Sound wave far detection imaging logging geological interpretation conclusion generation method and system

Abstract

The invention belongs to the technical field of oil exploration and development, and relates to a sound wave far detection imaging logging geological interpretation conclusion generation method and system. By mining the relation between a label matrix and a generated text, optimizing the characteristics of the label matrix, and adopting characteristic fusion, the intersection of image and text characteristics is realized, and the visual semantic information fusion is enhanced; the accuracy of the text on image description is improved, the acoustic remote detection imaging logging geological interpretation conclusion is quickly generated, the diversity of the conclusion is reduced, the interpretation coincidence rate is improved, and the interpretation analysis capability of non-acoustic logging professionals on acoustic remote detection imaging is enhanced.

Description

Technical Field

[0001] The invention belongs to the field of geophysical logging technology, relates to acoustic wave remote detection imaging logging technology, and specifically relates to a method and system for generating geological interpretation conclusions of acoustic wave remote detection imaging logging. Background Art

[0002] As an exploration tool that can see far through the well, acoustic wave remote detection imaging logging is widely used in the detection and discovery of concealed oil and gas reservoirs, wellside fractures and caves, and other small geological anomalies. It has shown unique application value in deep-sea tight oil and gas, shale oil and gas, and fracture-cavity oil and gas reservoirs. However, the current challenge of the multi-solution of the geological interpretation of acoustic wave remote detection imaging is that professionals are required to interpret and analyze the acoustic wave remote detection imaging, which limits the deepening and promotion of the application of this technology. To this end, people have built a "forward modeling-imaging-interpretation conclusion" library to compare the acoustic wave remote detection imaging results with the typical imaging in the built library to achieve the geological interpretation of acoustic wave remote detection imaging logging, but there are still technical obstacles such as the cumbersome and time-consuming interpretation process and the low interpretation conformity rate.

[0003] In recent years, the rapid development of artificial intelligence has provided a new research idea for image feature recognition. By learning the feature relationship between images and text reports through neural networks, a relatively accurate text explanation is automatically generated for a given image, providing assistance to non-professionals and reducing the multiple interpretation conclusions. It shows its technical superiority in the generation of medical image report diagnosis solutions. The text generation method is generally composed of a convolutional neural network and a natural language processing network. First, a convolutional neural network (hereinafter referred to as: CNN) or its variants is used to extract features from the input image. CNN is good at capturing low-level local features such as edges and textures in the image, and gradually extracting higher-level semantic information. In order to better capture the complex geological features in the image, the existing method will also introduce a multi-scale feature extraction module to ensure the balanced processing of details and overall information. In terms of text generation, early natural language processing networks are usually LSTM, BiLSTM, GRU, etc., and features are extracted through attention mechanisms. With the development of Transformer, Transformer and related deformation models are used as text feature extraction models and report generation architectures. This basic structure is often more suitable for traditional natural image short sentence description tasks. However, in acoustic wave long-range detection imaging data, normal samples (i.e. samples showing normal geological characteristics) account for a larger proportion of the total samples, while the number of abnormal samples (i.e. samples reflecting abnormal geological characteristics) is relatively small. This data imbalance problem makes the model pay more attention to conventional geological structural characteristics and ignore abnormal characteristics such as cracks and caves, resulting in the generated interpretation conclusions being more biased towards the description of non-feature areas, which fails to reach a geological interpretation conclusion that conforms to reality. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method and system for generating geological interpretation conclusions of acoustic wave long-range detection imaging well logging, mines the relationship between the label matrix and the generated text, optimizes the label matrix features, and adopts feature fusion to realize the intersection of image and text features, strengthens the fusion of visual semantic information, improves the accuracy of text description of the image, quickly generates geological interpretation conclusions of acoustic wave long-range detection imaging well logging, reduces the diversity of conclusions, improves the interpretation conformity rate, and thus improves the interpretation and analysis ability of non-acoustic logging professionals on acoustic wave long-range detection imaging.

[0005] In a first aspect, the present invention provides a method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging, the steps of which are as follows: Data acquisition step: acquiring acoustic wave remote detection imaging images and text sequences; Image feature extraction step: extract local features of the acoustic wave remote detection imaging image, and perform average pooling on the local features to extract global features; Label classification step: classify the local features to obtain label vectors; One-time encoding step: Encode the global features to obtain the hidden vector; Label feature fusion step: construct a label matrix, fuse the label matrix with the hidden vector and label vector to generate label matrix features; Text indexing step: index the text sequence through word vectors to obtain text features; Vector feature fusion step: fuse the text feature with the hidden vector to generate a fusion vector, and add the text feature and the fusion vector to obtain the text vector; Explanation conclusion generation step: decode the label matrix features and text vectors to generate a report and obtain the explanation conclusion; Secondary coding step: Encode text features to obtain enhanced text features; Text feature extraction step: pooling the enhanced text features to obtain a global text feature vector; Loss function design steps: minimize the text loss function by maximizing the similarity between the generated report and the target report by minimizing the cross entropy loss function, use the similarity between the label vector and the true vector as the label loss function, and use the cosine similarity between the label matrix feature and the global text feature vector as the text-label loss function; Optimization step: The interpretation conclusion is optimized by minimizing the total loss function formed by the text loss function, the label loss function and the text-label loss function to generate the final interpretation conclusion.

