A data processing method and electronic device

Abstract

This application provides a data processing method and electronic device. The device identifies geological information by attribute recognition, determines multiple recognition events, and filters them by confidence level, converting them into data packets carrying adjustment actions. At the same time, it calculates the difference between each data packet and the 3D geological model to determine the transmission method, prioritizing the uploading of key geological changes. After receiving the data packets, the cloud accurately locates them in the corresponding target area in the overall model using spatial anchor points and binds them. However, the bound target area may be associated with multiple data packets with inconsistent update directions. Therefore, it is necessary to construct conflict vectors and determine conflict weights for the data packets corresponding to the target area. Then, based on the adjustment actions of each data packet, the conflict vectors, and the conflict weights, the adjustment method of the target area is determined and a local update is performed. This can obtain accurate and traceable incremental update results for the model, facilitating efficient, accurate, and traceable local updates of the 3D geological model under weak network conditions.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a data processing method and electronic device. Background Technology

[0002] Geological information is the core basis for updating three-dimensional geological models and risk prevention and control. The timeliness and accuracy of its on-site acquisition directly affect safety prediction and handling decisions.

[0003] Currently, geological information acquisition and model updates mainly rely on on-site manual recording, photo taking, geophysical data uploading, and back-end manual secondary interpretation. This results in problems such as slow data uploading, key geological facts being buried in a large number of original files, model updates relying on manual sorting and secondary interpretation, and difficulty in timely detection and handling of conflicts between multiple data sources. As a result, it is difficult to meet the timeliness and traceability requirements of local incremental updates of three-dimensional geological models. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a data processing method and an electronic device to solve or partially solve the above-mentioned technical problems.

[0005] To achieve the above objectives, this application provides a data processing method applied to a device, wherein the device stores a three-dimensional geological model of a target geological area, and the method includes:

[0006] Obtain geological information for the target geological area; Geological information is used to identify attributes and determine multiple identification events; Determine the first confidence level for each identification event, filter all identification events according to the first confidence level to obtain undetermined identification events, and convert the undetermined identification events into data packets, wherein the data packets include adjustment actions; Determine the difference between each data packet and the three-dimensional geological model, and based on each data packet and its corresponding difference value, determine the transmission method for each data packet. Each data packet is transmitted to the cloud according to its corresponding transmission method. The overall model located in the cloud is then adjusted based on the adjustment actions corresponding to the data packets to obtain the adjusted overall model.

[0007] To achieve the above objectives, this application also provides a data processing method applied in the cloud, wherein the cloud stores an overall model, which is a three-dimensional geological model corresponding to the entire geological region, and the method includes: Acquire at least one data packet transmitted from the device, wherein the data packet includes an adjustment action; For each data packet: determine the target anchor point corresponding to the data packet in the overall model, combine the target anchor point with the data packet, and determine the target region corresponding to the data packet in the overall model; Bind the data packet to the corresponding target area; For each data packet corresponding to the target area, a collision vector is constructed and a collision weight is determined. Based on the adjustment action, collision vector and collision weight corresponding to each data packet, the adjustment method corresponding to the target area is determined. The target area is adjusted according to the corresponding adjustment method to obtain the adjusted overall model.

[0008] Based on the same inventive concept, this disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.

[0009] As can be seen from the above, the data processing method and electronic device provided in this application can perform attribute identification on the acquired geological information, determine multiple identification events, and filter each identification event according to a first confidence level. The filtered undetermined identification events are then converted into data packets carrying adjustment actions. At the same time, the difference between each data packet and the three-dimensional geological model is calculated to adaptively determine the corresponding transmission method. However, the data packets obtained in this way only contain adjustment actions and lack spatial correspondence with the overall model. Therefore, each data packet needs to be transmitted to the cloud according to the corresponding transmission method. The cloud determines the target anchor point corresponding to each data packet in the overall model and determines the corresponding target area by combining the target anchor point and the data packet. Then, the data packets are bound to the target area. However, the bound target area may be associated with multiple data packets at the same time and the update directions are inconsistent. Therefore, it is necessary to construct conflict vectors and determine conflict weights for the data packets corresponding to the target area. Then, the adjustment method of the target area is determined by comprehensively considering the adjustment actions of each data packet, the conflict vectors, and the conflict weights, and local adjustments are performed. This enables precise and independent incremental adjustments to each target area in the overall model, realizing local fine-grained updates of the overall model. The entire process is completed automatically, which facilitates efficient, accurate, and traceable updates of dynamic three-dimensional geological models under weak network conditions. Attached Figure Description To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A flowchart illustrating the data processing method of this application embodiment. Figure 1 ; Figure 2 A flowchart illustrating the data processing method of this application embodiment. Figure 2 ; Figure 3 This is a schematic diagram of the structure of a data processing apparatus according to an embodiment of this application. Figure 1 ; Figure 4 This is a schematic diagram of the structure of a data processing apparatus according to an embodiment of this application. Figure 2 ; Figure 5 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0012] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0013] Currently, information platforms for tunnel engineering typically categorize on-site data into types such as geological logging, advanced forecasting, monitoring and measurement, construction logs, image data, test data, and model data. On-site personnel upload photos of the tunnel face, surrounding rock grade, joint and fracture descriptions, groundwater conditions, and construction records via mobile terminals or web forms. The cloud platform then archives the data, generates reports, and provides visualizations.

[0014] These solutions address the difficulty of managing paper records, but their data organization remains centered on documents, forms, and reports. After a working face photo is uploaded, the platform knows there is a photo, but it may not know which lithological boundary in the photo requires modification of the 3D model; after an advance borehole record is uploaded, the platform knows there is data for a borehole, but it may not know which borehole depth corresponds to which model block.

[0015] Another type of existing approach focuses on the dynamic updating of 3D geological models. These typically involve re-interpolating stratigraphic interfaces, correcting fault boundaries, or adjusting rock mass properties based on newly added boreholes, geological profiles, geophysical inversion results, or construction data. The emphasis of this type of approach is on the model reconstruction algorithm, rather than how field data enters the model update process with minimal data transfer, minimal ambiguity, and executable semantics.

[0016] Other approaches use voxels, meshes, octrees, or 3D tiles to organize geological models and point cloud data, supporting spatial queries, local display, and local updates. These techniques can improve the efficiency of 3D model management, but they typically assume that update data has already been processed on the server side and do not address how to generate geological update events bound to affected voxels at the field end.

[0017] Existing tunnel risk early warning platforms also have the concept of "event reporting," such as water inrush events, deformation exceeding limits events, support abnormality events, and equipment failure events. These events mainly serve safety early warning, work order circulation, or emergency response, and do not directly carry 3D geological model update instructions, nor do they include model version dependencies, affected model blocks, and conflict merging rules.

[0018] In terms of data compression, point cloud compression, 3D tile compression, image compression, and differential transfer technologies are relatively mature. These methods mainly reduce data volume or improve visualization loading speed, but they often don't know which compressed information will change the geological model and which information is merely archived evidence. Simple compression does not guarantee that the cloud will automatically update the model accordingly.

