliers to ensure the accuracy and reliability of the analysis results, and normalizing the sensitive attribute values. Specifically, the following formula can be used for normalization:

[0069]

[0070] Among them, A represents the sensitive attribute value after normalization; A f Represents the sensitive attribute value before preprocessing; A max Indicates the maximum value of the sensitive attribute before preprocessing; A min Represents the minimum value of the sensitive attribute before preprocessing. This scheme is set up in this way to convert sensitive attribute values ​​of originally varying magnitudes into a comparable numerical range, thereby laying a solid foundation for subsequent key steps such as data analysis, pattern recognition, and reservoir characteristic prediction.

[0071] In an embodiment of the present application, optionally, a fuzzy self-organizing neural network for dessert recognition is trained using the training sample to obtain a trained dessert recognition model, including: initializing weight values ​​of the fuzzy self-organizing neural network for dessert recognition; and determining the degree of subordination between the training sample and the cluster center based on the following formula:

[0072]

[0073] Among them, x ij is the training sample, i = 1, 2, ..., N, i represents the i-th sensitive attribute value in the training sample, j = 1, 2, ..., M, j represents the number of sample data points in each earthquake attribute (the earthquake attribute is the sensitive attribute, that is, j can be the number of sensitive attribute values ​​of each sensitive attribute), K represents the number of cluster centers, σ represents the width of the Gaussian window, t represents the number of iterations, ω ik (t) is the weight;

[0074] The strength of the membership relationship between the training sample and the neurons in the membership layer is determined according to the following formula:

[0075]

[0076] The weights are updated based on the strength of the membership relationship between the training samples and the neurons in the membership layer:

[0077]

[0078] Among them, α represents the learning factor index;

[0079] If the updated weights and the weights before the update meet the preset conditions, it is determined that the dessert recognition model training is completed.

[0080] It should be noted that when initializing the neural network, a random number is used to determine the initial weight ω of the network cluster center. ik (t). Where 0<ω ik (t)<1, i=1, 2, ..., N, i represents the i-th sensitive attribute value in the training sample, k=1, 2, ..., K represents the number of cluster centers, and t=0 is set as the starting point of the entire initialization process.

[0081] In an embodiment of the present application, optionally, if the updated weights and the weights before the update meet preset conditions, it is determined that the dessert recognition model training is completed, including: calculating the difference between the updated weights and the weights before the update; if the absolute value of the difference is less than a preset threshold, it is determined that the dessert recognition model training is completed.

[0082] For example, to determine whether the network has reached a stable state, that is, to determine whether the sweet spot recognition model training is complete: the amplitude of the neuron weight adjustment is monitored. If the weight update amplitude is reduced to below a certain threshold, the network can be considered to have reached a stable state. Otherwise, if this condition is not met, it is necessary to continue iteration or re-determine the sensitive attributes after a certain number of iterations. The conditions for determining network stability are:

[0083] |ω ik (t+1)-ω ik (t)|<ε

[0084] Where ε is the preset threshold.

[0085] Specifically, when the network reaches a stable state, the clustering results are output, namely:

[0086]

[0087] Where yk represents the number of clustering results. The clustering results of the fuzzy self-organizing neural network can generally reflect the distribution range and thickness of sand bodies, and can also distinguish sweet spots by the number of clusters. The advantage of fuzzy self-organizing neural network clustering is that it does not require sample labels and performs cluster analysis based entirely on the characteristics of the data itself, which can reflect the planar variation trend of the seismic attributes of the horizon.

[0088] For example, Figure 6 It is a fuzzy self-organizing neural network topology graph.

[0089] In an embodiment of the present application, optionally, the method further includes: inputting the actual value of the sensitive attribute of each position in the target area into the trained dessert recognition model to obtain the dessert type of each position; and determining the dessert distribution information of the target area according to the dessert type of each position.

[0090] For example, during actual seismic data processing, sensitive attribute values ​​are extracted from the seismic data based on an appropriate time window. These extracted sensitive attribute values ​​are then subjected to necessary preprocessing steps to ensure data quality. After preprocessing, these data are input into a fuzzy self-organizing neural network to obtain cluster analysis results. Combined with auxiliary information from well logging data, each cluster category is analyzed in depth to interpret its potential sweet spot type. This allows the sweet spot type corresponding to the sensitive attribute value input into the model to be determined. This foundation allows the distribution and thickness characteristics of the sweet spots within the target area to be inferred.