[0006] In some embodiments, in the image feature extraction step, the acoustic wave long-range detection imaging image is input into an image feature extractor, the features output by the last convolutional layer of the image feature extractor are extracted as local features, the local features are input into a classifier for average pooling, and all features of each row are concatenated into a sequence as a global feature.

[0007] In some embodiments, in the label classification step, the local features are input into a classifier for classification to generate a label vector.

[0008] In some embodiments, in the label feature fusion step, the specific method of fusing the label matrix with the hidden vector and the label vector to generate the label matrix feature is: Calculate the similarity between the label matrix and the hidden vector It is expressed as:

[0009] In the formula, is the label matrix, each row vector of the label matrix , indicating the category The text feature vector of is the number of labels, is the hidden vector, is the hidden vector The transpose of is the dimension of the label matrix; Based on similarity Calculate the attention score with the label vector It is expressed as:

[0010] In the formula, is the normalized exponential function, is the label vector; Using Attention Score , normalization and residual connection fusion label features to obtain the label matrix feature representation as follows:

[0011]

[0012] In the formula, is the label matrix feature, is the normalization function, is the input variable, is the mean, is the standard deviation.

[0013] In some embodiments, in the text feature extraction step, a method for pooling the enhanced text features to obtain a global text feature vector is: pooling the enhanced text features, and taking the first vector of the enhanced text features as the global text feature vector.

[0014] In some embodiments, in the loss function design step, the minimized text loss function is expressed as:

[0015]

[0016] In the formula, To minimize the text loss function, The number of words to generate a report for, For the target report generated The probability distribution of words, for The target text of the moment, for The text of the moment, is the label matrix, This is an image of acoustic wave long-range detection.

[0017] In some embodiments, in the loss function design step, the label loss function is expressed as:

[0018] In the formula, is the label loss function, Predicted for an image Category labels, For Real The label of the category, is the number of labels.

[0019] In some embodiments, in the loss function design step, the text-label loss function is:

[0020]

[0021] In the formula, is the text-label loss function, is the global text feature vector, is the label matrix feature, For Real The label of the category, , indicating the category The text feature vector of is the dimension of the label matrix.

[0022] In some embodiments, in the optimization step, the total loss function is expressed as:

[0023] In the formula, is the total loss function, To minimize the text loss function, is the label loss function, is the text-label loss function, is the weighting factor of the label loss function, is the weighting factor of the text-label loss function.

[0024] In a second aspect, the present invention provides a system for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging, which is used to implement the method for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging described in the first aspect of the present invention, comprising: A data acquisition module, used to acquire acoustic wave remote detection imaging images and text sequences; Image feature extractor, extracting local features of acoustic wave remote detection imaging images; Classifier, average pooling local features to extract global features, and classify local features to obtain label vectors; Transformer neural network, including: Encoder: encodes the global features to obtain hidden vectors, indexes the text sequence through word vectors to obtain text features, encodes the text features to obtain enhanced text features, and pools the enhanced text features to obtain the global text feature vector; Decoder, built-in: The label attention module constructs a label matrix and fuses the label matrix with the hidden vector and label vector to generate label matrix features; The feature fusion module fuses the text features with the hidden vector to generate a fusion vector, and adds the text features to the fusion vector to obtain the text vector; The interpretation conclusion generation module decodes the label matrix features and text vectors to generate a report and obtain the interpretation conclusion; The loss function design module uses the similarity between the generated report and the target report as the minimized text loss function by minimizing the cross entropy loss function, the similarity between the label vector and the true vector as the label loss function, and the cosine similarity between the label matrix feature and the global text feature vector as the text-label loss function; The optimization module optimizes the explanation conclusion by minimizing the total loss function formed by the text loss function, the label loss function and the text-label loss function to generate the final explanation conclusion.

[0025] Compared with the prior art, the advantages and positive effects of the present invention are: (1) The method and system for generating geological interpretation conclusions of acoustic long-range imaging logging provided by the present invention optimize the features of the label matrix by mining the relationship between the label matrix and the generated text, and at the same time adopt feature fusion to realize the intersection of image and text features, strengthen the fusion of visual semantic information, and improve the accuracy of text description of images. By integrating the encoding, fusion and optimization mechanism of image features and text features, the geological interpretation conclusions of acoustic long-range imaging logging are automatically and quickly generated, the diversity of conclusions is reduced, the interpretation conformity rate is improved, and the interpretation and analysis ability of non-acoustic logging professionals on acoustic long-range imaging is enhanced.