[0019] In related technologies, cloud-based uploading of geological information for tunnel construction and model updates are typically achieved using mobile data acquisition and uploading systems or dynamic 3D geological model update systems for data archiving and model reconstruction. However, significant shortcomings remain: What dynamic 3D geological models truly need is not the original files themselves, but rather the facts within those files that will alter the model. For example, the appearance of lithological interfaces at a certain mileage section, a sudden change in drilling rate at a specific borehole depth, dense fractures appearing in borehole wall imaging, or an anomaly requiring an increase in the probability of water abundance. Current upload methods fail to translate these facts into executable model update units.

[0020] Most data acquisition systems are centered on files, forms, and reports, focusing on the complete archiving of data, but lack the accurate extraction of facts about local geological changes, and cannot achieve refined management such as "prioritizing the uploading of key events and driving incremental updates of models".

[0021] In practical engineering, the collected geological data may be multi-sourced and heterogeneous, large in volume, and subject to unstable on-site network bandwidth. Meanwhile, critical information such as tunnel face exposure, pre-drilling anomalies, and geophysical interpretation constitutes a small portion of the original files, making it difficult for existing methods to guarantee the timeliness and local accuracy of model updates. Directly synchronizing the entire model or uploading the entire file often requires significant bandwidth and time, and in this scenario, critical data is easily buried in massive amounts of original files, leading to update delays or conflict overwriting. Furthermore, geological data uploading and 3D geological model updates are usually performed separately. Indicators such as model changes, version dependencies, and multi-source data conflict handling rely heavily on manual judgment and secondary interpretation in the background, lacking an end-to-end automated process from on-site data to incremental model updates.

[0022] The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0023] The data processing method proposed in the embodiments of this application.

[0024] like Figure 1 As shown, this method is applied to the device side, which stores a three-dimensional geological model of the target geological area. The execution process of this method includes: The "device end" refers to the data acquisition and edge computing equipment deployed at the tunnel construction site, including but not limited to: tablets or mobile terminals used by on-site personnel, drilling rig acquisition units, ground-penetrating radar mainframes, geophysical data acquisition instruments, and edge computing gateways or industrial control computers deployed in on-site construction management stations or temporary computer rooms. The device end possesses data storage, local computing, and wireless communication capabilities, used for real-time identification, event packet encapsulation, and uploading and scheduling of geological information acquired on-site, and for maintaining version synchronization and feedback interaction with the overall model in the cloud.

[0025] The target geological region refers to the geological spatial range corresponding to a specific tunnel mileage section under construction within the current acquisition cycle. Each acquisition cycle corresponds to a fixed construction section interval, which is defined by the starting and ending mileages and associated with the local cross-sectional coordinates, borehole opening coordinates, and borehole direction parameters within that section. When an acquisition cycle ends, as the tunnel face advances to a new mileage section, the system enters the next acquisition cycle, and the target geological region is updated to the new construction section interval. The original region is archived and removed from the device's local active storage.

[0026] The 3D geological model, stored on the device, is a data copy of the overall model in the cloud within the target geological area. It corresponds to the construction segment of the current acquisition cycle. Within each segment, voxel or octree data structures are used to express attributes such as lithology type, fault distribution, water-bearing probability, fracturing probability, and model confidence. The 3D geological model on the device is synchronously loaded from the cloud at the beginning of the acquisition cycle and continuously updated as construction progresses by receiving version increments from the cloud. When the acquisition cycle ends and the target geological area changes, the device unloads the current model snapshot, loads the model snapshot corresponding to the new area, and begins the local storage and update closed loop for the next cycle.

[0027] Step 101: Obtain geological information of the target geological area.

[0028] In practice, multi-source raw data—geological information—of the target geological area is obtained through on-site survey equipment and data acquisition terminals.

[0029] The exploration equipment includes drilling rigs and their drilling measurement systems, ground-penetrating radar, seismic wave detectors, borehole imagers, and face exposure acquisition devices; the data acquisition terminals include speed sensors, torque sensors, drilling speed sensors, return water flow meters, and return water turbidity meters installed on the drilling rig, as well as high-definition industrial cameras for face image acquisition and mobile data entry devices for manual recording.

[0030] The multi-source raw data—geological information—includes photographs of the drilling face, borehole drilling parameters, geophysical anomaly data, and manually compiled information. Based on this geological information, the following can be obtained: continuously recorded drilling speed, torque, water return, and turbidity sequences during drilling; reflection intensity profiles and locations of anomalous phase axes obtained from ground-penetrating radar; seismic wave velocity distribution and reflection interface locations obtained from seismic wave detection; fracture trace length, opening, and infill type obtained from borehole imaging; lithological zoning boundary coordinates, structural plane orientation, water trace distribution areas, and area proportions revealed at the drilling face; and descriptions of surrounding rock grade, joint development degree, and water production status entered by on-site geological engineers.

[0031] In specific step 102, attribute identification is performed on the geological information to determine multiple identification events.

[0032] In practice, geological information collected on-site is used as input, and a pre-built attribute recognition rule engine is used for item-by-item recognition processing. By mapping multi-source heterogeneous geological information to a unified geological attribute space, multiple recognition events are generated. Each recognition event corresponds to a preliminarily determined geological object or anomaly feature, carrying its type label, location information, attribute information, and preliminary confidence level. Simultaneously, to ensure the traceability of subsequent processing, each recognition event records its data source identifier, collection time, and dependent model version information to support version dependency checks and event tracking.

[0033] Step 103: Determine the first confidence level for each identification event, filter all identification events according to the first confidence level to obtain undetermined identification events, and convert the undetermined identification events into data packets.

[0034] The data packet includes adjustment actions.

[0035] In practice, for each generated identification event, the first confidence level of the event is calculated by first calculating a weighted sum of multiple confidence factors and then mapping them. The first confidence level is used to characterize the credibility of the identification event and its value ranges from 0 to 1.

[0036] Subsequently, based on the statistical distribution characteristics of the first confidence level of all identified events within the current batch, each event is categorized and screened. Specifically, the judgment is made according to the confidence level: if the identified event is one that can be directly adopted or adopted after enhancement and upload, it is treated as a pending identified event and enters the subsequent upload and update process; if the identified event is a review queue event, it is not entered into the main model update process; if the identified event cannot be uploaded, it is rejected or archived as an observation.

[0037] Furthermore, for the selected pending events, they are encapsulated into lightweight geological event packets according to a preset data packet structure. These packets contain at least an adjustment action field. The adjustment action includes the type of operation performed by the identified event on the overall model, enabling the cloud to perform corresponding adjustments based on the adjustment action upon receiving the data packet.

[0038] Step 104: Determine the difference value between each data packet and the three-dimensional geological model, and determine the transmission method corresponding to each data packet based on the difference value.

[0039] In practice, for each generated lightweight geological event data package, before uploading it to the cloud, it is necessary to first assess the degree of difference between the geological information carried by the data package and the current version of the 3D geological model.

