Disclosure of Invention
In view of this, the disclosure provides a method and a system for monitoring the construction of sand control operation of an oil well, which can timely find the risk of sand production of a stratum, evaluate the risk level, and take corresponding countermeasures to reduce the risk of oil well safety accidents so as to ensure the oil well safety operation.
According to an aspect of the present disclosure, there is provided a method for monitoring construction of sand control operation of an oil well, comprising: acquiring a surface image of reservoir rock through a camera; performing image feature analysis on the surface image of the reservoir rock to obtain the Dan Biaoguan features of the reservoir rock; and determining a risk level tag value for formation sand production based on the reservoir rock Dan Biaoguan characteristics.
In the above oil well sand control operation construction monitoring method, performing image feature analysis on the surface image of the reservoir rock to obtain the Dan Biaoguan feature of the reservoir rock, including: performing image enhancement processing on the surface image of the reservoir rock to obtain an enhanced reservoir rock Dan Biaoguan image; and extracting features from the enhanced reservoir rock apparent image by an image feature extractor based on a deep neural network model to obtain the reservoir rock Dan Biaoguan features.
In the above method for monitoring the construction of sand control operation of an oil well, performing image enhancement processing on the surface image of the reservoir rock to obtain an enhanced image of the reservoir rock Dan Biaoguan, comprising: and carrying out bilateral filtering on the surface image of the reservoir rock to obtain an enhanced apparent image of the reservoir rock.
In the above oil well sand control operation construction monitoring method, extracting features from the enhanced reservoir rock apparent image by an image feature extractor based on a deep neural network model to obtain the reservoir rock Dan Biaoguan features, including: passing the enhanced reservoir rock apparent image through a pyramid network-based image feature extractor to obtain a reservoir rock apparent shallow feature map, a reservoir rock apparent middle layer feature map and a reservoir rock apparent deep feature map; arranging the reservoir rock apparent shallow layer feature map, the reservoir rock apparent middle layer feature map and the reservoir rock apparent deep layer feature map into a reservoir rock Dan Biaoguan multi-scale feature map along a channel dimension; and performing feature enhancement on the multi-scale feature map of the reservoir rock Dan Biaoguan to obtain a multi-scale enhanced feature map of the reservoir rock Dan Biaoguan as the reservoir rock Dan Biaoguan features.
In the above oil well sand control operation construction monitoring method, passing the enhanced reservoir rock apparent image through an image feature extractor based on a pyramid network to obtain a reservoir rock apparent shallow feature map, a reservoir rock apparent middle layer feature map and a reservoir rock apparent deep feature map, comprising: passing the enhanced reservoir rock apparent image through a first convolution module of the pyramid network-based image feature extractor to obtain the reservoir rock apparent shallow feature map; passing the enhanced reservoir rock apparent image through a second convolution module of the pyramid network-based image feature extractor to obtain the reservoir rock apparent mid-layer feature map; and passing the enhanced reservoir rock apparent image through a third convolution module of the pyramid network-based image feature extractor to obtain the reservoir rock apparent deep feature map.
In the above method for monitoring the construction of sand control operation of an oil well, performing feature enhancement on the multi-scale feature map of the reservoir rock Dan Biaoguan to obtain a multi-scale enhanced feature map of the reservoir rock Dan Biaoguan as the feature of the reservoir rock Dan Biaoguan, including: the reservoir rock Dan Biaoguan multiscale feature map is passed through a three-branch attention module to obtain the reservoir rock Dan Biaoguan multiscale enhanced feature map.
In the above method for monitoring the construction of sand control operation of an oil well, the method for obtaining the multi-scale strengthening characteristic map of the reservoir rock Dan Biaoguan by passing the multi-scale characteristic map of the reservoir rock Dan Biaoguan through a three-branch attention module comprises the following steps: passing the reservoir rock Dan Biaoguan multiscale feature map through a first attention branch to obtain a first multi-modal joint attention feature map; passing the reservoir rock Dan Biaoguan multiscale feature map through a second attention branch to obtain a second multi-modal joint attention feature map; passing the reservoir rock Dan Biaoguan multiscale feature map through a third attention branch to obtain a third multi-modal joint attention feature map; and merging the first multi-modal joint attention profile, the second multi-modal joint attention profile, and the third multi-modal joint attention profile to obtain the reservoir rock Dan Biaoguan multi-scale enhanced profile.