[0026] (2) The method and system for generating geological interpretation conclusions of acoustic long-range imaging logging provided by the present invention optimize the label matrix by designing a text-label loss function, and reduce the deviation in label matrix learning by constraining feature consistency and accuracy, thereby further improving the accuracy of text description of images. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flowchart of the method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to an embodiment of the present invention; Figure 2 It is a structural block diagram of the system for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to an embodiment of the present invention; Figure 3 A schematic diagram of the structure of the classifier according to an embodiment of the present invention; Figure 4 Schematic diagram of the structure of the label attention module according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of the feature fusion module according to an embodiment of the present invention; Figure 6 A schematic diagram of a loss value variation curve of the model described in an embodiment of the present invention on a Cave data set; Figure 7 A schematic diagram of a loss value variation curve of the model according to an embodiment of the present invention on the Crack dataset; Figure 8 It is a box plot of the model described in the embodiment of the present invention on the Cave data set; Fig. 9 It is a box plot of the model described in the embodiment of the present invention on the Crack data set; Fig.10 A schematic diagram of two data set images and interpretation conclusions according to an embodiment of the present invention; Fig.11 This is a comparison chart of the qualitative experimental results described in the embodiments of the present invention; Fig.12is the weighting factor of the label loss function described in the embodiment of the present invention Sensitivity diagram of ; Fig.13 is the weighting factor of the text-label loss function described in the embodiment of the present invention Sensitivity diagram of ; Fig.14 This is a diagram showing the results of acoustic wave remote detection processing of the full-depth MX-1 well in the Tarim Oilfield according to an embodiment of the present invention; Fig.15 This is a simulation diagram of acid fracturing fracture of the MX-1 well in an embodiment of the present invention; Fig.16 This is a graph showing the acid fracturing construction of the MX-1 well in the embodiment of the present invention.

[0028] In the figure, 1. Data acquisition module, 2. Image feature extractor, 3. Classifier, 4. Transformer neural network, 41. Encoder, 42. Decoder, 421. Label attention module, 422. Feature fusion module, 423. Explanation conclusion generation module. 424. Loss function design module, 425. Optimization module. DETAILED DESCRIPTION

[0029] The present invention will be described in detail below by way of exemplary embodiments in conjunction with the accompanying drawings. However, it should be understood that, without further description, elements, structures and features in one embodiment may also be beneficially combined in other embodiments.

[0030] Acoustic wave long-range detection imaging logging technology is a key technology used in oil and gas field development and geological structure analysis. It generates imaging images and related geological interpretation information through acoustic wave detection data. However, existing geological interpretation methods mostly rely on artificial experience or simple statistical models, which makes it difficult to fully utilize the complex information in the imaging data, resulting in low accuracy and consistency of the interpretation results. The present invention provides a method and system for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging. By extracting local features and global features of acoustic wave long-range detection imaging images, label classification and feature encoding, combined with text features obtained by indexing from text sequences, feature fusion strategies and optimization designs are used to generate high-quality geological interpretation reports. The core is to use deep learning methods (such as label classification, encoding and decoding architecture) and optimization designs based on loss functions to ensure that the generated interpretation conclusions are close to the real target in content. The following is a detailed description of the above-mentioned method and system for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging in conjunction with the accompanying drawings.

[0031] See also Figure 1 The first embodiment of the present invention provides a method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging, the steps of which are: S1. Data acquisition step: Acquire acoustic wave remote detection imaging images and text sequences.

[0032] Specifically, the text sequence is , , express The text of the moment, Represents the vocabulary consisting of all characters in the text sequence, Indicates the number of characters in the text sequence.

[0033] S2, image feature extraction step: extract local features of the acoustic wave long-range detection imaging image, and perform average pooling on the local features to extract global features.

[0034] Specifically, in some embodiments, the acoustic wave remote detection imaging image is input into an image feature extractor (e.g., a deep convolutional neural network) to extract the features output by the last convolutional layer of the image feature extractor. As local features, , and are the height, width and channels of the image respectively.

[0035] Input the local features into the classifier for average pooling, and concatenate all the features of each row into a sequence as the global feature ,in, , the process is as follows:

[0036] in, Represented as an image feature extractor, For acoustic wave remote detection imaging images, is the number of image blocks, It is Image block features.

[0037] Average pooling is used to obtain global features and improve the stability and expressiveness of features.

[0038] S3, label classification step: perform label classification on local features to obtain label vectors.

[0039] Specifically, in some embodiments, the local features Input into a classifier (e.g., a multi-classification neural network) to classify the input acoustic wave remote detection image.

[0040] Specifically, in some embodiments, the following Figure 3 The classifier structure shown performs label classification on local features. The classifier receives the local features extracted in step S1. , through the pooling layer and convolution layer, the local features It is converted into an image data with 2048 features and 1×1 dimension, and then passes through a fully connected layer with 1024 neurons. In order to reduce fitting, the Dropout technology is applied after the fully connected layer. Finally, it passes through an output layer with 14 neurons, each neuron corresponds to a classification label. The output layer uses Activation function, get the final classification result, generate the label vector of the image , is the number of image labels. The process is as follows:

[0041] in, and It is the trainable weight of the fully connected layer in the classifier, which is used to linearly transform the pooled image features and gradually extract deeper feature information. Represents local features An averaging operation is performed to reduce the dimension of image features while retaining the main feature information. It is an activation function, which represents the corrected linear unit and introduces nonlinear features. It is the activation function of the output layer, and the probability of each category is obtained.