[0040] This difference value is not a single indicator, but is calculated comprehensively from three dimensions: geometric information, attribute information, and confidence level information. Geometric difference reflects the spatial offset between the geological interface or anomaly boundary indicated by the data packet and the existing geometric boundary of the 3D geological model; attribute difference reflects the deviation between the attribute values ​​(lithology, water-bearing probability, or fracturing probability) carried by the data packet and the current predicted values ​​of the 3D geological model; and confidence level difference reflects the reliability of the data packet's identification.

[0041] By weighted fusion, the difference value of each data packet is obtained. Then, based on the difference values ​​of all data packets to be uploaded within the current collection period, the transmission mode corresponding to each data packet is determined: data packets with significant differences and a large impact on construction safety are classified into high-priority real-time transmission mode, while data packets with moderate differences or limited impact are classified into batch transmission or delayed transmission mode.

[0042] Step 105: Transmit each data packet to the cloud according to the corresponding transmission method, and adjust the overall model located in the cloud according to the adjustment action corresponding to the data packet to obtain the adjusted overall model.

[0043] In practice, for the already encapsulated data packets, the system sends them to the cloud according to their corresponding transmission method. The cloud then performs corresponding update operations on the affected model blocks in the overall model based on the adjustment actions carried in the data packets, including adding geological objects, correcting geometric boundaries, adjusting attribute probabilities, or freezing updates.

[0044] The above scheme enables the device to perform attribute identification on the acquired geological information of the target geological area, identify multiple identification events, and determine the first confidence level of each identification event. All identification events are then filtered according to the first confidence level, and the filtered pending identification events are converted into data packets carrying adjustment actions. However, these data packets only contain adjustment actions and lack spatial correspondence with the overall model in the cloud. Therefore, it is necessary to determine the difference value between each data packet and the 3D geological model, and adaptively determine the corresponding transmission method for each data packet based on the difference value. Then, each data packet is transmitted to the cloud according to the corresponding transmission method, so that the cloud can adjust the overall model based on the adjustment actions in the data packets. This ensures that each data packet is uploaded in an orderly manner as needed, achieving priority transmission and model-driven operation of key events under weak network conditions, facilitating accurate incremental updates of the overall model. The entire process is automated, bringing great convenience to users.

[0045] In some embodiments, the geological information includes multiple data to be processed, and step 102 includes: The data to be processed includes original files, waveform curves, or form records.

[0046] Step 1021: Tag all the data to be processed according to the preset data source identifier to obtain multiple first data with data source identifiers.

[0047] In practice, the device identification prefix, file format, and data field information of each piece of data to be processed are read and matched with the preset data source identifier in the database. Each piece of data to be processed is matched and labeled with the corresponding data source identifier to obtain the data to be processed with the data source identifier, which is the first data.

[0048] Among them, the data source identifier is a classification label used to mark the source type of geological information. Source Category {Working face exposure, advanced drilling, geophysical interpretation, borehole imaging, monitoring data, manual verification}.

[0049] The data includes: face exposure: information on lithology and water features at the face obtained by on-site personnel through photography and manual recording; advanced drilling: parameters such as drilling speed, torque, and water return obtained through the drilling rig's measurement-while-drilling system; geophysical interpretation: anomaly data interpreted through geophysical methods such as ground-penetrating radar and seismic wave detectors; borehole imaging: borehole wall fracture images and structural surface information obtained through borehole imaging instruments; monitoring data: data collected in real time by sensors deployed during construction; and manual verification: data verified and corrected by on-site geological engineers or experts.

[0050] Step 1022: Extract features from each first data point to obtain multiple feature data points. Classify all feature data points to obtain multiple geological object information corresponding to each first data point.

[0051] In practice, for each piece of first data that already has a data source identifier, feature extraction is performed according to the parsing rules corresponding to its data source identifier to obtain geological object information. .

[0052] The geological object information includes location data and multiple attribute values. Location data characterizes the spatial location of the geological object in the construction coordinate system, while attribute values ​​characterize quantitative features such as lithology, water-bearing probability, and fracturing probability. This allows for comparison of each geological object's information with the predicted object information at the corresponding location in the 3D geological model to determine adjustment actions and generate corresponding recognition events. Different feature extraction methods are used for different source categories.

[0053] For example, for the first data source identified as the tunnel face exposure, the tunnel face image data and manually recorded information are first read. Lithological boundary coordinates are extracted using edge detection algorithms, structural surface trace direction vectors are extracted using structural surface analysis, and the area and location of water-bearing areas are extracted using color segmentation and texture analysis. These extraction results are used as feature data. Then, the data is classified according to preset geological object mapping rules: if lithological boundary coordinates are extracted and there is an interface offset with adjacent mileage sections, it is classified as a lithological interface; if water-bearing areas are extracted and their area proportion exceeds a preset threshold, it is classified as water-rich areas; if multiple sets of structural surface traces are extracted and their density exceeds a preset threshold, it is classified as fractured areas.

[0054] For example, for the first data source identified as advanced drilling, the drilling speed sequence, torque sequence, return water volume sequence, and return water turbidity sequence are first read. A sliding window statistical analysis is performed on each sequence, extracting the mean, variance, abrupt change amplitude, and duration of the abrupt change within each window. These statistics are then used as feature data. Next, the data is classified according to a preset geological object mapping rule: if the drilling speed sequence shows a positive abrupt change within a continuous borehole depth range and the return water volume increases synchronously, it is classified as a fractured zone; if the return water turbidity sequence shows a continuous increase and the drilling speed does not change significantly, it is classified as a weak interlayer; if the torque sequence shows periodic fluctuations and the drilling speed fluctuation amplitude exceeds a preset threshold, it is classified as a fractured development section.

[0055] For example, for the first data source identified as geophysical interpretation, the reflection intensity profile or resistivity profile data is first read. The coordinates of the anomaly center are extracted through local extremum detection, and the boundary control points of the anomaly body are extracted through contour tracing. The corresponding reflection intensity or resistivity values ​​are then read, and these extraction results are used as feature data. Then, the anomaly body is classified according to preset geological object mapping rules: if the anomaly body is continuously distributed in a strip-like pattern and the reflection intensity exceeds a preset threshold, it is classified as a fault anomaly; if the anomaly body is distributed in a blocky pattern and the resistivity is lower than a preset threshold, it is classified as a water-rich anomaly; if the anomaly body is distributed in a scattered pattern and the reflection phase axis is discontinuous, it is classified as a fractured anomaly.

[0056] For example, for the first data source identified as borehole imaging, the fracture trace image is first read. The fracture trace length, aperture, and infill material type vectors are extracted through edge detection and line segment fitting, and these extracted results are used as feature data. Then, the data is classified according to preset geological object mapping rules: if the fracture trace length exceeds a preset threshold and the aperture is greater than a preset threshold, it is classified as an open fracture; if the infill material type contains clay minerals and the fracture density exceeds a preset threshold, it is classified as a fracture zone; if the fracture is closed and the trace direction is consistent with the direction of the regional principal stress, it is classified as a closed fracture.

