Well wall instability type prediction method, device and system

Abstract

The invention discloses a well wall instability type prediction method, device and system. In order to quickly analyze the classification characteristics of a large number of falling blocks, full-automatic sorting and three-dimensional scanning are carried out on the falling blocks on the well wall, and complete three-dimensional point cloud data of the falling blocks are obtained. The method comprises the following steps: preprocessing original point cloud data, and extracting feature parameters from the chipped point cloud data through a feature extraction algorithm; and training falling block data collected on site to obtain an optimal machine learning algorithm, optimizing a feature parameter improvement classification model, and integrally developing a well wall falling block image feature parameter intelligent extraction and classification system. Falling blocks generated by on-site well drilling are subjected to full-automatic scanning and intelligent classification, the types of the falling blocks and well wall instability reasons are judged, and on-site well wall instability diagnosis and timely implementation of prevention and control measures are supported.

Description

Technical Field

[0001] The present invention relates to the field of geological drilling technology, and in particular to a method, device and system for predicting the type of wellbore instability. Background Art

[0002] Current research shows that different causes of wellbore instability will produce different forms of falling blocks. Based on field experience and simulation analysis, some experts and scholars have linked the form of falling blocks with the wellbore instability mechanism and have made detailed classifications and descriptions of falling blocks produced by different drilling damage forms. Therefore, if the falling blocks returned during the drilling process can be analyzed in a timely manner, corresponding remedial measures can be taken during the drilling construction process based on the information of falling blocks to prevent further instability of the wellbore.

[0003] With the development of sensing technology and automation technology, computer vision and machine learning algorithms are gradually being used to replace manual judgment of the category of block drop. At present, a large number of studies have established the relationship between the morphology of block drop and wellbore instability, and the block drop categories have evolved from fuzzy to detailed, the classification indicators have evolved from qualitative to quantitative, and the classification methods have evolved from subjective judgment of experts and scholars to automatic classification by machine learning classification models.

[0004] However, the current block classification still mainly uses two-dimensional RGB images for image processing. Although this type of image has the characteristics of fast speed, it does not have depth information itself and is greatly affected by the shooting environment. It is mainly based on contour analysis and can extract fewer feature parameters, resulting in the proposed features being not comprehensive and lacking representativeness. Three-dimensional images can directly obtain the three-dimensional geometric shape of the block surface, thus providing richer content for model reconstruction and target recognition tasks. However, due to the different shapes and sizes of blocks and the large number of blocks, there is currently no on-site equipment that can automatically and quickly scan the three-dimensional shapes of batch blocks and extract shape features for classification, which restricts the timely implementation of on-site wellbore stability diagnosis and prevention and control measures. Summary of the invention

[0005] The present invention provides a method, device and system for predicting the type of wellbore instability, which can automatically and quickly scan the three-dimensional shapes of batch fallen blocks and extract shape features for classification.

[0006] According to one aspect of the present invention, a method for predicting the type of wellbore instability is provided, comprising:

[0007] Preprocessing the original point cloud data of the well wall debris obtained by three-dimensional scanning to obtain the debris point cloud data;

[0008] Extracting characteristic parameters from the dropped point cloud data, and constructing a characteristic data set according to the characteristic parameters;

[0009] By evaluating the alternative machine learning algorithms, a preferred classification model is determined, the feature data set is identified to determine the block loss classification, and the wellbore instability type is predicted based on the block loss classification.

[0010] Optionally, the preprocessing of the original point cloud data of at least one well wall block fragment obtained by three-dimensional scanning includes:

[0011] Converting the original point cloud data into a PLY file having at least one additional information and a PCD file including the three-dimensional coordinates of each point in the original point cloud data;

[0012] Use random sampling to reduce the amount of data;

[0013] Correct the three-dimensional coordinates of each point where the block is dropped through a preset reference coordinate system;

[0014] Identifying and removing outliers and noise in the raw point cloud data by statistical filtering;

[0015] Each point cloud is clustered based on Euclidean distance.

[0016] Optionally, extracting characteristic parameters from the lost block point cloud data and constructing a characteristic data set according to the characteristic parameters includes:

[0017] Extracting feature parameters from the dropped block point cloud data by using different feature extraction algorithms;

[0018] The data set is constructed according to the characteristic parameters, the number of the block drop point cloud and the classification parameters, and the classification parameters include four types of block drop: fragmentation, angular, plate-like and block-like.

[0019] Optionally, the feature extraction algorithm includes one or more of PCA bounding box method, triangular mesh method, three-axis projection method, local curvature method, model subdivision projection method, ratio of block volume to bounding box volume, ratio of projection area to circumscribed rectangle area, sphericity, roundness and three-axis ratio.