[0042] S4, one-time encoding step: Encode the global features to obtain the hidden vector.

[0043] Specifically, in some embodiments, a Transformer neural network is used to encode the global features to obtain a hidden vector, and the process is shown in the following formula:

[0044] In the formula, is the hidden vector, is a standard Transformer neural network encoder, is the number of image blocks, It is Image block features, It is the enhanced Image block features.

[0045] S5, label feature fusion step: construct a label matrix, and fuse the label matrix with the hidden vector and the label vector to generate label matrix features.

[0046] In order to solve the data imbalance problem, the label matrix is ​​introduced to enhance the text features. This can be achieved through the label attention module of the Transformer neural network. The structure of the label attention module can be found in Figure 4.

[0047] Specifically, the constructed label matrix is ​​expressed as , each row vector of the label matrix , indicating the category The text feature vector of is the number of labels. It should be noted that the label matrix is ​​a trainable label matrix, and the text information of different labels is learned during the training process. The label matrix is ​​randomly set.

[0048] Specifically, the specific method of fusing the label matrix with the hidden vector and the label vector to generate the label matrix features is: Calculate the similarity between the label matrix and the hidden vector It is expressed as:

[0049] In the formula, is the hidden vector The transpose of is the dimension of the label matrix; Based on similarity Calculate the attention score with the label vector It is expressed as:

[0050] In the formula, is the normalized exponential function, is the label vector; Using Attention Score , normalization and residual connection fusion label features to obtain the label matrix feature representation as follows:

[0051]

[0052] In the formula, is the label matrix feature, is the normalization function, is the input variable, is the mean, is the standard deviation.

[0053] Using the label matrix , hidden vector and label vector Calculate attention score jointly , and with the label matrix Weighted calculation, build feature maps for text features of the same label and suppress text features of other labels.

[0054] S6, text indexing step: index the text sequence through word vectors to obtain text features.

[0055] Specifically, in some embodiments, a Transformer neural network encoder is used as a text feature extractor to extract text sequences. Index the word vectors to get text features , the process is shown as follows:

[0056] in, Represents the word vector indexing process.

[0057] S7, vector feature fusion step: perform feature fusion of the text feature and the hidden vector to generate a fusion vector, and add the text feature and the fusion vector to obtain the text vector.

[0058] Specifically, in some embodiments, a Transformer neural network decoder is used for feature fusion, and a feature fusion module is introduced into the Transformer neural network decoder, such as Figure 5 As shown in the figure, the feature fusion module fuses the text features with the hidden vector to generate a fusion vector, which is then added to the text features through the residual layer MLP as the final output of the encoding stage and input into the Transformer neural network decoder. The feature fusion method effectively realizes the information complementarity of different features and enhances the richness of feature representation. The specific process is as follows:

[0059]

[0060] In the formula, for All text features before time, is a hidden vector, representing the enhanced image features, is the fusion vector, is the number of Chinese characters input to the decoder, is a text vector.

[0061] S8, explanation conclusion generation step: decode the label matrix features and text vectors to generate a report and obtain the explanation conclusion.

[0062] S9, secondary encoding step: encode the text features to obtain enhanced text features.

[0063] Specifically, the Transformer neural network encoder is used as a text feature extractor to encode text features to obtain enhanced text features. The process is as follows:

[0064] is the enhanced text feature output by the encoder, To enhance the length of text features, To enhance the text features feature vectors.

[0065] S10, text feature extraction step: pooling the enhanced text features to obtain a global text feature vector.

[0066] Specifically, the first feature vector of the enhanced text feature as the global text feature vector.

[0067] S11. Loss function design steps: The similarity between the generated report and the target report is maximized by minimizing the cross entropy loss function as the minimized text loss function, the similarity between the label vector and the true vector is used as the label loss function, and the cosine similarity between the label matrix feature and the global text feature vector is used as the text-label loss function.

[0068] Specifically, the minimized text loss function is expressed as:

[0069]

[0070] In the formula, To minimize the text loss function, The number of words to generate a report for, For the target report generated The probability distribution of words, for The target text of the moment, for The text of the moment, is the label matrix, This is an image of acoustic wave long-range detection.

[0071] The label loss function is expressed as:

[0072] In the formula, is the label loss function, Predicted for an image Category labels, For Real The label of the category, is the number of labels.

[0073] The text-label loss function is:

[0074]

[0075] In the formula, is the text-label loss function, is the global text feature vector, is the label matrix feature, For Real The label of the category, , indicating the category The text feature vector of is the dimension of the label matrix.

[0076] Improve the matching degree of generated text to target text by minimizing cross entropy loss. Ensure the accuracy of label assignment by optimizing label loss function. Optimize the learning of label matrix by text-label loss to strengthen the correlation between text content and label information.