[0057] For example, for the first data source identified as monitoring data, the real-time monitoring sequence (including data collected by sensors such as displacement gauges, pressure gauges, piezometers, and strain gauges) is first read. Trend analysis is performed on each monitoring sequence to extract the rate of change, cumulative change, acceleration characteristics, and frequency of exceeding limits. These statistics are used as feature data. Then, the data is classified according to preset geological object mapping rules: if the displacement rate continues to increase and the cumulative displacement exceeds a preset threshold, it is classified as a deformation acceleration zone; if the piezometer reading shows a step increase and is related to rainfall or construction disturbance time, it is classified as a water pressure anomaly zone; if the strain gauge reading shows a sudden change and coincides with the location of the adjacent geological interface, it is classified as a stress concentration zone.

[0058] For example, for the first data source identified as manually reviewed, the review form data entered by the on-site geological engineer is first read, including the surrounding rock level adjustment record, water discharge status description, joint development degree correction, and lithology confirmation information. The qualitative description code and quantitative correction value are extracted as feature data. Then, the data is classified according to the preset geological object mapping rules: if the surrounding rock level differs from the model prediction level by more than two levels, it is classified as surrounding rock level deviation; if the water discharge status is marked as "strain-like water discharge" or "water inrush," it is classified as water inrush risk; if the review record clearly describes "the existence of an unforeseen fault" or "lithological abrupt change," it is classified as an unforeseen geological body.

[0059] Information for each geological object: Step 1023: Obtain the predicted object information corresponding to the geological object information in the three-dimensional geological model, determine the deviation between the geological object information and the predicted object information, and determine the adjustment action.

[0060] In practice, based on the location data of the geological object information, a spatial search is performed on the 3D geological model stored on the device to obtain the predicted object information currently predicted by the model at that location. This predicted object information consists of the predicted data already stored at the corresponding spatial location in the 3D geological model, including at least the predicted location data and multiple predicted attribute values, reflecting the model's current version. The following is an estimate of the geological conditions at this location.

[0061] Then, the geological object information and the predicted object information are compared item by item to determine the deviation between them: the positional deviation is determined based on the location data and the predicted location data, which characterizes the degree of spatial offset between the actual location of the geological object and the model's predicted location; the attribute deviation is determined based on multiple attribute values ​​and multiple predicted attribute values, which characterizes the direction and magnitude of the difference between the actually observed geological attributes and the model's predicted attributes. Through the above comparisons, the deviation between the geological object information and the predicted object information in two dimensions—spatial location and attribute characteristics—is obtained.

[0062] Finally, based on the determined deviation and combined with preset judgment rules, the adjustment action corresponding to the current geological object information is determined from multiple candidate adjustment actions. {Add, Correct, Increase Probability, Decrease Probability, Request Review, Freeze Update}.

[0063] Step 1024: Determine the difference vector based on the deviation between the geological object information and the predicted object information.

[0064] In practice, the geometric positional deviation is determined based on the location data of the geological object information and the predicted location data of the predicted object information. , where the geometric position deviation is the Euclidean distance in three-dimensional space.

[0065] Then, based on the multiple attribute values ​​of the geological object information and the multiple predicted attribute values ​​of the predicted object information, the attribute deviations in each attribute dimension are calculated. The attribute probability deviations are then calculated by taking the square root of the sum of the squares of the attribute deviations. .

[0066] Next, based on the second confidence level of the geological object information and the predicted confidence level of the corresponding location in the three-dimensional geological model, the initial confidence level deviation of the model is determined. .

[0067] Finally, the geometric position bias, attribute bias, and confidence bias are weighted and fused, and the difference vector is obtained by taking the square root of the weighted sum of squares. :

[0068] in, Geometric position deviation (meters) This represents the attribute probability bias. This represents the initial confidence level bias. , , The weights are fixed.

[0069] Optionally, when historical verification results for a certain type of event indicate that the source is more accurately located geometrically, the system automatically increases... And correspondingly reduce other weights; when a certain type of event is more reliable in attribute identification, the system automatically increases the weight. This difference vector comprehensively characterizes the overall deviation between the identification results and the three-dimensional geological model; the larger the difference vector, the more significant the difference between the identification results and the model predictions.

[0070] Step 1025: Determine whether the difference vector exceeds the first threshold.

[0071] In practice, it is determined whether the difference vector exceeds a preset first threshold. .

[0072] Step 1026: In response to the difference vector exceeding the first threshold, an identification event is generated based on the data source identifier, geological object information, adjustment action, and difference vector.

[0073] In specific implementation, if Exceeding the first threshold Then As an output for identifying events.

[0074] The above scheme categorizes multi-source geological data based on preset data source identifiers. Differentiated feature extraction methods are used to extract geological object information containing location and multiple attribute values ​​from different sources, such as tunnel face exposure, advanced drilling, and geophysical interpretation. This information is then compared with the predicted object information in the 3D geological model using multi-dimensional deviations in spatial location and attribute probability. Geometric deviation, attribute deviation, and confidence deviation are fused to construct a difference vector. When the difference vector exceeds a preset threshold, an identification event containing deviation details and adjustment actions is automatically generated. This achieves automated interpretation, refined comparison, and anomaly-driven model update closed loop for multi-source heterogeneous geological information in tunnel construction. This effectively eliminates subjective differences in manual interpretation, accurately captures hidden conflicts between actual on-site exposure and model prediction, and drives continuous dynamic calibration and reliability improvement of the 3D geological model.

[0075] In some embodiments, the geological object information includes location data and multiple attribute values; the adjustment action includes at least one of the following: adding an action, correcting an action, increasing probability an action, decreasing probability an action, requesting review an action, and freezing update an action; step 1023 includes: Step 10231: Determine the second confidence level corresponding to the geological object information based on all the feature data corresponding to the geological object information.

[0076] In practice, multiple confidence parameters are extracted from all feature data corresponding to the geological object information. Each confidence parameter characterizes the credibility of the geological object information in a single dimension. The confidence weight corresponding to each confidence parameter is obtained, and each confidence parameter is multiplied by its corresponding confidence weight to obtain multiple weighted values. All weighted values ​​are summed to obtain an initial confidence value. The initial confidence value is then normalized to obtain a second confidence level ranging from 0 to 1.

[0077] Step 10232: Obtain anchor point coordinates based on location data, determine whether the 3D geological model includes anchor point coordinates, and obtain the determination result.

[0078] In practice, based on the location data in the geological object information, the location data is converted into three-dimensional geological model coordinates through the transformation matrix between the construction coordinate system and the three-dimensional geological model coordinate system to obtain the anchor point coordinates. The anchor point coordinates are then located in the three-dimensional geological model to determine whether the three-dimensional geological model includes the anchor point coordinates.

[0079] The location data includes: tunnel mileage station number, local cross-section coordinates or borehole opening coordinates and borehole orientation parameters.