In the above oil well sand control operation construction monitoring method, determining a risk level tag value of the formation sand production based on the reservoir rock Dan Biaoguan characteristics includes: performing feature distribution optimization on the multi-scale reinforcement feature map of the reservoir rock Dan Biaoguan to obtain an optimized multi-scale reinforcement feature map of the reservoir rock Dan Biaoguan; and passing the optimized reservoir rock Dan Biaoguan multiscale enhanced feature map through a classifier to obtain a classification result, wherein the classification result is used for representing a risk grade label value of the formation sand.
In the above method for monitoring the construction of sand control operation of an oil well, performing feature distribution optimization on the multi-scale reinforcement feature map of the reservoir rock Dan Biaoguan to obtain an optimized multi-scale reinforcement feature map of the reservoir rock Dan Biaoguan, including: performing channel dimension traversing flow form convex optimization on each feature matrix of the reservoir rock Dan Biaoguan multi-scale reinforcement feature map along the channel dimension by using the following optimization formula to obtain the optimized reservoir rock Dan Biaoguan multi-scale reinforcement feature map; wherein, the optimization formula is:wherein (1)>Is the individual feature matrix along the channel dimension of the reservoir rock Dan Biaoguan multiscale enhanced feature map,/for each feature matrix along the channel dimension>And->Column vectors and row vectors respectively obtained by linear transformation of global mean pooling vectors composed of global means of all feature matrices of the reservoir rock Dan Biaoguan multiscale enhanced feature map, and +_>Representing the spectral norms of the matrix +.>Representing vector multiplication, ++>Represents multiplication by location, and +.>Each feature matrix along the channel dimension of the multi-scale enhanced feature map for the optimized reservoir rock Dan Biaoguan.
According to another aspect of the present disclosure, there is provided an oil well sand control operation construction monitoring system, comprising: the image acquisition module is used for acquiring the surface image of the reservoir rock through the camera; the image feature analysis module is used for carrying out image feature analysis on the surface image of the reservoir rock to obtain the Dan Biaoguan features of the reservoir rock; and a risk level determination module for determining a risk level tag value for formation sand production based on the reservoir rock Dan Biaoguan characteristics.
In the above oil well sand control operation construction monitoring system, the image feature analysis module includes: the image enhancement processing unit is used for carrying out image enhancement processing on the surface image of the reservoir rock to obtain an enhanced reservoir rock Dan Biaoguan image; and an image feature extraction unit for extracting features of the enhanced reservoir rock apparent image by an image feature extractor based on a deep neural network model to obtain the reservoir rock Dan Biaoguan features.
According to an embodiment of the disclosure, a surface image of reservoir rock is first acquired through a camera, then image feature analysis is performed on the surface image of the reservoir rock to obtain reservoir rock Dan Biaoguan features, and then a risk level tag value of the formation sand production is determined based on the reservoir rock Dan Biaoguan features. Therefore, the risk of sand production of the stratum can be timely found, the risk level is evaluated, and corresponding countermeasures are adopted to reduce the risk of oil well safety accidents so as to ensure the oil well safety operation.
Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.
Description of the embodiments
The following description of the embodiments of the present disclosure will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some, but not all embodiments of the disclosure. All other embodiments, which can be made by one of ordinary skill in the art without undue burden based on the embodiments of the present disclosure, are also within the scope of the present disclosure.
As used in this disclosure and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Various exemplary embodiments, features and aspects of the disclosure will be described in detail below with reference to the drawings. In the drawings, like reference numbers indicate identical or functionally similar elements. Although various aspects of the embodiments are illustrated in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
In addition, numerous specific details are set forth in the following detailed description in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art have not been described in detail in order not to obscure the present disclosure.
