Distributed intelligent logging system and logging method

Abstract

The invention discloses a distributed intelligent well logging system and a well logging method, and relates to the technical field of intelligent well logging. The distributed intelligent well logging system comprises an intelligent well logging module, a ground intelligent control module and a communication module; the number of the intelligent logging modules is N, and N is an integer larger than 1; the intelligent well logging module is arranged underground and used for collecting, processing and uploading well logging data. The ground intelligent control module is arranged on the ground and is used for receiving and analyzing data uploaded by the intelligent logging module and sending an acquisition scheme to the intelligent logging module; the intelligent well logging module and the ground intelligent control module are in two-way communication through the communication module. The intelligent logging module comprises a data acquisition unit, a first processing unit and a first communication interface unit; the ground intelligent control module comprises a second communication interface unit, an intelligent analysis unit and a control instruction generation unit, underground intelligent preprocessing and ground real-time analysis are achieved through deep fusion of a distributed architecture and an artificial intelligence technology, and an acquisition strategy can be dynamically optimized according to an analysis result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent logging, in particular to a distributed intelligent logging system and a logging method. BACKGROUND

[0002] With the exploration and development of oil and gas fields extending to deep, complex reservoirs and unconventional fields, traditional logging technology is facing severe challenges. Most existing logging systems adopt a centralized architecture, in which downhole instruments are mainly responsible for raw data acquisition and upload all to the ground computer for processing and analysis through a cable. This mode has many inherent defects: first, the high-definition image data collected downhole is huge, while the cable transmission bandwidth is limited, resulting in low data transmission efficiency and significantly prolonging the operation time; second, the ground system cannot monitor the downhole data quality in real time, and only after the data is returned can image blurring, signal distortion and other problems be found, which requires re-measuring downhole, greatly increasing the operation cost and risk; third, data analysis and decision-making completely depend on the experience of ground personnel, the process is cumbersome and lagging, and cannot realize adaptive acquisition and real-time optimization for complex downhole geological conditions.

[0003] In recent years, although some research has tried to add preprocessing functions downhole to compress data, it is mostly limited to simple filtering or compression, lacking intelligent judgment of data value. At the same time, the ground analysis system lacks intelligence, and cannot form a closed-loop feedback from data perception to decision execution, making it difficult to dynamically adjust the acquisition strategy according to real-time geological information, restricting the improvement of exploration efficiency and effect.

[0004] Therefore, a distributed intelligent logging system and a logging method are provided to solve the above problems. SUMMARY

[0005] To solve the above problems, the present application provides a distributed intelligent logging system and a logging method, which realizes intelligent preprocessing downhole, real-time analysis on the ground, and dynamic optimization of acquisition strategy according to the analysis results through the deep integration of distributed architecture and artificial intelligence technology.

[0006] To achieve the above purpose, the present application provides a distributed intelligent logging system, which comprises an intelligent logging module, a ground intelligent control module and a communication module; the intelligent logging module is N, N is an integer greater than 1; The intelligent logging module is arranged downhole for collecting, processing and uploading logging data; The ground intelligent control module is arranged on the ground for receiving and analyzing the data uploaded by the intelligent logging module and sending a new acquisition scheme to the intelligent logging module; The intelligent logging module and the ground intelligent control module communicate bidirectionally through the communication module.

[0007] Preferably, the intelligent logging module comprises a data acquisition unit, a first processing unit and a first communication interface unit; The data acquisition unit is configured to acquire downhole data; the first processing unit is connected with the data acquisition unit and configured to pre-process the acquired downhole data; and the first communication interface unit is connected with the first processing unit and the communication module and configured to interact with the ground intelligent control module through the communication module.

[0008] Preferably, the ground intelligent control module comprises a second communication interface unit, an intelligent analysis unit and a control instruction generation unit; The second communication interface unit is configured to interact with the first communication interface unit through the communication module; the intelligent analysis unit is connected with the second communication interface unit and configured to analyze and process the received downhole data and generate a new logging data acquisition scheme based on the analysis result; and the control instruction generation unit is connected with the intelligent analysis unit and the second communication interface unit and configured to generate a corresponding control instruction according to the new logging data acquisition scheme and issue the control instruction to the intelligent logging module through the second communication interface unit and the communication module.