[0077] S12, optimization step: optimize the interpretation conclusion by minimizing the total loss function formed by the text loss function, the label loss function and the text-label loss function to generate the final interpretation conclusion. The total loss function comprehensively optimizes the text, label and text-label relationship to improve the quality of the final interpretation conclusion.

[0078] Specifically, the total loss function is expressed as:

[0079] In the formula, is the total loss function, To minimize the text loss function, is the label loss function, is the text-label loss function, is the weighting factor of the label loss function, is the weighting factor of the text-label loss function.

[0080] In the method for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging of the present invention, the relationship between the label matrix and the generated text is mined by a model, the label matrix features are optimized, and feature fusion is adopted at the same time to realize the intersection of image and text features, strengthen the fusion of visual semantic information, and improve the accuracy of text description of the image. By integrating the encoding, fusion and optimization mechanism of image features and text features, the geological interpretation conclusions of acoustic wave long-range detection imaging logging are automatically and quickly generated. The interpretation conclusions finally generated can simultaneously meet the optimization goals of text, label and text-label relationship, reduce the diversity of conclusions, and have a high interpretation compliance rate.

[0081] The second embodiment of the present invention provides a system for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging, which is used to implement the method for generating geological interpretation conclusions of acoustic wave long-range detection imaging logging described in the first aspect of the present invention, comprising: Data acquisition module 1, used to acquire acoustic wave remote detection imaging images and text sequences; Image feature extractor 2, extracting local features of acoustic wave remote detection imaging images; Classifier 3, performs average pooling on local features to extract global features, and performs label classification on local features to obtain label vectors; Transformer neural network 4, including: The encoder 41 encodes the global feature to obtain a hidden vector, indexes the text sequence through the word vector to obtain text features, encodes the text features to obtain enhanced text features, and pools the enhanced text features to obtain a global text feature vector; Decoder 42, comprising: The label attention module 421 constructs a label matrix, and performs feature fusion of the label matrix with the hidden vector and the label vector to generate label matrix features; A feature fusion module 422 performs feature fusion on the text feature and the hidden vector to generate a fusion vector, and adds the text feature and the fusion vector to obtain a text vector; The interpretation conclusion generation module 423 decodes the label matrix features and the text vector to generate a report and obtain the interpretation conclusion; The loss function design module 424 uses the similarity between the generated report and the target report maximized by minimizing the cross entropy loss function as the minimized text loss function, uses the similarity between the label vector and the true vector as the label loss function, and uses the cosine similarity between the label matrix feature and the global text feature vector as the text-label loss function; The optimization module 425 optimizes the interpretation conclusion by minimizing the total loss function formed by the text loss function, the label loss function and the text-label loss function to generate the final interpretation conclusion.

[0082] In the above-mentioned acoustic wave long-range detection imaging logging geological interpretation conclusion generation system of the present invention, an image feature extractor, a classifier, an encoder and a decoder of a Transformer neural network form a model, and the acoustic wave long-range detection imaging logging geological interpretation conclusion generation is realized through the model. The model uses a total loss function formed by minimizing a text loss function, a label loss function and a text-label loss function as a loss function. The model mines the relationship between the label matrix and the generated text, optimizes the label matrix features, and adopts feature fusion to achieve the intersection of image and text features, strengthen the fusion of visual semantic information, and improve the accuracy of the text description of the image. Through the encoding, fusion and optimization mechanism of comprehensive image features and text features, the acoustic wave long-range detection imaging logging geological interpretation conclusion is automatically and quickly generated. The interpretation conclusion finally generated can simultaneously meet the optimization goals of text, label and text-label relationship, reduce the diversity of conclusions, and have a high interpretation compliance rate.

[0083] The method for generating the geological interpretation conclusion of the acoustic wave remote detection imaging logging and the training method of the system described in the above embodiment of the present invention are: The experiment was conducted based on the Ubuntu 20.04 operating system, with a graphics card configuration of 12GB, NVIDIA GeForce RTX3060 GPU, and a parallel computing architecture of CUDA11.0. Use PyCharm 2021 development tools, python3.7 programming language, and Pytorch1.7.1 training framework. The experiment uses a pre-trained ResNet101 neural network to extract long-range detection image features, and then inputs the generated 512-dimensional image features into the fully connected layer to obtain a 14-dimensional label vector. Set the number of image feature input layers of the Transformer neural network to 3, the multi-head attention head to 8, the hidden dimension to 512, the input label matrix size to 14×512, and set the balance parameters and is 1 and 0.5. In the experiments of training Cave and Crack datasets, the Adam optimizer, 1e-4 model learning rate, and 5e-5 visual extraction learning rate are used for model optimization. The model loss changes on the two datasets are as follows: Figure 6 , 7 As shown in the figure, the Cave dataset was trained for 50 rounds, and the loss curve became stable after 10 rounds of iteration. The Crack dataset was trained for 30 rounds, and the loss curve became stable after 10 rounds of iteration, and the model achieved the best effect. Figure 8 , Fig. 9 The box plot in the figure shows the distribution of evaluation indicators during the model training process. Based on the above analysis, it can be concluded that the acoustic wave long-range detection imaging logging geological interpretation conclusion generation method proposed in the present invention and the Transformer neural network in the system show a stable advancement trend during the iteration process.