[0080] Step 10233: In response to the fact that the anchor point coordinates are not included in the three-dimensional geological model, the judgment result is negative, and the second confidence level is greater than the second threshold. If the first condition is met, the adjustment action is changed to a new action; or.

[0081] In practice, when the anchor point coordinates obtained from the location data cannot be located in the three-dimensional geological model, it indicates that no prediction object information has been established at that location in the current model version.

[0082] At this point, the system enters the new action determination process: First, it obtains the second confidence level corresponding to the geological object information. Then, it compares the second confidence level with a preset second threshold. If the second confidence level is greater than the second threshold, it is determined that the first condition is met, and the action is adjusted to a new action; if... It does not exist in the three-dimensional geological model and has a second confidence level. Second threshold, New addition.

[0083] The second threshold is the minimum confidence threshold for new actions.

[0084] For example, if a newly formed fracture zone not recorded in the three-dimensional geological model is revealed at the working face, and the confidence level of the fracture zone is greater than the second threshold, the system determines that there is no corresponding prediction object information at this location in the model, and the identification result is reliable, and the adjustment action is determined to be a new action.

[0085] Step 10234: In response to the inclusion of anchor point coordinates in the three-dimensional geological model, if the judgment result is yes, then proceed with the processing.

[0086] In practice, when the anchor point coordinates obtained from the location data can be located in the corresponding spatial position in the three-dimensional geological model, it indicates that the prediction object information has been established at that position in the current model version.

[0087] Step 10235: Obtain the prediction object information corresponding to the anchor point coordinates in the three-dimensional geological model.

[0088] The predicted object information includes predicted location data and multiple predicted attribute values.

[0089] In practice, for anchor point coordinates that have been confirmed to exist in the three-dimensional geological model, the anchor point coordinates are used as spatial retrieval keywords to query the spatial index structure of the three-dimensional geological model and obtain the predicted object information at the anchor point coordinates.

[0090] The predicted location data in the predicted object information includes: the coordinates of the interface control point currently stored in the model at the location of the anchor point, the coordinates of the anomaly center, or the coordinates of the lithological zone boundary. Multiple predicted attribute values ​​in the predicted object information include: quantitative attributes such as the probability of lithological category, water abundance, fracturing, faulting, and model confidence level currently stored in the model.

[0091] The three-dimensional geological model is organized using a voxel or octree data structure. Each voxel or octree node stores the corresponding spatial location and the geological attribute information at that location.

[0092] Step 10236: Determine the position deviation based on the position data and the predicted position data, and determine the attribute deviation based on multiple attribute values ​​and multiple predicted attribute values.

[0093] In practice, the Euclidean distance in three-dimensional space is calculated as the positional deviation based on the location data in the geological object information and the predicted location data in the predicted object information. Based on the multiple attribute values ​​in the geological object information and the multiple predicted attribute values ​​in the predicted object information, the deviation in each attribute dimension is calculated one by one, and the attribute deviation is obtained by calculating the square root of the sum of the squares of the attribute deviations.

[0094] Step 10237: In response to the position deviation and / or attribute deviation being less than the third threshold, if the second condition is satisfied, the adjustment action is a correction action; or...

[0095] In practice, the positional deviation and attribute deviation are compared with a preset third threshold. The third threshold is the deviation tolerance, which is used to determine whether the current recognition result only needs minor adjustments rather than significant changes.

[0096] Furthermore, if the location deviation is less than the location deviation tolerance and / or less than the attribute deviation tolerance, it indicates that the deviation between the identified event and the model prediction in terms of spatial location or attribute features is within an acceptable small range. Therefore, the current identification result is determined to be basically consistent with the model's existing predictions, but the model boundaries or attribute values ​​still need fine-tuning to better match the actual observed data. At this point, the second condition is triggered, determining the adjustment action as a correction action: if In a 3D geological model, there exist instances where the location or attribute deviations exceed the tolerance for deviation. Correction.

[0097] The correction action indicates that subsequent data packets carry the instruction to "locally fine-tune the geometric boundaries or attribute values ​​of the target model block in the direction of the recognition result".

[0098] Step 10238: In response to the attribute deviation being positive and the attribute value being greater than the predicted attribute value, if the third condition is met, then the adjustment action is to increase the probability action; or...

[0099] In practice, after confirming that the anchor point coordinates exist in the three-dimensional geological model and that the corresponding prediction object information has been obtained, the attribute deviation of each of the multiple attributes of the geological object information is calculated one by one. When the attribute deviation of a certain attribute is positive and the attribute value of the attribute is greater than the corresponding prediction attribute value, it indicates that the field identification results show an enhancing trend in the direction of the attribute, that is, the actual observed geological features exceed the current prediction level of the model.

[0100] At this point, the third condition is triggered, determining that the adjustment action is an action that increases the probability: if In a 3D geological model, there exists an attribute whose identification result enhances a certain attribute direction with a positive bias. Increase the probability.

[0101] Among them, the probability increase action instruction carries the instruction to "increase the probability value of the corresponding attribute of the target model block towards the recognition value" in subsequent data packets.

[0102] Step 10239: In response to the attribute deviation being negative and the attribute value being less than the predicted attribute value, if the fourth condition is met, then the adjustment action is to reduce the probability action; or...

[0103] In practice, after obtaining the predicted object information corresponding to the anchor point coordinates from the 3D geological model, the attribute values ​​in the geological object information are compared with the corresponding predicted attribute values. The difference in each attribute dimension is calculated one by one to obtain the attribute deviation. If the attribute deviation of a certain attribute dimension is negative, that is, the attribute value is less than the predicted attribute value, it indicates that the actual observation result shows a weakening trend in that attribute direction relative to the current prediction of the model.

[0104] At this point, obtain multi-source evidence supporting the location of the anchor point coordinates, including but not limited to: whether there are other historical events that also point to a weakening of the attribute direction, whether the geological rules consistent with the weakening direction are met, and whether there are spatially adjacent confirmed events that support the weakening direction.

[0105] If, within the historical adjustment records corresponding to the anchor point coordinates, there exists at least one piece of high-confidence evidence consistent with the weakening direction, then the fourth condition is satisfied, and the adjustment action is determined to be a probability-reducing action: if The model contains elements that weaken a certain attribute in a certain direction, with a negative bias, and for which there is already evidence to support this. Reduce the probability.

[0106] Among them, the probability reduction action indicates that subsequent data packets should carry an update instruction to "adjust the probability of the corresponding attribute of the target model block in the direction of decreasing".

[0107] Step 102310, in response to: the historical adjustment action corresponding to the anchor point coordinates in the three-dimensional geological model has a different adjustment direction than the probability-increasing action, and / or the historical adjustment action corresponding to the anchor point coordinates in the three-dimensional geological model has a different adjustment direction than the probability-reducing action, if the fifth condition is satisfied, then the adjustment action is a request for review action; or.

[0108] In practice, for anchor point coordinates where corresponding prediction object information already exists in the 3D geological model, the historical adjustment records for those anchor point coordinates are further obtained. These historical adjustment records include all adjustment actions performed at that location in previous model update cycles and their directions.