[0009] Preferably, the first processing unit specifically works as follows: receiving the downhole data acquired by the data acquisition unit; performing data cleaning, format standardization and compression processing on the downhole data; adding time stamp, depth identification and module identification metadata to the processed data; selecting the priority and transmission opportunity of the data to be uploaded according to the preset transmission strategy and downhole communication condition; sending the encapsulated data to the communication module through the first communication interface unit.

[0010] Preferably, the intelligent analysis unit comprises a data receiving and preprocessing subunit, an image quality intelligent evaluation subunit, a geological feature intelligent recognition subunit, a multi-source data fusion analysis subunit and a scheme decision and output subunit. The data receiving and preprocessing subunit is configured to receive the data encapsulated by the intelligent logging module, decode, format convert and time-depth align the data, and extract image data and environmental parameter data. The image quality intelligent evaluation subunit is connected with the data receiving and preprocessing subunit and configured to perform multi-dimensional quality analysis on the image data by using a multi-task deep convolutional neural network, generate quantitative scores of clarity, contrast and signal-to-noise ratio, judge whether the scores are lower than a preset threshold, and classify the image data into high-quality image data and low-quality image data according to the judgment result. The geological feature intelligent identification subunit is connected with the image quality intelligent evaluation subunit, and is configured to perform pixel-level semantic segmentation on high-quality image data output by the image quality intelligent evaluation subunit by using a hybrid architecture of a Transformer-UNet, identify geological features such as cracks, holes and bedding, and calculate reservoir parameters such as crack density, porosity and permeability; The multi-source data fusion analysis subunit is connected with the image quality intelligent evaluation subunit and the geological feature intelligent identification subunit, and is configured to perform spatio-temporal registration and fusion analysis based on image recognition results and environmental parameter data by using a Bayesian network algorithm, and generate a comprehensive evaluation report of a downhole geological environment in combination with a geological model and historical data. The scheme decision and output subunit is connected with the multi-source data fusion analysis subunit, and is configured to generate a new logging data acquisition scheme according to the comprehensive evaluation report, and send the new acquisition scheme to the control instruction generation unit.

[0011] Preferably, the multi-task deep convolutional neural network takes a ResNet network as a backbone network, includes a global average pooling layer and a fully connected regression branch, the global average pooling layer is connected with the backbone network and is configured to convert a feature map into a feature vector, and the fully connected regression branch is connected with the global average pooling layer and is configured to calculate and output three quantitative scores of image clarity, contrast and signal-to-noise ratio.

[0012] Preferably, the scheme decision and output subunit further includes a human-computer interaction interface, which is configured to present the new logging data acquisition scheme to an operator in a visualized form, and receive a modification confirmation instruction issued by the operator based on the comprehensive evaluation report.

[0013] Preferably, the communication module is a single-core armored cable, and the bidirectional data transmission and power supply between the downhole and the ground are realized by cable telemetry technology.

[0014] A distributed intelligent logging method, comprising the following steps: S1: collecting downhole data and performing preprocessing; S2: analyzing and processing the downhole data in S1, and generating a new logging data acquisition scheme based on the analysis results; and sending corresponding control instructions to the intelligent logging module according to the new logging data acquisition scheme; S3: the intelligent logging module executes the control instructions in S2.

[0015] Preferably, S2 specifically comprises: S21: decoding, format conversion and time-depth alignment processing of data, and extraction of image data and environmental parameter data; S22: Multi-dimensional quality analysis of image data is performed using a multi-task deep convolutional neural network to generate three quantitative scores of definition, contrast, and signal-to-noise ratio, and whether the scores are lower than a preset threshold is determined, and according to the determination result, the image data is divided into high-quality image data and low-quality image data; S23: Pixel-level semantic segmentation is performed on the high-quality image data in S22 using a Transformer-UNet hybrid architecture to identify cracks, holes, and bedding geological features, and reservoir parameters such as crack density, porosity, and permeability are calculated; S24: Based on the image recognition result and environmental parameter data, a Bayesian network algorithm is used for spatio-temporal registration and fusion analysis, and combined with a geological model and historical data, a comprehensive evaluation report of the downhole geological environment is generated; S25: A new logging data acquisition scheme is generated according to the comprehensive evaluation report; S26: Control instructions are generated according to the new logging data acquisition scheme.