Aiming at the technical problems, the technical concept of the present disclosure is to collect the surface image of the reservoir rock through a camera, and introduce image processing and analysis technology at the rear end to analyze the surface image of the reservoir rock to judge the risk level of the sand production of the stratum, thereby realizing the automatic monitoring of the sand control operation construction of the oil well. Through the method, the problems of low efficiency caused by expert intervention and low accuracy caused by incomplete monitoring and subjective factors can be avoided, so that the risk of sand production of the stratum can be timely found, the risk level is evaluated, corresponding countermeasures are adopted to reduce the risk of oil well safety accidents, and oil well safety operation is ensured.
FIG. 1 illustrates a flow chart of a method of monitoring construction of an oil well sand control operation according to an embodiment of the present disclosure. Fig. 2 shows a schematic architecture diagram of a method of monitoring construction of sand control operations of an oil well according to an embodiment of the present disclosure. As shown in fig. 1 and 2, the construction monitoring method for sand control operation of an oil well according to an embodiment of the present disclosure includes the steps of: s110, acquiring surface images of reservoir rock through a camera; s120, performing image feature analysis on the surface image of the reservoir rock to obtain the Dan Biaoguan features of the reservoir rock; and S130, determining a risk level tag value of the sand produced in the stratum based on the characteristics of the reservoir rock Dan Biaoguan.
Specifically, in the technical solution of the present disclosure, first, a surface image of reservoir rock acquired by a camera is acquired. It should be understood that during the actual process of acquiring the surface image of the reservoir rock, the image may be affected by external light and shadows, so that the quality of the surface image of the reservoir rock is low, and there may be problems such as noise, blurring or excessively low contrast in the image, which may greatly affect the performance of the subsequent machine vision algorithm, thereby affecting the accuracy of apparent detection of the reservoir rock, and reducing the accuracy of risk level assessment of the formation sand. Therefore, the surface image of the reservoir rock needs to be image enhanced prior to analysis in order to better identify and analyze the apparent details and texture features of the reservoir rock for more accurate formation sand risk level assessment. In particular, in one specific example of the present disclosure, bilateral filtering may be employed as an image enhancement method, which enables unnecessary image details such as noise and smoothing to be removed while preserving as much as possible the apparent structural details and edges of reservoir rock. Meanwhile, the bilateral filter can be based on a Gaussian filter function distributed in a spatial domain according to the characteristics of the image, so that the problem that pixels far away affect edge pixels is effectively solved, the purpose of protecting the edges from noise removal is achieved, namely, the relation among the pixels is utilized to adjust filter parameters, so that an enhancement effect which is more accurate and suitable for actual demands is obtained, and the image quality and the accuracy and reliability of an analysis result are improved.
Feature mining of the enhanced reservoir rock apparent image is then performed using a convolutional neural network model-based feature extractor that has excellent performance in implicit feature extraction of the image. In particular, it is considered that in actually performing the apparent feature detection of the reservoir rock, not only the structural and compositional deep feature information of the reservoir rock but also the features such as the texture and shape of the surface thereof should be focused more. Therefore, in order to further improve the accuracy and sufficiency of the apparent condition detection of the reservoir rock, in the technical solution of the present disclosure, the enhanced reservoir rock apparent image is further passed through an image feature extractor based on a pyramid network to obtain a reservoir rock apparent shallow feature map, a reservoir rock apparent middle layer feature map and a reservoir rock apparent deep feature map. By the method for extracting the multi-level features, the feature information of different levels of the image can be gradually extracted, so that the apparent state feature information of the reservoir rock can be better understood and represented, and the risk level of the sand production of the stratum can be accurately evaluated.