[0084] Figure 6 is the loss value change curve of the model on Cave. Figure 5 As shown in the figure, the Cave dataset is trained for 50 rounds, and the loss curve becomes stable after 10 rounds of iterations.

[0085] Figure 7 is the loss value change curve of the model on Crack. Figure 6 As shown in the figure, the number of training rounds for the Crack dataset is 30. After 10 rounds of iterations, the loss curve tends to be stable and the model achieves the best effect.

[0086] Figure 8 This is a box plot of the model's evaluation indicators on the Cave dataset. Fig. 9 The box plot of the evaluation index of the model on the Crack dataset shows the distribution of the evaluation index during the model training process. It can be seen that the model of the present invention presents a stable advancement trend during the iteration process.

[0087] In order to verify the effectiveness of the method and system for generating geological interpretation conclusions of acoustic wave remote detection imaging logging described in the above embodiment of the present invention, the following specific embodiments are used for description.

[0088] Embodiment: A finite difference numerical simulation method is used. According to the characteristics of the fracture-cave carbonate reservoir in the Tarim Oilfield, in order to simplify the model complexity, the fracture-cave aggregate model is split into fracture and cave models, including a cave model and a crack model.

[0089] According to the parameter settings in Table 1, finite difference numerical simulation was carried out, and acoustic remote detection migration imaging was performed on it. The size of the geological model was set to 30 m×20 m to capture and simulate the geological structural characteristics in a larger range. In the cave model, 25,200 cave models were obtained by changing the size and number of caves, the distance from the well axis, and the type of cave filling. The fracture model considered the inclination, width, position and number of fractures, simulated the morphological changes of fractures under different geological conditions, and obtained 2,640 simulation models. Table 1 shows the design details of the two models, and the formation elastic parameter settings in the numerical simulation are shown in Table 2. After the geological model was constructed, the borehole acoustic field numerical simulation was carried out to obtain the dipole array waveform data. The corresponding model acoustic remote detection result map was obtained by using the migration imaging technology. Each geological model corresponds to an interpretation conclusion, thereby constructing the cave (Cave) data set and the fracture (Crack) data set, which consist of 25,200 and 2,640 remote detection images and corresponding interpretation conclusion reports, respectively. The reports are manually annotated and describe the geological characteristics and parameters (such as cave location, size, shape, etc.) in detail.

[0090] Table 1

[0091] Table 2

[0092] Table 3 shows the information of images, reports and average word count in the dataset. The dataset intuitively shows the features of caves and cracks in the geological structure.

[0093] Table 3

[0094] The experiment split the data into training set, validation set and test set according to 70%-10%-20%, as shown in Table 4.

[0095] Table 4

[0096] Fig.10 The following table shows examples of imaging images and reports in the dataset, from left to right, model type, remote detection imaging image, and interpretation conclusion. For example, the first row shows an imaging image of a single cave, and the corresponding interpretation is: there is a single cave 10 meters away from the well axis. It should be emphasized that in order to achieve the final on-site rapid interpretation and analysis, more extensive and more numerical calculation results that conform to the geological model are still needed.

[0097] Fig.11 The following is a comparison of the qualitative experimental results of the method of the present invention and the system. Fig.11As shown in the figure, from top to bottom, the simulated image of a single cave near the well, the simulated image of a single through-well fracture, and the simulated image of a cave group near the well are respectively displayed. The heat map is output by the last hidden layer of the image feature extraction module, reflecting the model's focus area on the input image and its feature capture ability. The left column is the original image, the middle column shows the R2Gen method heat map, and the right column shows the heat map obtained by the method and system of the present invention. By comparing the heat map results, it is found that the text generated by the method and system of the present invention can accurately capture and describe the location and characteristics of the fractures and caves near the well. For example, in the first row, the high-heat area of ​​the method and system of the present invention fully represents the location and size information of a single cave, while the heat map generated by the R2Gen method fails to fully cover the cave area. In the second row, the method and system of the present invention relatively completely identify the through-well fracture, while the R2Gen method makes an incorrect judgment on the irrelevant area outside the fracture. In the heat map of the third row, the method and system of the present invention generate multiple high-heat areas within the range of 5-15 meters from the well axis, which can clearly represent the cave group within this range, while the heat map generated by the R2Gen method has insufficient coverage and fails to accurately display the location and range of the cave group. Through qualitative analysis of the heat map results, the method and system of the present invention outperform the R2Gen method in capturing the details of geological features, accurately describing the locations and characteristics of fractures and caves, and maintaining consistency in a variety of complex situations.