[0016] Therefore, the present application adopts the above-mentioned distributed intelligent logging system and logging method, which has the following beneficial effects: (1) The present application cleans, compresses and intelligently selects the original data through the downhole first processing unit, effectively reduces the redundant data transmission, relieves the cable bandwidth pressure, and shortens the single logging cycle. The image quality intelligent evaluation subunit can identify low-quality data in real time and trigger an early warning, avoiding repeated downhole operations due to unqualified data, greatly saving non-production time and comprehensive cost.

[0017] (2) The present application constructs a real-time intelligent closed loop of downhole acquisition-ground analysis-decision feedback. The ground intelligent analysis unit dynamically generates an optimized acquisition scheme based on multi-source data fusion analysis and geological feature recognition, and sends control instructions to the downhole for execution. This makes data acquisition change from a fixed program to target-driven, which can adaptively adjust the acquisition parameters according to the encountered geological targets (such as cracks and holes), maximize the information value of unit data, and improve the pertinence and effectiveness of exploration.

[0018] (3) The present application uses a multi-task deep convolutional neural network to evaluate image quality, objectively quantifies definition, contrast and signal-to-noise ratio, and ensures the reliability of the subsequent analysis data source. The Transformer-UNet hybrid architecture is used to realize pixel-level geological feature segmentation, accurately identify complex structures such as cracks and holes, and automatically calculate key parameters such as crack density and porosity, reducing the subjective error of human interpretation and improving the accuracy and efficiency of reservoir evaluation.

[0019] (4) The human-computer interaction interface of the decision-making and output sub-unit of the present invention visualizes the AI ​​analysis results and optimization schemes to the operators, allowing expert experience to intervene in the review and correction, realizing the complementary advantages of artificial intelligence and human experts, and ensuring the scientificity and reliability of the final decision.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 This is a structural diagram of a distributed intelligent logging system according to the present invention; Figure 2 This is a flowchart illustrating a distributed intelligent logging method according to the present invention. Detailed Implementation

[0022] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0024] The terms "comprising" or "including" as used in this invention mean that the element preceding the term encompasses the element listed after the term, and do not exclude the possibility of encompassing other elements. Terms such as "inner," "outer," "upper," and "lower" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. In this invention, unless otherwise explicitly specified and limited, the term "attached" and similar terms should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can refer to a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication of two elements or the interaction relationship between two elements. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0025] Example A distributed intelligent logging system, such as Figure 1 As shown, it includes an intelligent logging module and a surface intelligent control module; there are N intelligent logging modules, where N is an integer greater than 1; The intelligent logging module is installed downhole and is used to collect, process, and upload logging data; Specifically, the intelligent logging module includes a data acquisition unit, a first processing unit, and a first communication interface unit; The data acquisition unit is used to acquire downhole data; the first processing unit, connected to the data acquisition unit, is used to preprocess the acquired downhole data; the first communication interface unit, connected to the first processing unit and the communication module, is used to interact with the ground intelligent control module through the communication module.

[0026] The specific working process of the first processing unit is as follows: Receive downhole data collected by the data acquisition unit; Data cleaning, format standardization, and compression are performed on downhole data; Add timestamps, depth identifiers, and module identifier metadata to the processed data; Based on the preset transmission strategy and downhole communication conditions, select the priority and transmission timing of the data to be uploaded; The encapsulated data is sent to the communication module through the first communication interface unit.