Further, after the shallow layer characteristic, the middle layer characteristic and the deep layer characteristic information about the apparent state of the reservoir rock in the enhanced reservoir rock apparent image are obtained respectively, the reservoir rock apparent shallow layer characteristic image, the reservoir rock apparent middle layer characteristic image and the reservoir rock apparent deep layer characteristic image are required to be fused so as to preserve multi-level information in the image to evaluate the risk level of the formation sand. Thus, in the technical solution of the present disclosure, the reservoir rock apparent shallow layer feature map, the reservoir rock apparent middle layer feature map and the reservoir rock apparent deep layer feature map are arranged along the channel dimension into a reservoir rock Dan Biaoguan multi-scale feature map, so as to better describe the features and properties of the reservoir rock.
The reservoir rock Dan Biaoguan multiscale feature map is then passed through a three-branch attention module to obtain a reservoir rock Dan Biaoguan multiscale enhanced feature map. It should be understood that, because the multi-scale feature map of the reservoir rock Dan Biaoguan includes complementary information between shallow features, middle features and deep features related to the apparent state of the reservoir rock in the enhanced apparent image of the reservoir rock, in the technical solution of the present disclosure, the three-branch attention module is further used to enhance interactions of features at various positions in the multi-scale feature map of the reservoir rock Dan Biaoguan in a spatial dimension, interactions in the spatial dimension and a channel dimension, and interactions between different channel dimensions to more effectively mine and attach importance to complementary feature information, so as to be beneficial for expressing the apparent state features of the reservoir rock more fully, thereby improving accuracy of evaluation of the stratum sand production risk level.
Accordingly, as shown in fig. 3, performing image feature analysis on the surface image of the reservoir rock to obtain reservoir rock Dan Biaoguan features, including: s121, performing image enhancement processing on the surface image of the reservoir rock to obtain an enhanced reservoir rock Dan Biaoguan image; and S122, performing feature extraction on the enhanced reservoir rock apparent image through an image feature extractor based on a deep neural network model to obtain the reservoir rock Dan Biaoguan features.
More specifically, in step S121, an image enhancement process is performed on the surface image of the reservoir rock to obtain an enhanced reservoir rock Dan Biaoguan image, including: and carrying out bilateral filtering on the surface image of the reservoir rock to obtain an enhanced apparent image of the reservoir rock. It should be noted that bilateral filtering is an image processing technology, which performs smoothing processing on an image while maintaining image edge information, and combines information of a spatial domain and a gray domain, and adjusts weights of filters by considering spatial distances and gray differences between pixels, thereby enhancing and denoising the image. The main purpose of bilateral filtering is to smooth an image while preserving edge information, which can effectively remove noise in the image while keeping details and edges of the image clear. Compared with other filtering methods, the bilateral filtering can better maintain the structure and texture information of the image, and the blurring effect possibly caused by the traditional smoothing filter is avoided. The surface image of the reservoir rock is subjected to bilateral filtering treatment, so that the apparent image of the reservoir rock can be enhanced, the apparent image is clearer, the details are more prominent, noise interference possibly existing is removed, the image quality is improved, and a more reliable data base can be provided for subsequent analysis and treatment of the reservoir rock.
More specifically, in step S122, as shown in fig. 4, feature extraction is performed on the enhanced reservoir rock apparent image by an image feature extractor based on a deep neural network model to obtain the reservoir rock Dan Biaoguan features, including: s1221, enabling the enhanced reservoir rock apparent image to pass through an image feature extractor based on a pyramid network to obtain a reservoir rock apparent shallow feature map, a reservoir rock apparent middle layer feature map and a reservoir rock apparent deep feature map; s1222, arranging the reservoir rock apparent shallow layer feature map, the reservoir rock apparent middle layer feature map and the reservoir rock apparent deep layer feature map into a reservoir rock Dan Biaoguan multi-scale feature map along a channel dimension; and S1223, performing feature enhancement on the multi-scale feature map of the reservoir rock Dan Biaoguan to obtain a multi-scale enhanced feature map of the reservoir rock Dan Biaoguan as the feature of the reservoir rock Dan Biaoguan. It should be noted that the pyramid network is a deep neural network model, which processes images through multi-level feature extractors to obtain feature representations of different scales, the structure of the pyramid network is similar to that of a pyramid, the bottom is an original image, and the top is the most abstract feature representation. The pyramid network is mainly used for extracting multi-scale characteristics of the image, so that detail and structural information of different layers in the image are captured. By extracting features at different levels, the pyramid network can take into account both local details and global structure, enabling the network to better understand the content and semantics of the image. In step S1221, the enhanced reservoir rock Dan Biaoguan image is extracted into a shallow, a middle and a deep feature map, respectively, by a pyramid network based image feature extractor. These feature maps correspond to different scales, respectively, and may provide feature representations of different levels. The shallow, middle and deep feature maps are arranged along the channel dimension to form a reservoir rock Dan Biaoguan multi-scale feature map. By doing so, the characteristic information of different scales can be integrated together, so that more comprehensive information is provided for subsequent characteristic processing and analysis.