[0091] Example 3

[0092] Figure 7 This is a schematic diagram of the structure of a dessert recognition device provided in Example 3 of this application. The device can execute the dessert recognition method provided in any embodiment of the present invention and has the corresponding functional modules and beneficial effects of the execution method. Figure 7 As shown, the device includes:

[0093] The pre-stack forward modeling module 310 is used to perform pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism models are established based on the sweet spot type, rock physical parameters, and contact relationships between rocks in the target area;

[0094] A sensitive attribute determination module 320 is configured to extract attributes from the simulated seismic data to obtain initial attributes, and determine sensitive attributes for model training from the initial attributes;

[0095] The neural network training module 330 is used to determine each sensitive attribute value as a training sample, and use the training sample to train the fuzzy self-organizing neural network for dessert recognition to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the dessert type in the target area.

[0096] The technical solution of the embodiment of the present application includes: a pre-stack forward modeling module 310, which is used to perform pre-stack forward modeling on a plurality of pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism model is established based on the sweet spot type, rock physical parameters, and contact relationship between rocks in the target area; a sensitive attribute determination module 320, which is used to extract attributes from the simulated seismic data to obtain initial attributes, and determine sensitive attributes for model training from the initial attributes; a neural network training module 330, which is used to determine each sensitive attribute value as a training sample, and train a fuzzy self-organizing neural network for sweet spot identification through the training sample to obtain a trained sweet spot identification model; the trained sweet spot identification model is used to identify the sweet spot type in the target area. This technical solution uses sensitive attributes to identify the sweet spot type, can accurately identify the distribution of sweet spots, and provides technical support for oil and gas exploration and development.

[0097] In the embodiment of the present application, optionally, the sensitive attribute determination module 320 is specifically configured to:

[0098] Obtaining the change trend of each initial attribute, and determining the initial attribute with a monotonically increasing or monotonically decreasing change trend as a sensitive attribute; and / or

[0099] Determine the first change trend of the same initial attribute on the well and the second change trend on the sweet spot mechanism model;

[0100] If the first change trend and the second change trend meet the correlation condition, the initial attribute is determined as a sensitive attribute.

[0101] In the embodiment of the present application, optionally, the device further includes:

[0102] The sensitive attribute re-determination module is used to execute the step of determining the sensitive attributes for model training from the initial attributes if the fuzzy self-organizing neural network for dessert recognition does not meet the loss function convergence condition during the iteration process, and re-train the fuzzy self-organizing neural network for dessert recognition based on the new sensitive attribute values.

[0103] In the embodiment of the present application, optionally, the multiple dessert mechanism models include at least two of the following:

[0104] Single-layer dessert thickness change model, single-layer dessert position change model, double-layer dessert double-layer position change model, multi-layer dessert position change model, dessert type change model and dessert cumulative thickness constant layer number change model.

[0105] In the embodiment of the present application, optionally, the neural network training module 330 includes:

[0106] A weight initialization unit, used to initialize the weight values ​​of the fuzzy self-organizing neural network for dessert recognition;

[0107] The membership degree determination unit is used to determine the membership degree of the training sample and the cluster center based on the following formula:

[0108]

[0109] Where xij is the training sample, i = 1, 2, ..., N, i represents the i-th sensitive attribute value in the training sample, j = 1, 2, ..., M, j represents the number of sample data points in each earthquake attribute, K represents the number of cluster centers, σ represents the width of the Gaussian window, t represents the number of iterations, ω ik (t) is the weight;

[0110] The membership relationship strength determination unit is used to determine the membership relationship strength between the training sample and the membership layer neurons according to the following formula:

[0111]

[0112] The weight update unit is used to update the weights based on the strength of the membership relationship between the training samples and the neurons in the membership layer:

[0113]

[0114] Among them, α represents the learning factor index;

[0115] The training completion determination unit is used to determine that the dessert recognition model training is completed if the updated weights and the weights before the update meet a preset condition.

[0116] In the embodiment of the present application, optionally, the training completion determination unit is specifically configured to:

[0117] Calculate the difference between the updated weight and the weight before the update;

[0118] If the absolute value of the difference is less than the preset threshold, it is determined that the dessert recognition model training is completed.

[0119] In the embodiment of the present application, the apparatus may further include: a dessert type determination module configured to input the actual value of the sensitive attribute of each position in the target area into the trained dessert recognition model to obtain the dessert type of each position;

[0120] The distribution information determination module is used to determine the dessert distribution information of the target area according to the dessert type at each location.