[0027] The ground-based intelligent control module is located on the ground and is used to receive and analyze the data uploaded by the intelligent logging module and send new acquisition plans to the intelligent logging module. Specifically, the ground intelligent control module includes a second communication interface unit, an intelligent analysis unit, and a control command generation unit; The second communication interface unit is used to interact with the first communication interface unit through the communication module; the intelligent analysis unit, connected to the second communication interface unit, is used to analyze and process the received downhole data and generate a new logging data acquisition scheme based on the analysis results; the control command generation unit, connected to the intelligent analysis unit and the second communication interface unit, is used to generate corresponding control commands according to the new logging data acquisition scheme and send them to the intelligent logging module through the second communication interface unit and the communication module.

[0028] The intelligent analysis unit includes a data receiving and preprocessing subunit, an image quality intelligent assessment subunit, a geological feature intelligent identification subunit, a multi-source data fusion analysis subunit, and a scheme decision and output subunit; The data receiving and preprocessing subunit is used to receive the data packaged by the intelligent logging module, decode the data, convert the format and perform time-depth alignment processing, and extract image data and environmental parameter data; The image quality intelligent assessment subunit is connected to the data receiving and preprocessing subunit. It is used to perform multi-dimensional quality analysis on image data using a multi-task deep convolutional neural network, generate three quantitative scores of sharpness, contrast and signal-to-noise ratio, determine whether the score is lower than a preset threshold, and classify the image data into high-quality image data and low-quality image data according to the judgment result. The multi-task deep convolutional neural network uses ResNet as its backbone network, including a global average pooling layer and a fully connected regression branch. The global average pooling layer is connected to the backbone network and is used to convert feature maps into feature vectors. The fully connected regression branch is connected to the global average pooling layer and is used to calculate and output three quantitative scores of image sharpness, contrast and signal-to-noise ratio.

[0029] Specifically, the multi-task deep convolutional neural network uses a ResNet network with the original top-level fully connected classifier removed as a shared backbone, retaining all convolutional layers from the first convolutional layer to the fifth residual block, as well as normalization and activation layers, with an output space size of 7. 7. A high-order feature map with 2048 channels; a parameterless kernel with a size of 7 is then appended to this feature map. A global average pooling layer with a step size of 1 and a stride of 7 achieves a mean of 1 across both width and height dimensions. 1 A fixed-length global feature vector of 2048 is then flattened to 2048 dimensions and sequentially passed through a shared dimensionality reduction module: a random deactivation layer with a deactivation rate of 0.5, a fully connected layer with an input dimension of 2048 and an output dimension of 512, a modified linear unit activation function, and another random deactivation layer with a deactivation rate of 0.5, forming a 512-dimensional shared representation. Finally, three structurally identical but independent fully connected regression branches are derived in parallel. Each branch contains only one linear mapping layer with an input dimension of 512 and an output dimension of 1. Without activation, it directly outputs scalar prediction values, corresponding to the image sharpness score, contrast score, and signal-to-noise ratio score, respectively, achieving end-to-end multi-task joint training.

[0030] The intelligent geological feature recognition subunit is connected to the intelligent image quality assessment subunit. It is used to perform pixel-level semantic segmentation on the high-quality image data output by the intelligent image quality assessment subunit using the Transformer-UNet hybrid architecture, identify geological features such as fractures, pores, and bedding, and calculate reservoir parameters such as fracture density, porosity, and permeability. Specifically, high-quality images are fed into the Transformer-UNet hybrid architecture: a convolutional front end, 3×3 convolutional layers + batch normalization + ReLU, which maps the three-channel image into a 64-channel feature map with invariant spatial dimensions. The convolutional path uses two consecutive 3×3 convolutions (downsampling with a stride of 2) to extract local textures. The feature map is sliced ​​into 8×8 non-overlapping windows, flattened into 64-dimensional tokens, and then processed by a 4-head self-attention + feedforward network to learn the global context. The outputs of the two paths are concatenated along the channel dimension and then fused using a 1×1 convolution to obtain the final feature of this layer. With each layer, the number of channels doubles (64→128→256→512→1024), and the spatial size is halved (H / 2, H / 4, H / 8, H / 16, H / 32), forming a five-level multi-scale feature.