In step S1223, the multi-scale feature map of the reservoir rock Dan Biaoguan is feature enhanced to obtain a multi-scale enhanced feature map of the reservoir rock Dan Biaoguan. This step can further enhance the useful information in the profile and enhance the expressive power of the features, thereby better characterizing and characterizing the reservoir rock.
More specifically, in step S1221, as shown in fig. 5, passing the enhanced reservoir rock apparent image through a pyramid network-based image feature extractor to obtain a reservoir rock apparent shallow feature map, a reservoir rock apparent middle layer feature map, and a reservoir rock apparent deep feature map, comprising: s12211 passing the enhanced reservoir rock apparent image through a first convolution module of the pyramid network-based image feature extractor to obtain the reservoir rock apparent shallow feature map; s12212 passing the enhanced reservoir rock apparent image through a second convolution module of the pyramid network-based image feature extractor to obtain the reservoir rock apparent middle layer feature map; and S12213 passing the enhanced reservoir rock apparent image through a third convolution module of the pyramid network-based image feature extractor to obtain the reservoir rock apparent deep feature map. It will be appreciated that feature extraction is an important task in image processing, which converts the original image into a more representative feature map for subsequent analysis and processing, and after enhancement of the reservoir rock Dan Biaoguan image, by feature extraction via a pyramid network-based image feature extractor, shallow, middle and deep features of the reservoir rock appearance can be obtained. Passing the enhanced reservoir rock apparent image through a first convolution module of an image feature extractor based on a pyramid network to obtain a shallow feature map of the reservoir rock apparent, the shallow feature map typically containing some low-level image features, such as edges, textures, etc., which can be used to characterize the reservoir rock detail information; passing the enhanced reservoir rock apparent image through a second convolution module of the pyramid network-based image feature extractor to obtain a middle layer feature map of the reservoir rock Dan Biaoguan, the middle layer feature map typically containing some middle-level image features, such as shapes, corner points, etc., which may be used to characterize the morphology and structural information of the reservoir rock; the enhanced reservoir rock apparent image is passed through a third convolution module of the pyramid network-based image feature extractor to obtain a deep feature map of the reservoir rock Dan Biaoguan, which typically contains some high-level image features, such as portions, whole bodies, etc., that can be used to characterize the overall features and combination information of the reservoir rock. By extracting the characteristic diagrams of different layers, the characteristics of the reservoir rock Dan Biaoguan can be obtained more comprehensively and diversified, the characteristics and the properties of the reservoir rock can be represented more accurately, and the characteristic diagrams can be used for subsequent tasks such as classification, segmentation and detection, and the understanding and analysis capability of the reservoir rock are improved.