[0020] Optionally, the preferred classification model is evaluated in the following way:

[0021] Divide the data set into 5 subsets, take each of the subsets as a test set in turn, and use the remaining subsets as training sets to train each candidate model to obtain training results, test the training results by using the subsets as the test set, and reflect the performance evaluation results of each candidate model through confusion matrix and accuracy;

[0022] A final evaluation result of each candidate model is obtained according to the five performance evaluation results of each candidate model, and the preferred classification model is determined according to the final evaluation result.

[0023] Optionally, the method further includes:

[0024] Determine the corresponding relationship between the amount of reduction in the number of characteristic parameters and the degree of reduction in the accuracy of the performance evaluation results;

[0025] The number of the characteristic parameters is reduced based on the corresponding relationship.

[0026] Optionally, the predicting of the wellbore instability type according to the block falling classification includes:

[0027] The lithology of the wellbore block fall is determined, and the wellbore instability type is determined based on the lithology and the block fall classification.

[0028] According to another aspect of the present invention, a device for predicting the type of wellbore instability is provided, comprising:

[0029] A point cloud data processing unit is used to pre-process the original point cloud data of the well wall debris obtained by three-dimensional scanning to obtain the debris point cloud data;

[0030] A feature data set construction unit, used to extract feature parameters from the dropped point cloud data, and construct a feature data set according to the feature parameters;

[0031] The classification and identification unit is used to determine the preferred classification model by evaluating the alternative machine learning algorithms, identify the feature data set to determine the block loss classification, and predict the wellbore instability type according to the block loss classification.

[0032] According to another aspect of the present invention, a wellbore instability type prediction system is provided, characterized in that it comprises: a wellbore instability type prediction device, a flexible vibration plate, an automatic feeding bin, a visual device, a SCARA robot, a conveyor belt and a three-dimensional scanning device;

[0033] The automatic feeding bin is used to automatically vibrate the fallen blocks that need to be divided to the flexible vibration plate;

[0034] The flexible vibration plate is used to separate the fallen pieces by vibration;

[0035] The visual system is used to determine the position information of each fallen block and transmit the position information to the SCARA robot;

[0036] The SCARA robot is used to grab the fallen blocks from the well wall according to the position information and place them on the conveyor belt, so that the conveyor belt drives the fallen blocks to move;

[0037] The three-dimensional scanning device is used to obtain the original point cloud data of the well wall fallen block when the well wall fallen block moves to the preset work station in the conveyor belt;

[0038] The wellbore instability type prediction device is used to execute a wellbore instability type prediction method as described in any one of claims 1-7.

[0039] According to another aspect of the present invention, there is provided an electronic device, the electronic device comprising:

[0040] At least one processor; and a memory in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for predicting the type of wellbore instability described in any embodiment of the present invention.

[0041] 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 method for predicting the type of wellbore instability described in any embodiment of the present invention when executed.

[0042] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the method for predicting the type of wellbore instability according to any embodiment of the present invention is implemented.

[0043] The technical solution of the embodiment of the present invention performs fully automatic scanning and intelligent classification of the fallen blocks generated by on-site drilling, realizes on-site detection, identification, and diagnosis of well wall stability, derives the cause of well wall instability, and proposes well wall instability prevention and control measures in real time according to the cause of well wall instability, thereby reducing the drilling complexity caused by well wall instability, improving drilling efficiency, and saving drilling costs.

[0044] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, 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.

[0046] Figure 1 is a flow chart of a method for predicting wellbore instability type provided in Embodiment 1 of the present invention;

[0047] Figure 2 It is a structural schematic diagram of a wellbore instability type prediction device provided by the second embodiment of the present invention;

[0048] Figure 3 It is a structural schematic diagram of a wellbore instability type prediction system provided by the third embodiment of the present invention;

[0049] Figure 4 It is a schematic diagram of the structure of an electronic device for implementing the method for predicting the type of wellbore instability according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 described embodiments 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 creative work should fall within the scope of protection of the present invention.

[0051] It should be noted that the terms "first", "second", etc. in the specification 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 data 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 that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0052] Embodiment 1

[0053] Figure 1 This is a flow chart of a method for predicting the type of wellbore instability provided by the first embodiment of the present invention. This embodiment is applicable to the case of predicting the cause of wellbore instability. The method can be executed by a wellbore instability type prediction device. The wellbore instability type prediction device can be implemented in the form of hardware and / or software. The wellbore instability type prediction device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0054] S110, pre-processing the original point cloud data of the fallen blocks on the well wall obtained by three-dimensional scanning to obtain the fallen block point cloud data.