[0031] Symmetrical four-level upsampling is adopted. Each layer first uses 2×2 deconvolution to double the spatial size and halve the channels. Then, it is concatenated with the features at the same level as the encoder in the channel dimension. Then, two 3×3 convolutions + ReLU are used for refinement. The last layer uses 1×1 convolution to map 64 channels into 4 channels for output, which correspond to the pixel-level probabilities of four categories: background, cracks, holes, and layering.

[0032] A hybrid loss method combining cross-entropy and Dice is used to assign higher weights (5:1) to fractures and pores to address sample imbalance. Data augmentation includes random rotation, brightness variation, and simulated mud spots to improve the model's robustness under varying downhole lighting conditions.

[0033] The multi-source data fusion analysis subunit is connected to the image quality intelligent assessment subunit and the geological feature intelligent recognition subunit. It is used to perform spatiotemporal registration and fusion analysis based on image recognition results and environmental parameter data, and to generate a comprehensive assessment report of the downhole geological environment by combining geological models and historical data. Specifically, a unified coordinate framework based on depth and time is constructed. Data from different sources and at different times (image recognition results, environmental parameters) are precisely aligned to the same downhole location point using their timestamps and depth markers. For example, it ensures that at a point "well depth 1505.5 meters," the features identified by the image, the temperature and pressure data at that point, and the geological model prediction results corresponding to that depth are correctly correlated. This is the foundation for all subsequent fusion analyses.

[0034] Constructing a knowledge network: A network model is built in advance based on geological knowledge and expert experience. Network nodes include environmental parameters (temperature, pressure), image quality indicators, image recognition results, and the final actual geological properties (lithology, fractures, stability, etc.). The connections between nodes represent causal or influencing relationships between them; for example, "high drilling fluid viscosity" leads to "low image clarity"; "low image clarity" increases the "image recognition error rate"; and "actual lithology" and "image clarity" together determine the "image recognition result."

[0035] The spatiotemporally registered data is input into the network. Image quality indicators (e.g., "clarity = poor") are input into the corresponding nodes. Preliminary identification results (e.g., "identified as mudstone") are input into the corresponding nodes. Environmental parameters (e.g., "pressure = 50 MPa") are input into the corresponding nodes. The geological model and historical data for this depth point are retrieved from the geological database as prior probabilities. For example, based on data from neighboring wells, the probability of encountering sandstone at this stratum is 70%, and the probability of encountering mudstone is 30%. A Bayesian inference algorithm is fused to perform probability propagation calculations throughout the network. Its core function is to quantify and manage the uncertainty at each stage. For example, it calculates: "Given poor image quality and prior knowledge indicating that this area is mostly sandstone, what is the probability that the actual lithology is still sandstone, even though the image initially identifies it as mudstone?" Finally, it outputs an optimal estimate that integrates all information, along with a confidence level. Based on the high-confidence result obtained after fusion analysis, which removes most contradictions and information noise, this sub-unit calls the report generation template.

[0036] A comprehensive assessment report of the underground geological environment typically includes: Comprehensive lithology columnar section: Displays the final identified lithology and its confidence level at different depths.

[0037] Geological feature analysis: description of features such as fractures, faults, and cavities, occurrence measurement and reliability assessment.

[0038] Wellbore stability assessment: Based on lithology, fracture development and ground pressure data, predict and warn of wellbore collapse risk.

[0039] Data quality explanation: Attach the image quality assessment results to explain the low confidence of some conclusions in the report (e.g., "The confidence of the lithology judgment is low because the image of segment XX is blurry").

[0040] Decision recommendations: Provide drilling engineers with data-driven recommendations, such as "It is recommended to reduce the drilling rate at a depth of 1508 meters and carefully observe the mud return situation."

[0041] The scheme decision and output subunit is connected to the multi-source data fusion analysis subunit. It is used to generate a new well logging data acquisition scheme based on the comprehensive evaluation report and send the new acquisition scheme to the control command generation unit.

[0042] The scheme decision and output subunit also includes a human-machine interface, which is used to present the new logging data acquisition scheme to the operator in a visual form and receive modification confirmation instructions issued by the operator based on the comprehensive evaluation report.