More specifically, in step S1223, feature strengthening the multi-scale feature map of the reservoir rock Dan Biaoguan to obtain a multi-scale strengthening feature map of the reservoir rock Dan Biaoguan as the reservoir rock Dan Biaoguan feature includes: the reservoir rock Dan Biaoguan multiscale feature map is passed through a three-branch attention module to obtain the reservoir rock Dan Biaoguan multiscale enhanced feature map. It is worth mentioning that the three-branch attention module is an attention mechanism module for image processing, which is used for enhancing the multi-scale characteristic map of the reservoir rock Dan Biaoguan. The three-branch attention module can learn the importance weight of each pixel point in the feature diagrams with different scales, so that the most representative feature is selected, the influence of redundant information and noise is reduced, and the robustness and the expression capability of the feature are improved. The multi-scale feature map of the reservoir rock Dan Biaoguan contains feature information on different scales, and the three-branch attention module can fuse the features to obtain more comprehensive and rich feature expression, and the detail and structure information of the reservoir rock can be better captured by comprehensively considering the feature weights on different scales. The reservoir rock Dan Biaoguan multiscale feature map can be further enhanced by the processing of the three-branch attention module, which means that the useful information in the feature map will be enhanced, making the features more prominent and obvious, facilitating subsequent classification, segmentation or other image processing tasks. The three-branch attention module plays roles in feature selection, scale fusion and feature reinforcement in the processing of the multi-scale feature map of the reservoir rock Dan Biaoguan, and can improve the expression capability and the discrimination of the apparent features of the reservoir rock.
Further, as shown in fig. 6, passing the reservoir rock Dan Biaoguan multiscale feature map through a three-branch attention module to obtain the reservoir rock Dan Biaoguan multiscale enhanced feature map includes: s12231, enabling the reservoir rock Dan Biaoguan multiscale feature map to pass through a first attention branch to obtain a first multi-mode joint attention feature map; s12232, enabling the reservoir rock Dan Biaoguan multiscale feature map to pass through a second attention branch to obtain a second multi-mode joint attention feature map; s12233, enabling the reservoir rock Dan Biaoguan multiscale feature map to pass through a third attention branch to obtain a third multi-mode joint attention feature map; and S12234, fusing the first multi-modal joint attention profile, the second multi-modal joint attention profile, and the third multi-modal joint attention profile to obtain the reservoir rock Dan Biaoguan multi-scale reinforcement profile.
Further, the multi-scale reinforced feature map of the reservoir rock Dan Biaoguan is passed through a classifier to obtain a classification result, wherein the classification result is used for representing a risk grade label value of the sand production of the stratum. That is, classification processing is performed by using the characteristic-enhanced multi-scale characteristic information related to the apparent state of the reservoir rock, so that the risk level of the formation sand is evaluated and judged based on the actual condition of the apparent state of the reservoir rock. Specifically, in the technical scheme of the disclosure, the label of the classifier is a risk level label value of the sand production of the stratum, so that after the classification result is obtained, the risk level of the sand production of the stratum can be evaluated to realize automatic monitoring of sand control operation construction of the oil well.
Accordingly, as shown in fig. 7, determining a risk level tag value for formation sand production based on the reservoir rock Dan Biaoguan characteristics includes: s131, performing feature distribution optimization on the multi-scale reinforcement feature map of the reservoir rock Dan Biaoguan to obtain an optimized multi-scale reinforcement feature map of the reservoir rock Dan Biaoguan; and S132, enabling the optimized reservoir rock Dan Biaoguan multiscale reinforcement feature map to pass through a classifier to obtain a classification result, wherein the classification result is used for representing a risk grade label value of the sand production of the stratum.
In particular, in the technical solution of the present disclosure, when the enhanced reservoir rock apparent image is passed through an image feature extractor based on a pyramid network, the reservoir rock apparent shallow feature map, the reservoir rock apparent middle layer feature map and the reservoir rock apparent deep feature map express image semantic features based on different feature extraction scales of the pyramid network in different depth dimensions, respectively, and when the reservoir rock apparent shallow feature map, the reservoir rock apparent middle layer feature map and the reservoir rock apparent deep feature map are arranged along a channel dimension to obtain the reservoir rock Dan Biaoguan multi-scale feature map, there is also a feature expression difference caused by a feature depth dimension and a spatial correlation dimension between feature matrices of the reservoir rock Dan Biaoguan multi-scale feature map.