[0055] Specifically, the point cloud data of the lost blocks is obtained through three-dimensional scanning, and the preprocessing methods of format conversion, downsampling, coordinate conversion, statistical filtering and clustering are adopted on the basis of the original point cloud to prepare for the subsequent extraction of the lost block feature parameters.

[0056] In an embodiment of the present invention, preprocessing the original point cloud data of at least one well wall block obtained by three-dimensional scanning includes:

[0057] Converting the original point cloud data into a PLY file having at least one additional information and a PCD file including the three-dimensional coordinates of each point in the original point cloud data;

[0058] Use random sampling to reduce the amount of data;

[0059] Correct the three-dimensional coordinates of each point where the block is dropped through a preset reference coordinate system;

[0060] Identifying and removing outliers and noise in the raw point cloud data by statistical filtering;

[0061] Each point cloud is clustered based on Euclidean distance.

[0062] Specifically, the original data of the block drop obtained by scanning the automatic capture system of the well wall block drop image is in STL format, and the surface of the three-dimensional object is characterized by triangular facets as the basic unit. In order to meet the requirements of subsequent data processing and block drop feature extraction, the block drop point cloud data is converted into a PLY file with additional information such as normals and a PCD file including the three-dimensional coordinates of the points.

[0063] The amount of data for a single dropped point cloud is generally around 20,000. Random sampling is used to reduce the amount of data. The random sampling algorithm traverses the original point cloud data and determines the number of points to be selected based on the preset sampling rate or number of sampling points. The algorithm then randomly selects these points as sampling results to form new, lower-density point cloud data.

[0064] To correct the coordinates of the missing point cloud, you first need to determine a reference coordinate system. Then, by comparing the feature points in the point cloud data with the corresponding points in the reference coordinate system, you can calculate the necessary transformation parameters, including the translation vector and rotation matrix. These parameters are used to adjust the coordinates of each point in the point cloud data so that it is aligned with the reference coordinate system.

[0065] The point cloud statistical filtering and denoising can identify and remove outliers and noise in the point cloud data through statistical filtering methods to improve the data quality.

[0066] Clustering of point cloud data, clustering point clouds based on the Euclidean distance between points. Through the Euclidean clustering algorithm, point cloud data is clustered according to its geometric features, thereby achieving segmentation and analysis of point clouds.

[0067] S120: extracting feature parameters from the dropped point cloud data, and constructing a feature data set according to the feature parameters.

[0068] The point cloud data is a three-dimensional data set composed of a large number of discrete points, but not every point of data is useful for the classification of the lost blocks. It is necessary to extract representative and relevant features from the original data.

[0069] In an embodiment of the present invention, extracting characteristic parameters from the lost block point cloud data and constructing a characteristic data set according to the characteristic parameters includes:

[0070] Extracting feature parameters from the dropped block point cloud data by using different feature extraction algorithms;

[0071] The data set is constructed according to the characteristic parameters, the number of the block drop point cloud and the classification parameters, and the classification parameters include four types of block drop: fragmentation, angular, plate-like and block-like.

[0072] In an embodiment of the present invention, the feature extraction algorithm includes one or more of the following: PCA bounding box method, triangular mesh method, three-axis projection method, local curvature method, model subdivision projection method, ratio of block volume to bounding box volume, ratio of projection area to circumscribed rectangle area, sphericity, roundness and three-axis ratio.

[0073] Exemplarily, 32 feature parameters are extracted from the dropped block point cloud data through feature extraction algorithms such as PCA bounding box method, triangular mesh method, three-axis projection method, local curvature method, model subdivision projection method, ratio of dropped block volume to bounding box volume, ratio of projection area to circumscribed rectangle area, sphericity, roundness and three-axis ratio.

[0074] The bounding box construction method based on PCA can realize the extraction of the length, width and thickness of the dropped point cloud. First, PCA is used to determine the center of mass and establish the calculation benchmark. Then the covariance matrix is ​​calculated. The main axis direction of the bounding box is determined by solving the eigenvalues ​​and eigenvectors of the covariance matrix. After determining the main axis direction of the bounding box, the X axis is selected to represent the length, the Y axis to represent the width, and the Z axis to represent the thickness.