[0121] A dessert recognition device provided in an embodiment of the present application can execute a dessert recognition method provided in any embodiment of the present invention, and has corresponding functional modules and beneficial effects of the execution method.

[0122] Example 4

[0123] Figure 8 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0124] like Figure 8 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0125] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0126] The processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any other suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the dessert recognition method.

[0127] In some embodiments, the dessert recognition method can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the dessert recognition method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the dessert recognition method in any other suitable manner (e.g., via firmware).

[0128] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0129] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0130] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0131] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0132] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0133] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0134] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0135] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for identifying desserts, characterized in that: include: Perform pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism models are established based on the sweet spot type, rock physical parameters, and contact relationships between rocks in the target area; Extracting attributes from the simulated seismic data to obtain initial attributes, and determining sensitive attributes for model training from the initial attributes; Each sensitive attribute value is determined as a training sample, and a fuzzy self-organizing neural network for dessert recognition is trained using the training sample to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the dessert type in the target area.

2. The method according to claim 1, characterized in that Determine sensitive attributes for model training from the initial attributes, including: Obtaining the change trend of each initial attribute, and determining the initial attribute with a monotonically increasing or monotonically decreasing change trend as a sensitive attribute; and / or Determine the first change trend of the same initial attribute on the well and the second change trend on the sweet spot mechanism model; If the first change trend and the second change trend meet the correlation condition, the initial attribute is determined as a sensitive attribute.

3. The method according to claim 1, characterized in that After determining each sensitive attribute value as a training sample and training a fuzzy self-organizing neural network for dessert recognition using the training sample, the method further includes: If the fuzzy self-organizing neural network for dessert recognition does not meet the loss function convergence condition during the iteration process, the step of determining the sensitive attributes for model training from the initial attributes is performed, and the fuzzy self-organizing neural network for dessert recognition is trained again based on the new sensitive attribute values.

4. The method according to claim 1, wherein The multiple sweet spot mechanism models include at least two of the following: Single-layer dessert thickness change model, single-layer dessert position change model, double-layer dessert double-layer position change model, multi-layer dessert position change model, dessert type change model and dessert cumulative thickness constant layer number change model.

5. The method according to claim 1, wherein The fuzzy self-organizing neural network for dessert recognition is trained using the training samples to obtain a trained dessert recognition model, including: Initialize the weight values ​​of the fuzzy self-organizing neural network for dessert recognition; The degree of subordination of the training sample to the cluster center is determined based on the following formula: Among them, x ij is the training sample, i = 1, 2, ..., N, i represents the i-th sensitive attribute value in the training sample, j = 1, 2, ..., M, j represents the number of sample data points in each earthquake attribute, K represents the number of cluster centers, σ represents the width of the Gaussian window, t represents the number of iterations, ω ik (t) is the weight; The strength of the membership relationship between the training sample and the neurons in the membership layer is determined according to the following formula: The weights are updated based on the strength of the membership relationship between the training samples and the neurons in the membership layer: Among them, α represents the learning factor index; If the updated weights and the weights before the update meet the preset conditions, it is determined that the dessert recognition model training is completed.

6. The method according to claim 5, characterized in that If the updated weights and the weights before the update meet the preset conditions, the dessert recognition model training is determined to be complete, including: Calculate the difference between the updated weight and the weight before the update; If the absolute value of the difference is less than the preset threshold, it is determined that the dessert recognition model training is completed.

7. The method according to claim 1, characterized in that The method further includes: inputting actual values ​​of sensitive attributes of each position in the target area into a trained dessert recognition model to obtain a dessert type at each position; Determine the sweet spot distribution information of the target area according to the sweet spot type of each location.

8. A dessert identification device, characterized in that: include: The pre-stack forward modeling module is used to perform pre-stack forward modeling on multiple pre-established sweet spot mechanism models to obtain simulated seismic data; the sweet spot mechanism model is established based on the sweet spot type, rock physical parameters and contact relationship between rocks in the target area; A sensitive attribute determination module is used to extract attributes from the simulated seismic data to obtain initial attributes, and to determine sensitive attributes for model training from the initial attributes; The neural network training module is used to determine each sensitive attribute value as a training sample, and train the fuzzy self-organizing neural network for dessert recognition through the training sample to obtain a trained dessert recognition model; the trained dessert recognition model is used to identify the dessert type in the target area.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the dessert identification method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the dessert identification method according to any one of claims 1 to 7 when executed.

Images