[0043] The intelligent logging module and the ground intelligent control module communicate bidirectionally through a communication module.

[0044] The communication module is a single-core armored cable, which enables bidirectional data transmission and power supply between the mine and the surface through cable remote transmission technology.

[0045] Example 1 A distributed intelligent logging method, such as Figure 2 As shown, it includes the following steps: S1: Collect downhole data and perform preprocessing; S2: Analyze and process the downhole data in S1, and generate a new logging data acquisition scheme based on the analysis results; generate corresponding control commands according to the new logging data acquisition scheme and send them to the intelligent logging module; S2 specifically includes: S21: Decode the data, convert the format, and perform time-depth alignment to extract image data and environmental parameter data; S22: Use a multi-task deep convolutional neural network to perform multi-dimensional quality analysis on image data, generate three quantitative scores of sharpness, contrast and signal-to-noise ratio, determine whether the score is lower than the preset threshold, and classify the image data into high-quality image data and low-quality image data according to the judgment result. S23: The Transformer-UNet hybrid architecture is used to perform pixel-level semantic segmentation on the high-quality image data in S22, identify geological features such as fractures, pores, and bedding, and calculate reservoir parameters such as fracture density, porosity, and permeability. S24: Based on image recognition results and environmental parameter data, a Bayesian network algorithm is used to perform spatiotemporal registration and fusion analysis, and combined with geological models and historical data, a comprehensive assessment report of the underground geological environment is generated. S25: Based on the comprehensive assessment report, generate a new well logging data acquisition plan; S26: Generate control commands based on the new logging data acquisition scheme.

[0046] S3: The intelligent logging module executes the control commands in S2.

[0047] Therefore, the present invention adopts the above-mentioned distributed intelligent logging system and logging method. Through the deep integration of distributed architecture and artificial intelligence technology, it not only effectively solves the core pain points of traditional logging systems in terms of data transmission, quality control and decision lag, but also realizes adaptive optimization and closed-loop management of the logging process through distributed intelligent architecture, providing a brand-new technical means for efficient and accurate exploration of oil and gas fields.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A distributed intelligent logging system, characterized in that, It includes an intelligent logging module, a surface intelligent control module, and a communication module; there are N intelligent logging modules, where N is an integer greater than 1; The intelligent logging module is installed downhole and is used to collect, process, and upload logging data; The ground-based intelligent control module is located on the ground and is used to receive and analyze the data uploaded by the intelligent logging module and send new acquisition plans to the intelligent logging module. The intelligent logging module and the ground intelligent control module communicate bidirectionally through a communication module.

2. The distributed intelligent logging system according to claim 1, characterized in that: The intelligent logging module includes a data acquisition unit, a first processing unit, and a first communication interface unit; The data acquisition unit is used to acquire downhole data; the first processing unit, connected to the data acquisition unit, is used to preprocess the acquired downhole data; the first communication interface unit, connected to the first processing unit and the communication module, is used to interact with the ground intelligent control module through the communication module.

3. The distributed intelligent logging system according to claim 1, characterized in that: The ground intelligent control module includes a second communication interface unit, an intelligent analysis unit, and a control command generation unit; The second communication interface unit is used to interact with the first communication interface unit through the communication module; the intelligent analysis unit is connected to the second communication interface unit and is used to analyze and process the received downhole data and generate a new logging data acquisition scheme based on the analysis results. The control command generation unit, connected to the intelligent analysis unit and the second communication interface unit, is used to generate corresponding control commands based on the new logging data acquisition scheme, and send them to the intelligent logging module through the second communication interface unit and the communication module.

4. A distributed intelligent logging system according to claim 2, characterized in that, The specific working process of the first processing unit is as follows: Receive downhole data collected by the data acquisition unit; Data cleaning, format standardization, and compression are performed on downhole data; Add timestamps, depth identifiers, and module identifier metadata to the processed data; Based on the preset transmission strategy and downhole communication conditions, select the priority and transmission timing of the data to be uploaded; The encapsulated data is sent to the communication module through the first communication interface unit.