[0075] A closed reconstruction method based on a triangular meshing algorithm is used to construct a block drop surface model composed of many triangles to calculate the block drop surface area. The projected convex pentahedron algorithm is used to divide the model into a series of convex pentahedrons formed by the projection of triangular facets on the central plane, and then the projection area of ​​these convex pentahedrons on the central plane is calculated. The sum of the volumes of the convex pentahedrons of all triangles on the projection plane is the block drop volume. This method not only takes into account the morphological characteristics of the model, but also ensures the accuracy and efficiency of the calculation.

[0076] The projection area and perimeter of the horizontal plane, longitudinal section and cross section are obtained by projecting the block drop point cloud data to the planes of x=0, y=0 and z=0 respectively, and then the projected plane point clouds are gridded and fitted. The projection perimeter is approximated by accumulating the outer lengths of all grid boundaries, and the sum of the grid areas is approximately the projection area.

[0077] The most prominent plane of the point cloud with missing blocks is also the focus of attention in the classification process. Based on the principles of mathematical optimization and geometry, the plane model that best represents the surface shape of the point cloud is extracted from the three-dimensional point cloud data. For the extracted surface points, the least squares method is used for plane fitting to find a plane model that minimizes the sum of the distances from the surface points to the plane. The parameters of the plane model are adjusted through an iterative optimization algorithm until the preset fitting accuracy is achieved, and then the area of ​​the fitting plane is calculated.

[0078] Based on the local curvature method, the surface roughness of the block, the most significant plane roughness, and the plane curvature dispersion are calculated. Take n points in the neighborhood of a point P in the point cloud, and fit these n points into a quadratic surface. Then the local surface at point P can be approximated by this quadratic surface, and then the coefficients of this quadratic surface are solved by the principle of least squares method to obtain the surface equation. At the same time, for the maximum fitting plane, the mean value of the curvature value and the dispersion are calculated as important indicators for block classification.

[0079] The aforementioned parameters are mainly some size parameters. In addition, some dimensionless derivative parameters are introduced to describe the shape of the block.

[0080] PCA is used to reduce the dimensionality of the original data while retaining the main features in the data. First, the covariance matrix of the point cloud data is calculated, and the eigenvalue decomposition of the matrix is ​​performed to obtain three eigenvalues ​​and corresponding eigenvectors arranged from large to small. These three eigenvectors represent the three main directions of data distribution, among which the eigenvector corresponding to the largest eigenvalue is the main direction of data distribution; next, the vertical plane of the midpoint of the main direction is found. First, the midpoint in the main direction is determined, which can be obtained by averaging all points in the main direction; then, according to the perpendicular relationship between the plane and the straight line, the normal vector of the vertical plane should be orthogonal to the main direction vector. Therefore, a vector orthogonal to the main direction vector is constructed as the normal vector of the vertical plane. Finally, a unique plane can be determined through the normal vector and the midpoint in the main direction. This plane is the fitted vertical plane, and then the point cloud is divided into two parts based on this plane to obtain the volume and calculate the volume ratio.

[0081] The ratio of the dropped block volume to the volume of the minimum circumscribed cuboid, and the ratio of the dropped block projection to the minimum circumscribed rectangle area are calculated to measure the approximation between the dropped block and the cuboid. The ratio of the dropped block volume to the bounding box volume, and the ratio of the projection area to the circumscribed rectangle area are obtained to characterize the shape of the dropped block.

[0082] Sphericity is used to describe how close a drop is to a sphere, and is calculated using the volume and surface area of ​​the drop. For projections in the three main directions, roundness is used to characterize how close it is to a circle.

[0083] The description of the classification of dropped blocks focuses more on the morphology of the dropped blocks rather than the size. This is because the size of the dropped blocks is affected by many factors. Therefore, compared with the three-axis size information such as the length, width and height of the dropped blocks, shape information such as the aspect ratio, length-to-height ratio and width-to-height ratio is more worthy of attention.

[0084] Exemplarily, 20 basic feature parameters and 12 secondary feature parameters are extracted from the block point cloud data through 10 feature extraction algorithms, including PCA bounding box method, triangular mesh method, three-axis projection method, local curvature method, model partitioning projection method, ratio of block volume to bounding box volume, ratio of projection area to circumscribed rectangle area, sphericity, roundness and three-axis ratio, for a total of 32 feature parameters. Among them, the basic feature parameters are feature parameters that can be directly obtained through the feature extraction algorithm, and the secondary feature parameters are feature parameters obtained by comparing and calculating other feature parameters. The block point cloud after preprocessing is calculated by the above algorithm to extract the corresponding feature parameters. After removing invalid values ​​or null values ​​(deleting block point clouds with too small size and too many missing blocks), a data set consisting of 978 blocks, 32 feature parameters and 1 classification parameter (including block types, such as fragmented, angular, plate-like, blocky, etc.) is formed.