[0109] Then, historical events that belong to either probability-increasing or probability-decreasing actions and have a confidence level higher than a preset high-confidence threshold are selected from the historical adjustment records. The adjustment direction corresponding to the current identification result is compared with the adjustment direction of the selected historical adjustment actions. Here, the adjustment direction is either the "positive" direction of the probability-increasing action or the "negative" direction of the probability-decreasing action.

[0110] Finally, if the direction of attribute deviation in the current identification result is opposite to the direction of adjustment in the historical adjustment actions, it is determined that the current identification result contradicts the existing high-weight events. At this point, the fifth condition is triggered, determining the adjustment action as a request for review: if... The deviation exists in the 3D geological model, but its direction contradicts the existing high-weight events. Request for review.

[0111] The request for review action indicates that subsequent data packets should carry the instruction to "not directly modify the model, but instead generate a review task and wait for manual confirmation".

[0112] Step 102311: In response to the failure to meet any of the first, second, third, fourth, and fifth conditions, the adjustment action is changed to a freeze update action.

[0113] In practice, after determining the above five conditions in sequence, if the current geological object information does not meet any of the conditions, then the identification result is determined to be unnecessary to drive model updates, and the adjustment action is identified as a freeze update action. If a 3D geological model exists but does not satisfy any of the above conditions, then... Updates frozen.

[0114] The freeze update action indicates that subsequent data packets should carry the instruction that "the currently identified event will not drive model modification, but will only be archived as an observation record".

[0115] The above scheme sets six-level stratification criteria for the 3D geological model. Logical arbitration is performed by combining the confidence level of geological object information, spatial location deviation, multi-attribute deviation direction, and historical adjustment records. This achieves refined, differentiated, and conflict-avoidance management of multi-source identification event-driven model updates. It can automatically incorporate high-confidence new discoveries into the model, fine-tune boundaries and attribute probabilities, and sensitively capture the enhancement and weakening trends of geological features. It can also trigger manual review or freeze erroneous updates when contradictory signals occur. This effectively avoids model oscillations and error accumulation caused by low-quality data or occasional conflicts, and significantly improves the accuracy and long-term stability of the 3D geological model in the tunnel construction process.

[0116] In some embodiments, the identified event includes multiple confidence parameters, each of which can affect a first confidence level. Step 103 includes: Step 1031: For each identified event: sum the confidence parameters and their corresponding confidence weights to obtain the first confidence level.

[0117] In practice, the formula for calculating the first confidence level is as follows: .

[0118] in, The confidence level parameter – the reliability of the data source – is calculated using the following formula: .

[0119] Specifically, Based on the reliability of the source. Among them: face exposure = 0.80, advanced drilling = 0.75, borehole imaging = 0.70, geophysical interpretation = 0.55, monitoring = 0.60, manual verification = 0.90.

[0120] Specifically, The calculation formula for the equipment calibration status is as follows: .in, This is the most recent calibration date.

[0121] Specifically, The accuracy of the coordinate system calibration is based on the nominal error in the construction survey report.

[0122] in, The confidence parameter is feature clarity.

[0123] Specifically, for image data, the calculation formula is: ,in, This represents the number of low-contrast pixels in the image. This represents the total number of pixels in the image. This is the complement of the proportion of low-contrast pixels. The less low-contrast pixels there are, the closer this value is to 1, and the higher the image clarity.

[0124] Specifically, for curve data, the calculation formula is: .in, The signal-to-noise ratio of the object currently being processed. Let be the signal-to-noise ratio of the (i+1)th processed object.

[0125] Specifically, regarding form data, This equals the percentage of required fields filled.

[0126] in, The confidence parameter – spatial anchor integrity – is calculated using the following formula: .

[0127] Specifically, The total number of required spatial fields for this source type is as follows: 4 for face events: mileage, cross-section x, cross-section y, number of calibration points; 5 for borehole events: borehole mileage, borehole cross-section coordinates, borehole azimuth, borehole dip angle, borehole depth. The number of fields for which valid values ​​have been provided. If Then it will directly return a "Insufficient spatial anchor points" receipt.

[0128] in, The confidence parameter – multi-source support level – is calculated using the following formula: .

[0129] Specifically, To satisfy the total number of events that meet the spatial proximity condition, Spatial distance The set of events within the scope is all events related to the target event. spatial distance Less than the nearest neighbor threshold The set of events, The spatial proximity radius threshold; For direction cosine similarity, To iterate through each event in the collection.

[0130] in, The confidence parameter – time freshness – is calculated using the following formula: .

[0131] Specifically, This represents the time (in hours) since the latest model version was released. .

[0132] in, The confidence level parameter – the status of manual review – is calculated using the following formula: .

[0133] Specifically, This indicates that manual verification has been performed. 0.75 indicates that cross-checking has been performed. 0.50 indicates that only automatic annotation is used. Weighting of the qualifications of the reviewers.

[0134] in, The confidence parameter – degree of conflict – is calculated using the following formula: This refers to the percentage of confirmed conflict events within a given spatial area.

[0135] Specifically, Within the specified spatial range, this represents the total number of neighboring events that have been identified as conflicting with target event i. This represents the total number of all neighboring events within the same spatial range, regardless of whether they conflict with the target event.

[0136] In addition, the confidence weights corresponding to each confidence parameter are automatically corrected based on the verification feedback accumulated during construction: For example: Successful verification feedback: For each type of event, count the number of times that type of event has been verified as valid through subsequent construction in history. In this case, successful verification means that after the event is updated, the results of subsequent construction excavation or drilling confirmation are consistent with the direction of the event judgment and the deviation is within an acceptable range. At this time, Adjustment formula: .

[0137] For example: Rejection Feedback: Counts the number of times this type of event was manually rejected or invalidated. and total number of times in history Conflict penalty weight Adjust proportionally: .

[0138] Step 1032: Determine the mean and standard deviation for all first confidence levels.

[0139] In practice, the average value corresponding to all first confidence levels The calculation formula is: ,in, The total number of events identified in the current data collection period; The first confidence score of the i-th identified event. The standard deviation of all first confidence scores. The calculation formula is: .

[0140] Step 1033: Obtain the first standard deviation coefficient and the second standard deviation coefficient.

[0141] In practice, the first and second standard deviation coefficients are used to determine the screening interval. For example, if the first standard deviation coefficient is 1.5 and the second standard deviation coefficient is 0.5: Step 1034: In response to the first confidence level being greater than or equal to the product of the mean minus the standard deviation and the coefficient of the second standard deviation, and the first confidence level being less than the product of the mean plus the standard deviation and the coefficient of the second standard deviation, the first confidence level is adjusted according to the attenuation coefficient, and the adjustment action corresponding to the first confidence level is changed to a request for review action; or.