5. A distributed intelligent logging system according to claim 3, characterized in that, The intelligent analysis unit includes a data receiving and preprocessing subunit, an image quality intelligent assessment subunit, a geological feature intelligent identification subunit, a multi-source data fusion analysis subunit, and a scheme decision and output subunit; The data receiving and preprocessing subunit is used to receive the data packaged by the intelligent logging module, decode the data, convert the format and perform time-depth alignment processing, and extract image data and environmental parameter data; The image quality intelligent assessment subunit is connected to the data receiving and preprocessing subunit. It is used to perform multi-dimensional quality analysis on image data using a multi-task deep convolutional neural network, generate three quantitative scores of sharpness, contrast and signal-to-noise ratio, determine whether the score is lower than a preset threshold, and classify the image data into high-quality image data and low-quality image data according to the judgment result. The intelligent geological feature recognition subunit is connected to the intelligent image quality assessment subunit. It is used to perform pixel-level semantic segmentation on the high-quality image data output by the intelligent image quality assessment subunit using the Transformer-UNet hybrid architecture, identify geological features such as fractures, pores, and bedding, and calculate reservoir parameters such as fracture density, porosity, and permeability. The multi-source data fusion analysis subunit is connected to the image quality intelligent assessment subunit and the geological feature intelligent identification subunit. It is used to perform spatiotemporal registration and fusion analysis based on image recognition results and environmental parameter data, using Bayesian network algorithms, and combined with geological models and historical data to generate a comprehensive assessment report of the downhole geological environment. The scheme decision and output subunit is connected to the multi-source data fusion analysis subunit. It is used to generate a new well logging data acquisition scheme based on the comprehensive evaluation report and send the new acquisition scheme to the control command generation unit.

6. A distributed intelligent logging system according to claim 5, characterized in that: The multi-task deep convolutional neural network uses ResNet as its backbone network, including a global average pooling layer and a fully connected regression branch. The global average pooling layer is connected to the backbone network and is used to convert feature maps into feature vectors. The fully connected regression branch is connected to the global average pooling layer and is used to calculate and output three quantitative scores of image sharpness, contrast and signal-to-noise ratio.

7. A distributed intelligent logging system according to claim 5, characterized in that: The scheme decision and output subunit also includes a human-machine interface, which is used to present the new logging data acquisition scheme to the operator in a visual form and receive modification confirmation instructions issued by the operator based on the comprehensive evaluation report.

8. A distributed intelligent logging system according to claim 1, characterized in that: The communication module is a single-core armored cable, which enables bidirectional data transmission and power supply between the mine and the surface through cable remote transmission technology.

9. A distributed intelligent logging method, employing the distributed intelligent logging system described in any one of claims 1-8, characterized in that, Includes the following steps: S1: Collect downhole data and perform preprocessing; S2: Analyze and process the downhole data in S1, and generate a new logging data acquisition scheme based on the analysis results; generate corresponding control commands according to the new logging data acquisition scheme and send them to the intelligent logging module; S3: The intelligent logging module executes the control commands in S2.

10. A distributed intelligent logging method according to claim 9, characterized in that, S2 specifically includes: S21: Decode the data, convert the format, and perform time-depth alignment to extract image data and environmental parameter data; S22: Use a multi-task deep convolutional neural network to perform multi-dimensional quality analysis on image data, generate three quantitative scores of sharpness, contrast and signal-to-noise ratio, determine whether the score is lower than the preset threshold, and classify the image data into high-quality image data and low-quality image data according to the judgment result. S23: The Transformer-UNet hybrid architecture is used to perform pixel-level semantic segmentation on the high-quality image data in S22, identify geological features such as fractures, pores, and bedding, and calculate reservoir parameters such as fracture density, porosity, and permeability. S24: Based on image recognition results and environmental parameter data, a Bayesian network algorithm is used to perform spatiotemporal registration and fusion analysis, and combined with geological models and historical data, a comprehensive assessment report of the underground geological environment is generated. S25: Based on the comprehensive assessment report, generate a new well logging data acquisition plan; S26: Generate control commands based on the new logging data acquisition scheme.

Images