[0085] S130, determining a preferred classification model by evaluating alternative machine learning algorithms, identifying the feature data set to determine the block loss classification, and predicting the wellbore instability type according to the block loss classification.

[0086] Exemplarily, in order to find a classification model suitable for the block loss classification problem, 10 different machine learning algorithms were used to test the database composed of the obtained feature parameters, including random forest classifier, K nearest neighbor classifier, support vector machine classifier, decision tree classifier, adaptive boosting algorithm, gradient boosting algorithm, extreme gradient boosting algorithm, multilayer perceptron classifier, logistic regression classifier and naive Bayes classifier.

[0087] In the embodiment of the present invention, the preferred classification model is evaluated in the following way:

[0088] Divide the data set into 5 subsets, take each of the subsets as a test set in turn, and use the remaining subsets as training sets to train each candidate model to obtain training results, test the training results by using the subsets as the test set, and reflect the performance evaluation results of each candidate model through confusion matrix and accuracy;

[0089] A final evaluation result of each candidate model is obtained according to the five performance evaluation results of each candidate model, and the preferred classification model is determined according to the final evaluation result.

[0090] The accuracy of each model is verified by using the five-fold cross validation method. The original data set (such as the data set consisting of 32274 data points mentioned above) is divided into five equal and similar subsets, four of which are used to train the model, and the remaining subset is used to test the model. The whole process is repeated five times, each time a different subset is selected as the test set, and the rest is used as the training set. Finally, the five performance evaluation results are averaged to obtain the final evaluation result.

[0091] The confusion matrix and accuracy are used to characterize the accuracy of different classifiers for the block loss classification problem. The confusion matrix is ​​a visual two-dimensional matrix, where the horizontal axis represents the predicted category and the vertical axis represents the true category. The data on the diagonal represent samples with correct predictions, and the data on the non-diagonal represent samples with incorrect predictions. The confusion matrix can be used to understand whether the classification model is correct and what types of errors have occurred in the classification model.

[0092] For the same sample library, the choice of algorithm has a great impact on the trained machine learning classifier. The accuracy of different algorithms ranges from 73.0% to 91.5%. For example, the extreme gradient boosting algorithm has the highest accuracy, reaching 91.5%, so the model classifier adopts the extreme gradient boosting algorithm.

[0093] In an embodiment of the present invention, the method may further include the following steps:

[0094] Determine the corresponding relationship between the amount of reduction in the number of characteristic parameters and the degree of reduction in the accuracy of the performance evaluation results;

[0095] The number of the characteristic parameters is reduced based on the corresponding relationship.

[0096] Data and features determine the upper limit of machine learning, while models and algorithms can only approach this upper limit. Therefore, for the classification of drilling block drop, the design and selection of feature parameters are crucial. In order to remove impurities and redundancy in the original data, it is necessary to design more efficient feature parameters to characterize the relationship between the problem to be solved and the prediction model. In addition, the increase in feature parameters will slow down the calculation speed and may cause overfitting problems. Therefore, it is necessary to reduce the number of feature parameters and extract the parameters that best characterize the characteristics of different categories without affecting the classification effect as much as possible.

[0097] According to the order of feature parameter weight values ​​from low to high, the number of feature parameters is gradually reduced, and the above feature parameters are increased and decreased in the two processes. The accuracy is positively correlated with the number of feature parameters. The more feature parameters there are, the greater the corresponding accuracy is. However, in the process of reduction, the accuracy is not significantly reduced at the beginning. For example, when the feature parameters are reduced by 20, the accuracy is only reduced by 1.5%, but when the feature parameters are reduced from 12 to 11, the accuracy is reduced by 2.1%, which sacrifices too much accuracy. Therefore, 12 morphological features including aspect ratio, width-to-height ratio, length-to-height ratio, Z-axis area ratio, lower volume, plane discreteness, bounding box volume, X-axis area ratio, volume ratio, Z-axis projection area, height, and two-side volume ratio are finally selected as the feature parameters for block drop classification. The calculation speed is doubled, and the accuracy is only reduced by 3%. Finally, these 12 feature parameters are used as the input of the extreme gradient boosting classification model to obtain the optimal drilling block drop classification algorithm.

[0098] In an embodiment of the present invention, the wellbore instability type is predicted according to the block falling classification, including:

[0099] The lithology of the wellbore block fall is determined, and the wellbore instability type is determined based on the lithology and the block fall classification.

[0100] Specifically, the blocks generated by 12 groups of field drilling and indoor tests were fully automatically scanned and intelligently classified. Each group of tests included 14-20 blocks, and the lithology of the blocks included coal rock, shale, mudstone, sandstone and dolomite.