Further, when the multi-scale feature map of the reservoir rock Dan Biaoguan passes through the three-branch attention module, whether the spatial attention mechanism of the local spatial feature distribution of the enhanced feature matrix or the channel attention mechanism of the overall feature distribution of some feature matrices in the enhanced channel dimension is the feature expression difference between the feature matrices of the multi-scale enhanced feature map of the reservoir rock Dan Biaoguan is further increased, so that the manifold geometry difference of the feature manifold expressions of the feature matrices causes the problem of poor manifold geometry continuity of the multi-scale enhanced feature map of the reservoir rock Dan Biaoguan, and the accuracy of the classification result obtained by the classifier is affected. Thus, the applicant of the present disclosure addresses each feature matrix along the channel dimension of the reservoir rock Dan Biaoguan multiscale enhanced feature map, e.g., denoted asAnd performing channel dimension traversal flow form convex optimization of the feature map.
Accordingly, in one specific example, the feature distribution optimization of the reservoir rock Dan Biaoguan multiscale reinforcement feature map to obtain an optimized reservoir rock Dan Biaoguan multiscale reinforcement feature map includes: performing channel dimension traversing flow form convex optimization on each feature matrix of the reservoir rock Dan Biaoguan multi-scale reinforcement feature map along the channel dimension by using the following optimization formula to obtain the optimized reservoir rock Dan Biaoguan multi-scale reinforcement feature map; wherein, the optimization formula is: Wherein (1)>Is the individual feature matrix along the channel dimension of the reservoir rock Dan Biaoguan multiscale enhanced feature map,/for each feature matrix along the channel dimension>And->Column vectors and row vectors respectively obtained by linear transformation of global mean pooling vectors composed of global means of all feature matrices of the reservoir rock Dan Biaoguan multiscale enhanced feature map, and +_>Representing the spectral norms of the matrix +.>Representing vector multiplication, ++>Represents multiplication by location, and +.>Each feature matrix along the channel dimension of the multi-scale enhanced feature map for the optimized reservoir rock Dan Biaoguan.
Here, the channel dimension traversal manifold optimization of the reservoir rock Dan Biaoguan multiscale enhanced feature map determines the base dimension of the feature matrix manifold by structuring the maximum distribution dense direction of the modulated feature matrices and traverses the feature matrix manifold along the channel direction of the reservoir rock Dan Biaoguan multiscale enhanced feature map to constrain each feature matrix by stacking the base dimension of the traversal manifold along the channel directionConvex optimization of the continuity of the represented traversing manifold, thereby realizing the optimization of the feature matrix +.>The geometric continuity of the high-dimensional feature manifold of the multi-scale enhanced feature map of the reservoir rock Dan Biaoguan composed of the traversing manifold to promote the accuracy of the classification result obtained by the classifier. In this way, the stratum sand production risk level can be automatically evaluated based on the apparent state condition of the reservoir rock, so that the automatic monitoring of the oil well sand production operation construction is realized, and corresponding stratum sand production risks of different levels can be adopted in time The countermeasures of the method can reduce the risk of oil well safety accidents and ensure the safety operation of the oil well.
Further, passing the optimized reservoir rock Dan Biaoguan multiscale enhanced feature map through a classifier to obtain a classification result, wherein the classification result is used for representing a risk grade label value of the sand production of the stratum, and the classification result comprises the following steps: expanding the optimized reservoir rock Dan Biaoguan multiscale reinforcement feature map into optimized classification feature vectors according to row vectors or column vectors; performing full-connection coding on the optimized classification feature vector by using a full-connection layer of the classifier to obtain a coding classification feature vector; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
It should be appreciated that the role of the classifier is to learn classification rules with a given class, known training data, and then classify (or predict) unknown data. Logistic regression (logistics), SVM, etc. are commonly used to solve the classification problem, and for multi-classification problems (multi-class classification), logistic regression or SVM can be used as well, but multiple bi-classifications are required to compose multiple classifications, but this is error-prone and inefficient, and the commonly used multi-classification method is the Softmax classification function.