[0098] Table 5 shows the comparative experimental results of different models on two data sets. As shown in Table 5, among these comparative methods, the traditional basic image captioning method ATT2IN directly applied to the generation of interpretation conclusions of remote detection images failed to achieve satisfactory results, while the R2Gen and R2GenCMN models using memory networks only considered using memory records to generate information. Although good results were achieved, they did not solve the data bias problem that conventional geological structure data accounted for a large proportion while abnormal geological structure feature data such as caves and cracks accounted for a small proportion. The CMCL model gradually improves the model performance by estimating the instance training difficulty and model capability. Prior Guided Trans introduces more complex prior knowledge. Compared with the above work, the model structure proposed in the present invention is simpler and achieves better or the same results. In general, the method and system of the present invention are superior to most comparative methods in all indicators.

[0099] Table 5

[0100] Table 6 and Table 7 are the ablation experiment results of the model proposed by the method and system of the present invention on the Cave dataset and the Crack dataset, respectively. Base is a Transformer-based encoder-decoder model, experiment (a) is the result of adding a feature fusion module on the basis of the base model, experiment (b) adds a label feature matrix, experiment (c) optimizes the label matrix through text-label loss, and experiment (d) is the model proposed by the method and system of the present invention. Comparing the results of base and (a) of the two dataset experiments, it is shown that the Transformer network with added feature fusion has significant improvements in most indicators. This is because the traditional Transformer model only fuses image features in the multi-head attention module, while the far-detection image contains a large amount of complex geological information. The encoder of the Transformer tends to capture the coarse-grained features and dependencies of the image, ignoring the fine-grained geological anomaly information. Therefore, a feature fusion module is added to integrate the visual and text features through the label matrix, so that the module pays attention to important visual features and suppresses irrelevant information, thereby providing a richer feature representation for the model. In the experiment (b), the label matrix is ​​first initialized using a normal distribution, and then the features of different label reports are learned in subsequent training. Compared with the results of the base, each evaluation index has been significantly improved, proving that the label matrix trained by the experiment can capture and memorize the various features of geological anomalies, provide richer feature representations in the decoding process, and improve the quality and reliability of the generated text. Experiment (c) further optimizes the label matrix by adding text-label loss, and the experimental indicators are significantly improved compared with experiment (b). Since geological anomalies have complex and diverse features, the model may make errors when learning the label matrix by relying solely on traditional loss functions. The text-label loss function can further optimize the label matrix, reduce the deviation in label matrix learning by constraining feature consistency and accuracy, and improve the overall model performance.

[0101] Table 6

[0102] Table 7

[0103] Fig.12 , Fig.13 They are the experimental results of the sensitivity of two corresponding parameters in the model. and The two parameters balance the text loss, label loss, and text-label loss. In order not to affect the performance of the model in generating text reports, the label loss and text-label loss parameters are set smaller than the text loss parameter. The impact of the corresponding parameters on the results is tested on the Cave dataset. Response parameters The sensitivity of Fig.12 As shown on the left. The experimental results show that with Increase, the size of each indicator fluctuates greatly until When it reaches 1, the values ​​of most indicators reach their maximum and tend to be stable. The sensitivity of Fig.13 As shown on the right. When the value increases from 0.1 to 0.5, the values ​​of most evaluation indicators gradually increase to the highest level. When the size of increases to 1 again, the size of the evaluation index gradually decreases. It is a parameter for adjusting the optimization matrix loss. If the parameter value is too small, the matrix cannot accurately reflect the label vector. If the parameter value is too large, the model will introduce too much noise information, resulting in performance degradation. Experimental results show that 1. When the value is 0.5, the best result is achieved.

[0104] Figures 14 to 16 This is a diagram of the analysis results of a field example.

[0105] Fig.14 This is the result map of acoustic wave long-range detection of the MX-1 well in Tarim Oilfield (7650-7740m). From left to right in the figure are the dipole variable density full-wave waveform, the monopole full-wave waveform, the acoustic wave time difference and the corresponding STC correlation map, the horizontal well acoustic wave long-range detection imaging map (up and down and left and right) and the extracted imaging heat map. Figure 1 As shown in the method flow chart, the present invention uses a sliding window to segment the original acoustic wave remote detection imaging image. The starting position of the window is set at the top of the image (usually the starting depth), the window size is 2 meters, the step length is 1 meter, and it slides gradually from top to bottom. Each image slice cut out is used as an independent image and analyzed by the method and system of the present invention. The heat map generated in the image feature encoding stage is re-spliced ​​according to its position in the original image to ensure that the geological features within the entire depth range are fully presented. From the results of the thermal map imaging, this layer section has developed well fractures (7660-7680m), high-angle fractures beside the well (7680-7710m) and caves beside the well (7680-7730m).

[0106] Fig.15 This is the acid fracturing fracture simulation of Well MX-1. It can be seen that the layer with the longest fracture extension corresponds to the layer with developed fractures near the well.

[0107] Fig.16 This is the acid fracturing construction curve of Well MX-1. It can be found that there are many points where the pressure drops rapidly during the fracturing process. Analysis shows that this is the response of communicating natural fractures.