[0101] 182 drilling blocks in 12 groups are automatically divided and sent to the conveyor belt with approximately equidistant spacing of about 300mm. The blocks to be tested are transported to the 3D scanning area through the automatic control conveyor belt in turn, and then automatically perform a full range of 3D scanning, automatically perform data splicing of multiple scanning surfaces, and obtain complete 3D point cloud data of all 182 blocks.

[0102] The intelligent extraction system of the feature parameters of the well wall block image was used to pre-process the point cloud data obtained by three-dimensional scanning for each block, including format conversion, downsampling, coordinate conversion, statistical filtering and clustering. The corresponding feature parameters were extracted for each block after processing, and the 12 optimized feature parameters were extracted to form a data set consisting of 182 blocks, 12 feature parameters and 1 classification parameter, a total of 182×13=2366 data. As shown in Table 1 below, the intelligent classification system of the feature parameters of the well wall block image was used to classify the blocks, and the manual recognition classification results were compared with the system intelligent classification results. Among them, the results of 4 groups reached 100%, and the lowest accuracy of the remaining groups was 85.7%, and the average accuracy reached 93.9%.

[0103]

[0104] Table 1: Comparison table of block type recognition accuracy and wellbore instability types

[0105] The method for predicting the type of wellbore instability provided by the embodiment of the present invention realizes the intelligent extraction and classification of the characteristic parameters of the wellbore block falling image, performs fully automatic scanning and intelligent classification of the block falling generated by on-site drilling, realizes on-site detection, identification, and diagnosis of wellbore stability, derives the cause of wellbore instability, and proposes wellbore instability prevention and control measures in real time according to the cause of wellbore instability, thereby reducing the drilling complexity caused by wellbore instability, improving drilling efficiency, and saving drilling costs.

[0106] Embodiment 2

[0107] Figure 2 is a schematic diagram of the structure of a device for predicting the type of wellbore instability provided by the second embodiment of the present invention. Figure 2 As shown, the device comprises:

[0108] The point cloud data processing unit 210 is used to pre-process the original point cloud data of the well wall debris obtained by three-dimensional scanning to obtain the debris point cloud data;

[0109] A feature data set construction unit 220 is used to extract feature parameters from the lost block point cloud data and construct a feature data set according to the feature parameters;

[0110] The classification identification unit 230 is used to identify the block loss classification of the feature data set by evaluating the alternative machine learning algorithms and determining the preferred classification model, and predict the wellbore instability type according to the block loss classification.

[0111] Optionally, the point cloud data processing unit 210 is specifically configured to execute:

[0112] Converting the original point cloud data into a PLY file having at least one additional information and a PCD file including the three-dimensional coordinates of each point in the original point cloud data;

[0113] Use random sampling to reduce the amount of data;

[0114] Correct the three-dimensional coordinates of each point where the block is dropped through a preset reference coordinate system;

[0115] Identifying and removing outliers and noise in the raw point cloud data by statistical filtering;

[0116] Each point cloud is clustered based on Euclidean distance.

[0117] Optionally, the feature data set construction unit 220 is specifically configured to execute:

[0118] Extracting feature parameters from the dropped block point cloud data by using different feature extraction algorithms;

[0119] The data set is constructed according to the characteristic parameters, the number of the block drop point cloud and the classification parameters, and the classification parameters include four types of block drop: fragmentation, angular, plate-like and block-like.

[0120] Optionally, the feature extraction algorithm includes one or more of PCA bounding box method, triangular mesh method, three-axis projection method, local curvature method, model subdivision projection method, ratio of block volume to bounding box volume, ratio of projection area to circumscribed rectangle area, sphericity, roundness and three-axis ratio.

[0121] Optionally, the classification identification unit 230 is used to evaluate and obtain a preferred classification model in the following manner:

[0122] Divide the data set into 5 subsets, take each of the subsets as a test set in turn, and use the remaining subsets as training sets to train each candidate model to obtain training results, test the training results by using the subsets as the test set, and reflect the performance evaluation results of each candidate model through confusion matrix and accuracy;

[0123] A final evaluation result of each candidate model is obtained according to the five performance evaluation results of each candidate model, and the preferred classification model is determined according to the final evaluation result.

[0124] Optionally, the classification and identification unit 230 is further configured to execute:

[0125] Determine the corresponding relationship between the amount of reduction in the number of characteristic parameters and the degree of reduction in the accuracy of the performance evaluation results;

[0126] The number of the characteristic parameters is reduced based on the corresponding relationship.