[0142] In specific implementation, if The cloud only changes the affected area markers, without directly modifying attribute values ​​or geometric boundaries. This is based on the attenuation coefficient. Perform confidence adjustment on the affected model part based on the first confidence level: At the same time, the corresponding adjustment action is changed to a request for review action, and a review task is generated by the cloud. If no verification confirmation is received within a collection cycle, the identification event will be automatically downgraded and archived.

[0143] Step 1035: In response to the first confidence level being greater than or equal to the product of the mean minus the standard deviation and the first standard deviation coefficient, and the first confidence level being less than the product of the mean minus the standard deviation and the second standard deviation coefficient, the identification event corresponding to the first confidence level is filtered out; or.

[0144] In specific implementation, if This identified event will be filtered out. It only records the event reference list and does not affect the model state.

[0145] Step 1036: In response to the first confidence level being less than the product of the mean minus the standard deviation and the first standard deviation coefficient, the identification event corresponding to the first confidence level is filtered out.

[0146] In specific implementation, if The identified event is filtered out, and the reason for rejection is recorded.

[0147] In addition, if If the identified event matches, it is directly uploaded to the queue; if The identified event is then matched against the real-time upload queue. The real-time upload queue requires full verification and binding in the cloud before adoption, while the direct upload queue does not require full verification and binding.

[0148] Furthermore, for the identified events in the direct upload queue and the real-time upload queue, the completeness of the corresponding spatial anchor point is determined: if there is an identified event... If the event is not identified, the event will be filtered out and a "spatial anchor point insufficient" receipt will be returned.

[0149] Step 1037: Filter all the identification events to obtain the pending identification events, and convert the pending identification events into data packets.

[0150] In practice, all identified events are filtered according to the above conditions to obtain pending identified events. These pending identified events are then encapsulated into lightweight geological event packets using a preset data packet structure. .

[0151] in, : Identify event types, which are triplet codes of "source layer + object layer + action layer", such as "advanced drilling - fracture zone - increased probability".

[0152] in, : Identify event value vectors, which carry location data of specific identified events, such as interface coordinates, probability values, and range parameters.

[0153] in, : Identify event space anchor points, which are composed of "mileage station number + local cross-section coordinates or borehole coordinates + borehole direction + borehole depth range".

[0154] in, : Identifies the event time, which is a timestamp accurate to the second.

[0155] in, The identification time data source identifier is composed of "device identity identifier + data collection team + data collection method".

[0156] in, First confidence level.

[0157] in, : Identify event evidence summaries, including thumbnail indexes, fragments of anomalous curves, or structured summaries of geophysical feature values.

[0158] in, : The set of part numbers of candidate recognition events that affect the model.

[0159] in, The adjustment instruction includes the attribute update target value, the set of geometric control points, the confidence adjustment amount, and the review request flag.

[0160] in, Version dependencies, including , representing the dependent model version number and spatial index version number.

[0161] Specifically, the dependent model version number is the version number of the model used to process the data packet, ensuring that results from different model versions can be distinguished and traced. If the model is subsequently upgraded, this version number can be used to determine whether the old data needs to be reprocessed or reinterpreted for compatibility.

[0162] Specifically, the spatial index version number refers to the version number of the model in the cloud. This spatial index version number is updated over time or according to accuracy requirements. This version number ensures that the correct lookup table is used when querying spatial proximity relationships and calculating spatial anchor points, avoiding positioning errors caused by inconsistent index versions.

[0163] The above scheme calculates the first confidence level and then sets a dynamic screening interval based on the statistical mean and standard deviation. High-confidence events are directly adopted, medium-confidence events are attenuated and transferred to manual review, and low-confidence events are directly screened out. The screened events are then encapsulated into structured data packets carrying spatial anchors, version dependencies, and adjustment instructions. This achieves closed-loop management of the entire process of multi-source geological identification events, from credible quantitative assessment and adaptive screening to traceable output. This enables model updates to automatically filter low-quality noise and contradictory signals, accurately extract high-value geological discoveries, and thus significantly reduce the burden of manual review while ensuring the accuracy of model evolution. This significantly improves the real-time calibration efficiency and long-term operation and maintenance stability of 3D geological models in complex tunnel construction scenarios.

[0164] In some embodiments, the predicted object information further includes a prediction confidence level; the data packet includes multiple transmission parameters, and step 104 includes: Step 1041: For each data packet: determine the confidence deviation based on the first confidence level and the predicted confidence level corresponding to the data packet; sum the location deviation, attribute deviation, confidence deviation and their corresponding transformation weights by weight to obtain the model difference value; sum the model difference value, each transmission parameter and its corresponding transmission weight by weight to obtain the transmission priority value.

[0165] In practice, firstly, the confidence bias is obtained by subtracting the first confidence level from the predicted confidence level corresponding to the data packet. Then, obtain the previously calculated geometric position deviation. Attribute probability bias Confidence bias And the corresponding transformation weights, weighted summation, to obtain the model difference value. .

[0166] in, The change in confidence level. , , These are the weights corresponding to geometric position deviation, attribute probability deviation, and confidence deviation, respectively.

[0167] Finally, the multiple transmission parameters, model difference values, and corresponding transmission weights of the data packets are obtained, and then weighted and summed to obtain the transmission priority value. The calculation formula is as follows:

[0168] in, - Geological Risk Level: This is the maximum risk level among all model blocks in the affected model block set. The risk level is quantified as follows: karst = 1.0, fault = 0.8, water-rich = 0.7, fractured = 0.6, intact = 0.3.

[0169] Specifically, a model block refers to the smallest adjustment unit in a three-dimensional geological model. It is obtained by meshing the three-dimensional geological model and is usually represented as a voxel or an octree node.

[0170] in, - The urgency of construction is calculated using the following formula: ,in This is the distance between the current working face mileage and the minimum mileage of the affected model block. The closer you get to the working face, the more urgent the situation becomes.

[0171] in, - Model block sensitivity, calculated using the following formula: , is the highest risk multiplied by the maximum uncertainty in the affected model block. Specifically, For model blocks Geological risk level, It iterates through all model blocks in the set and takes the maximum value. It is the set of model blocks that represent the impact of the i-th identified event. yes The j-th model block in It is a model block The model confidence level.

[0172] in, This refers to the degree of multi-source support.

[0173] in, - Event packet upload length, calculated using the following formula: Specifically, This refers to the length of the header field; It is the length of the main field; It is the original length of the evidence; This is the evidence compression factor.

[0174] Specifically, The calculation formula is: . , is the basic compression coefficient; is the risk sensitivity coefficient; and are the compression boundaries; is the lower limit of the compression coefficient; is the upper limit of the compression coefficient; means to limit within the interval, is a truncation function. If x < a, then take a; if x > b, then take b; otherwise take x.

[0175] Among them, - Version lag penalty: If an event depends on an old version and the current version is , then the calculation formula is: , the absolute difference between the current version and the dependent version . Specifically, is the maximum threshold for allowing version lag.

[0176] In addition, are the transmission weights corresponding to multiple transmission parameters and model difference values respectively, which are adjusted in real time according to the environmental state.