It is worth mentioning that the fully connected layer (Fully Connected Layer) is a common layer type in the neural network, also called dense connected layer or fully connected layer, and is used for connecting each element of the input data to each neuron of the output layer, in the image processing task, the image features are usually unfolded into one-dimensional vectors, and then encoded and processed through the fully connected layer, and the fully connected layer can learn higher-level feature representation in the input data, so that the network can better understand and classify the input data. The full-connection coding (Fully Connected Encoding) refers to coding input data through a full-connection layer to obtain coding classification feature vectors, wherein the full-connection coding process can be understood as mapping the input data into a high-dimensional feature space, and performing linear transformation and nonlinear activation through learned weights and offsets to obtain more abstract and advanced feature expression on the input data, and the coding classification feature vectors can better represent the features of the input data and provide more differentiated feature expression for subsequent classification tasks. In the processing of the reservoir rock Dan Biaoguan multiscale enhanced feature map, fully connected layers and fully connected encodings are used to convert the optimized feature map into encoded classification feature vectors. By linear transformation and nonlinear activation of the fully connected layers, more representative feature information can be extracted. And then, inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain a final classification result, wherein the final classification result is used for representing the risk level label value of the layer sand. The full-connection layer and the full-connection coding can improve the understanding and classifying ability of the classifier to the characteristics, thereby improving the accuracy and reliability of the classifying result.
In summary, according to the oil well sand control operation construction monitoring method based on the embodiment of the disclosure, the risk of sand production of a stratum can be found timely, the risk level is evaluated, and accordingly corresponding countermeasures are taken to reduce the risk of oil well safety accidents so as to ensure oil well safety operation.
FIG. 8 illustrates a block diagram of an oil well sand control operation construction monitoring system 100 according to an embodiment of the present disclosure. As shown in fig. 8, an oil well sand control operation construction monitoring system 100 according to an embodiment of the present disclosure includes: an image acquisition module 110 for acquiring a surface image of the reservoir rock by a camera; the image feature analysis module 120 is used for performing image feature analysis on the surface image of the reservoir rock to obtain the Dan Biaoguan features of the reservoir rock; and a risk level determination module 130 for determining a risk level tag value for formation sand production based on the reservoir rock Dan Biaoguan characteristics.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described oil well sand control operation construction monitoring system 100 have been described in detail in the above description of the oil well sand control operation construction monitoring method with reference to fig. 1 to 7, and thus, repetitive descriptions thereof will be omitted.
As described above, the oil well sand control operation construction monitoring system 100 according to the embodiment of the present disclosure may be implemented in various wireless terminals, such as a server having an oil well sand control operation construction monitoring algorithm, or the like. In one possible implementation, the well sand control operation construction monitoring system 100 according to embodiments of the present disclosure may be integrated into a wireless terminal as one software module and/or hardware module. For example, the well sand control operation construction monitoring system 100 may be a software module in the operating system of the wireless terminal or may be an application developed for the wireless terminal; of course, the well sand control operation construction monitoring system 100 may also be one of a number of hardware modules of the wireless terminal.
Alternatively, in another example, the well sand control operation construction monitoring system 100 and the wireless terminal may be separate devices, and the well sand control operation construction monitoring system 100 may be connected to the wireless terminal through a wired and/or wireless network and transmit interactive information in a agreed data format.
Fig. 9 illustrates an application scenario diagram of an oil well sand control operation construction monitoring method according to an embodiment of the present disclosure. As shown in fig. 9, in this application scenario, first, a surface image of reservoir rock is acquired by a camera (e.g., D shown in fig. 9), and then the surface image of reservoir rock is input into a server (e.g., S shown in fig. 9) where an oil well sand control operation monitoring algorithm is deployed, wherein the server can process the surface image of reservoir rock using the oil well sand control operation monitoring algorithm to obtain a classification result for a risk level tag value representing formation sand production.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.