[0127] Optionally, the classification and identification unit 230, when performing the prediction of the wellbore instability type according to the block falling classification, specifically performs:

[0128] The lithology of the wellbore block fall is determined, and the wellbore instability type is determined based on the lithology and the block fall classification.

[0129] The wellbore instability type prediction device provided in the embodiment of the present invention can execute the wellbore instability type prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0130] Embodiment 3

[0131] Figure 3 is a schematic diagram of the structure of a wellbore instability type prediction system provided by the third embodiment of the present invention. Figure 3 As shown, the system includes: a well wall instability type prediction device 310, a flexible vibration plate 320, an automatic feeding bin 330, a visual device 340, a SCARA robot 350, a conveyor belt 360 and a three-dimensional scanning device 370;

[0132] The automatic feeding bin 330 is used to automatically vibrate the fallen blocks that need to be divided to the flexible vibration plate 320;

[0133] A flexible vibration plate 320, used to separate the individual fallen pieces by vibration;

[0134] A visual system 340 is used to determine the position information of each fallen block and transmit the position information to the SCARA robot 350;

[0135] The SCARA robot 350 is used to grab the fallen blocks from the well wall according to the position information and place them on the conveyor belt 360 so that the conveyor belt 360 can drive the fallen blocks to move;

[0136] The three-dimensional scanning device 370 is used to obtain the original point cloud data of the well wall fallen block when the well wall fallen block moves to the preset station in the conveyor belt 360;

[0137] The wellbore instability type prediction device 310 is used in a wellbore instability type prediction method provided in any one of the above embodiments.

[0138] Since the wellbore stability problem during drilling is extremely time-sensitive, it is required to monitor the wellbore stability in real time to detect it early and deal with it as soon as possible. In the early stage of wellbore instability, the wellbore blocks are brought to the ground by the drilling fluid, which can reflect the cause and type of wellbore instability. Therefore, it is necessary to judge the type of wellbore instability based on the large number of blocks that come up during drilling, and take corresponding measures to prevent further instability of the wellbore. In order to quickly analyze the classification characteristics of a large number of blocks, an automatic image capture system for wellbore blocks was developed to perform fully automatic sorting and three-dimensional scanning of wellbore blocks to obtain complete three-dimensional point cloud data of the blocks.

[0139] Separate the well wall blocks of different sizes and shapes, and feed them to the conveyor belt with roughly equal spacing of about 300mm. Use flexible vibration plate + automatic feeding bin + vision + SCARA robot (with bionic gripper). First, load the blocks to be separated into the feeding bin, which vibrates to the flexible vibration plate; the vibration plate starts to vibrate as needed to separate the blocks; the vision system takes pictures to confirm the position of the independent blocks, and transmits the position information to the robot. The robot with grippers grabs the independent blocks and places them on the conveyor line.

[0140] After the measured block is transported to the 3D scanning area through the intelligent control conveyor belt, the command that the measured block has reached the specified position is obtained and transmitted to the robot's fully automatic 3D scanning system to send a 3D scanning command. The robot performs an all-round scan according to the pre-programmed scanning path, automatically stitches the data of multiple scanning surfaces, obtains complete 3D data, and saves the 3D scanning data to the specified folder. After the measurement is completed, the scanning system automatically creates a new scanning project to prepare for the next block scan, and at the same time sends instructions to the conveyor belt, which moves to the next station to scan the next block. After the point cloud scan is completed, the software will automatically generate a non-closed mesh model composed of triangular facets, and save the obtained point cloud data as an STL format file.

[0141] The wellbore instability type prediction system provided by the embodiment of the present invention can realize fully automatic sorting and three-dimensional scanning of wellbore broken blocks on site, intelligently extract and classify feature parameters of three-dimensional image data, determine the cause of wellbore instability, and propose corresponding measures to prevent further instability of the wellbore.

[0142] Embodiment 4

[0143] Figure 4 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 required herein.

[0144] like Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and 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 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0145] A number of 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 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 through a computer network such as the Internet and / or various telecommunication networks.

[0146] The processor 11 may be a variety of general and / or special processing components 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 special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a wellbore instability type prediction method.

[0147] In some embodiments, the wellbore instability type prediction method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the wellbore instability type prediction method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the wellbore instability type prediction method in any other appropriate manner (e.g., by means of firmware).

[0148] Various implementations of the systems and techniques described above herein 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 chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including 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.

[0149] 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, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0150] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0151] To provide interaction with a user, the systems and techniques described herein may 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 a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0152] The systems and techniques described herein may be implemented in a computing system that includes backend 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 frontend components (e.g., a user computer with a graphical user interface or a 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 backend components, middleware components, or frontend components. The components of the system may 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.