[0108] The above embodiments are used to explain the present invention rather than to limit the present invention. Any modification and change made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging, characterized in that: The steps are: Data acquisition step: acquiring acoustic wave remote detection imaging images and text sequences; Image feature extraction step: extract local features of the acoustic wave remote detection imaging image, and perform average pooling on the local features to extract global features; Label classification step: classify the local features to obtain label vectors; One-time encoding step: Encode the global features to obtain the hidden vector; Label feature fusion step: construct a label matrix, fuse the label matrix with the hidden vector and label vector to generate label matrix features; Text indexing step: index the text sequence through word vectors to obtain text features; Vector feature fusion step: fuse the text feature with the hidden vector to generate a fusion vector, and add the text feature and the fusion vector to obtain the text vector; Explanation conclusion generation step: decode the label matrix features and text vectors to generate a report and obtain the explanation conclusion; Secondary coding step: Encode text features to obtain enhanced text features; Text feature extraction step: pooling the enhanced text features to obtain a global text feature vector; Loss function design steps: minimize the text loss function by maximizing the similarity between the generated report and the target report by minimizing the cross entropy loss function, use the similarity between the label vector and the true vector as the label loss function, and use the cosine similarity between the label matrix feature and the global text feature vector as the text-label loss function; Optimization step: The interpretation conclusion is optimized by minimizing the total loss function formed by the text loss function, the label loss function and the text-label loss function to generate the final interpretation conclusion.

2. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the image feature extraction step, the acoustic wave long-range detection imaging image is input into the image feature extractor, the features output by the last convolutional layer of the image feature extractor are extracted as local features, the local features are input into the classifier for average pooling, and all the features of each row are concatenated into a sequence as the global feature.

3. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the label classification step, the local features are input into the classifier for classification to generate a label vector.

4. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the label feature fusion step, the specific method of fusing the label matrix with the hidden vector and the label vector to generate the label matrix feature is: Calculate the similarity between the label matrix and the hidden vector It is expressed as: In the formula, is the label matrix, each row vector of the label matrix , indicating the category The text feature vector of is the number of labels, is the hidden vector, is the hidden vector The transpose of is the dimension of the label matrix; Based on similarity Calculate the attention score with the label vector It is expressed as: In the formula, is the normalized exponential function, is the label vector; Using Attention Score , normalization and residual connection fusion label features to obtain the label matrix feature representation as follows: In the formula, is the label matrix feature, is the normalization function, is the input variable, is the mean, is the standard deviation.

5. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the text feature extraction step, the method of pooling the enhanced text features to obtain the global text feature vector is: pooling the enhanced text features, and taking the first vector of the enhanced text features as the global text feature vector.

6. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the loss function design step, the minimized text loss function is expressed as: In the formula, To minimize the text loss function, The number of words to generate a report for, For the target report generated The probability distribution of words, for The target text of the moment, for The text of the moment, is the label matrix, This is an image of acoustic wave long-range detection.

7. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the loss function design step, the label loss function is expressed as: In the formula, is the label loss function, Predicted for an image Category labels, For Real The label of the category, is the number of labels.

8. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the loss function design step, the text-label loss function is: In the formula, is the text-label loss function, is the global text feature vector, is the label matrix feature, For Real The label of the category, , indicating the category The text feature vector of is the dimension of the label matrix.

9. The method for generating geological interpretation conclusions of acoustic wave remote detection imaging logging according to claim 1, characterized in that: In the optimization step, the total loss function is expressed as: In the formula, is the total loss function, To minimize the text loss function, is the label loss function, is the text-label loss function, is the weighting factor of the label loss function, is the weighting factor of the text-label loss function.

10. A system for generating geological interpretation conclusions of acoustic wave long-range imaging logging, used to implement the method for generating geological interpretation conclusions of acoustic wave long-range imaging logging as claimed in any one of claims 1 to 9, characterized in that: include: A data acquisition module, used to acquire acoustic wave remote detection imaging images and text sequences; Image feature extractor, extracting local features of acoustic wave remote detection imaging images; Classifier, average pooling local features to extract global features, and classify local features to obtain label vectors; Transformer neural network, including: Encoder: encodes the global features to obtain hidden vectors, indexes the text sequence through word vectors to obtain text features, encodes the text features to obtain enhanced text features, and pools the enhanced text features to obtain the global text feature vector; Decoder, built-in: The label attention module constructs a label matrix and fuses the label matrix with the hidden vector and label vector to generate label matrix features; The feature fusion module fuses the text features with the hidden vector to generate a fusion vector, and adds the text features to the fusion vector to obtain the text vector; The interpretation conclusion generation module decodes the label matrix features and text vectors to generate a report and obtain the interpretation conclusion; The loss function design module uses the similarity between the generated report and the target report as the minimized text loss function by minimizing the cross entropy loss function, the similarity between the label vector and the true vector as the label loss function, and the cosine similarity between the label matrix feature and the global text feature vector as the text-label loss function; The optimization module optimizes the explanation conclusion by minimizing the total loss function formed by the text loss function, the label loss function and the text-label loss function to generate the final explanation conclusion.

Images