[0153] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0154] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0155] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting the type of wellbore instability, characterized in that: include: Preprocessing the original point cloud data of the well wall debris obtained by three-dimensional scanning to obtain the debris point cloud data; Extracting characteristic parameters from the dropped point cloud data, and constructing a characteristic data set according to the characteristic parameters; By evaluating the alternative machine learning algorithms, a preferred classification model is determined, the feature data set is identified to determine the block loss classification, and the wellbore instability type is predicted based on the block loss classification.

2. The method according to claim 1, characterized in that The preprocessing of the original point cloud data of at least one well wall block obtained by three-dimensional scanning includes: Converting the original point cloud data into a PLY file having at least one additional information and a PCD file including the three-dimensional coordinates of each point in the original point cloud data; Use random sampling to reduce the amount of data; Correct the three-dimensional coordinates of each point where the block is dropped through a preset reference coordinate system; Identifying and removing outliers and noise in the raw point cloud data by statistical filtering; Each point cloud is clustered based on Euclidean distance.

3. The method according to claim 1, characterized in that The step of extracting characteristic parameters from the lost point cloud data and constructing a characteristic data set according to the characteristic parameters includes: Extracting feature parameters from the dropped block point cloud data respectively by using different feature extraction algorithms; The data set is constructed according to the characteristic parameters, the number of the block drop point cloud and the classification parameters, and the classification parameters include four types of block drop: fragmentation, angular, plate-like and block-like.

4. The method according to claim 3, characterized in that The feature extraction algorithm includes PCA bounding box method, triangular mesh method, three-axis projection method, local curvature method, model segmentation projection method, ratio of dropped block volume to bounding box volume, ratio of projection area to circumscribed rectangle area, sphericity, roundness and three-axis ratio.

5. The method according to claim 1, characterized in that The preferred classification model is evaluated in the following way: Divide the data set into 5 subsets, take each of the subsets as a test set in turn, and use the remaining subsets as training sets to train each candidate model to obtain training results, test the training results by using the subsets as the test set, and reflect the performance evaluation results of each candidate model through confusion matrix and accuracy; A final evaluation result of each candidate model is obtained according to the five performance evaluation results of each candidate model, and the preferred classification model is determined according to the final evaluation result.

6. The method according to claim 1, characterized in that Further including: Determine the corresponding relationship between the amount of reduction in the number of characteristic parameters and the degree of reduction in the accuracy of the performance evaluation results; The number of the characteristic parameters is reduced based on the corresponding relationship.

7. The method according to claim 1, characterized in that The prediction of the wellbore instability type according to the block falling classification includes: The lithology of the wellbore block fall is determined, and the wellbore instability type is determined based on the lithology and the block fall classification.

8. A device for predicting the type of wellbore instability, characterized in that: include: A point cloud data processing unit is used to pre-process the original point cloud data of the well wall debris obtained by three-dimensional scanning to obtain the debris point cloud data; A feature data set construction unit, used to extract feature parameters from the dropped point cloud data, and construct a feature data set according to the feature parameters; The classification and recognition unit is used to evaluate the alternative machine learning algorithms, determine the preferred classification model, identify the feature data set to determine the block loss classification, and predict the wellbore instability type according to the block loss classification.

9. A wellbore instability type prediction system, characterized in that: include: Well wall instability type prediction device, flexible vibration plate, automatic feeding bin, visual device, SCARA robot, conveyor belt and 3D scanning device; The automatic feeding bin is used to automatically vibrate the fallen blocks that need to be divided to the flexible vibration plate; The flexible vibration plate is used to separate the fallen pieces by vibration; The visual system is used to determine the position information of each fallen block and transmit the position information to the SCARA robot; The SCARA robot is used to grab the fallen block from the well wall according to the position information and place it on the conveyor belt, so that the conveyor belt drives the fallen block to move; The three-dimensional scanning device is used to obtain the original point cloud data of the well wall fallen block when the well wall fallen block moves to the preset work station in the conveyor belt; The wellbore instability type prediction device is used to execute a wellbore instability type prediction method as described in any one of claims 1-7.

10. An electronic device, characterized in that: The electronic device comprises: At least one processor; and a memory in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for predicting the type of wellbore instability according to any one of claims 1 to 7.

11. 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 method for predicting the type of wellbore instability according to any one of claims 1 to 7 when executed.

12. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method for predicting the type of wellbore instability according to any one of claims 1 to 7 is implemented.

Images