[0177] For example, adjustment close to a geological hazard: The three-dimensional distance from the construction site location to the nearest known major geological anomaly . If , then the adjustment formula is:

[0178]

[0179] Among them, is the influence radius, with a default of 100 meters; is the corresponding sensitivity coefficient.

[0180] In addition, after the weights are increased, normalization is performed to ensure .

[0181] For example, network bandwidth adaptive adjustment: Monitor the current upload bandwidth and the historical average bandwidth ratio. If , then the adjustment formula is: . That is, the lower the bandwidth, the greater the upload length penalty weight (but the upper limit is 3 times), so as to preferentially upload short event packets. Normalization is performed after the increase.

[0182] For example, receipt feedback on the quality of spatial anchor points. If in the most recent event packets there are more than If the ratio receives a "spatial anchor point insufficiency" response, it will be temporarily increased. and The adjustment formula is as follows: When continuous If the event packet does not receive the receipt, it will revert to its original value.

[0183] For example, geophysical event verification feedback: the cumulative number of times such geophysical events are verified as valid by subsequent borehole drilling. The adjustment formula is as follows: .

[0184] Step 1042: Sort all transmission priority values ​​in descending order to obtain a transmission priority value sequence.

[0185] In practice, all data packets in the current acquisition period are collected, transmission priority values ​​are calculated, and they are sorted from largest to smallest to obtain a transmission priority value sequence: .

[0186] Step 1043: The transmission priority value located in the first proportion range of the transmission priority value sequence is taken as the first level, the transmission priority value located in the second proportion range of the transmission priority value sequence is taken as the second level, and the transmission priority value located in the third proportion range of the transmission priority value sequence is taken as the third level.

[0187] In practice, the first percentage range is: The second percentage range is: The third percentage range is: .in, , All are quartiles.

[0188] Step 1044: The data packets corresponding to the first-level transmission priority value are transmitted in real time; or...

[0189] In practice, the data packets corresponding to the first-level transmission priority value are transmitted in real time.

[0190] Step 1045: The data packets corresponding to the second-level transmission priority value are transmitted periodically; or...

[0191] In practice, second-level data packets are collected at certain intervals, and then all the collected second-level data packets are uploaded together after the collection period ends.

[0192] Step 1046: The data packets corresponding to the third-level transmission priority value are transmitted after a preset delay.

[0193] In practice, a delay time is set, and the third-level data packets are uploaded at the corresponding delay time.

[0194] In addition, the number of first-level data packets , The minimum number of data packets included in each priority level, taken from the first priority value in the transmission priority sequence. A number of data packets are classified as first-level; if the number of data packets in third-level category is... After taking the transmission priority value sequence The data packets are classified as Level 3.

[0195] The above scheme integrates the deviation between the first confidence level of the data packet and the model prediction confidence level, and calculates the model difference value by weighting it with geometric position deviation and attribute deviation. It also integrates multi-dimensional transmission parameters and dynamically adjusts the transmission weight based on real-time environmental feedback such as network bandwidth, distance to disaster bodies, and historical feedback. After calculating the transmission priority value, the data packets are arranged in descending order, and the data packets are divided into three levels of differentiated transmission: real-time, periodic, and delayed based on quartile proportions. This enables intelligent priority sorting and adaptive hierarchical uploading of 3D geological model update events under limited network conditions. This allows high-urgency and high-risk model conflicts to be synchronized in real time to support construction safety decisions, while low-value events are reasonably postponed to save bandwidth. Thus, while ensuring the real-time nature of model evolution, it significantly improves the efficiency and reliability of data transmission in field tunnel construction scenarios.

[0196] like Figure 2 As shown, this method is applied in the cloud, where the overall model is stored. The overall model is a three-dimensional geological model corresponding to the entire geological region. The execution process of this method includes: In this context, "cloud" refers to the backend service platform deployed on remote servers or cloud computing centers, including but not limited to: cloud server clusters, distributed storage systems, data management platforms, model computing engines, and geological analysis workstations and visualization terminals deployed in cloud-based operation and maintenance centers. The cloud possesses massive data storage, high-performance parallel computing, and multi-terminal communication capabilities, used to receive data packets uploaded from devices, perform spatial binding and conflict handling, drive local incremental updates of the overall model, and send version increments and event receipts to devices.

[0197] The overall geological area refers to the entire geological space area corresponding to the entire tunnel project, covering the entire construction area from the starting point to the end point, including completed sections, sections under construction, and sections to be constructed.

[0198] The overall model refers to a three-dimensional geological model covering the entire tunnel engineering route, stored in the cloud. It serves as the complete data source for the three-dimensional geological model of the target geological area on the equipment side, and uses voxel or octree data structures to organize the geological attribute information of the entire line. The three-dimensional geological model stored on the equipment side is a data copy of the overall model within the target geological area, and the two are kept synchronized and traced through version numbers.

[0199] Step 201: Obtain at least one data packet transmitted by the device.

[0200] The data packet includes adjustment actions.

[0201] In practice, the cloud acquires data packets uploaded from the device using different transmission methods.

[0202] Step 202: For each data packet: Determine the target anchor point corresponding to the data packet in the overall model, combine the target anchor point with the data packet, and determine the target region corresponding to the data packet in the overall model.

[0203] In practice, for each data packet, the location data it carries is first mapped to the overall model coordinate system using a coordinate transformation matrix to obtain the target anchor point. Then, the basic influence radius is adaptively adjusted based on the data packet's first confidence level to form a second influence radius. Next, using the target anchor point as the center and the second influence radius as the search range, candidate regions are queried in the spatial index of the overall model. The three-dimensional distance from the center of each candidate region to the anchor point is calculated, and candidate regions with a distance not greater than the second influence radius are identified as the target region corresponding to that data packet. Through this spatial positioning and distance filtering, a precise mapping of the data packet from the field coordinates to the model space is achieved, providing an accurate spatial correspondence for the subsequent binding of data packets to target regions.

[0204] Each target region is a model block. The overall model is organized using a voxel or octree data structure. Each voxel or octree node is a model block, which is the smallest spatial storage and update unit in the overall model. It corresponds to a specific location in the real geological space and stores attribute information such as lithology type, water-bearing probability, fracturing probability, and model confidence level at that location.

[0205] Step 203: Bind the data packet to the corresponding target area.

[0206] In practice, data packets are associated with corresponding target areas and bound together.

[0207] Step 204: Construct conflict vectors and determine conflict weights for at least one data packet corresponding to the target area. Based on the adjustment action, conflict vector, and conflict weight corresponding to each data packet, determine the adjustment method corresponding to the target area.

[0208] In practice, a conflict vector is constructed based on all data packets corresponding to the target area. Then, for each data packet, its multiple conflict parameters are multiplied by their corresponding first weights, and the results are weighted and summed to obtain the conflict weight of that data packet. Next, historical conflict records, current model attribute data, and the constructed conflict vectors for the target area are obtained, multiplied by their corresponding second weights, and weighted and summed to determine the dynamic conflict threshold for the target area. Finally, the change in conflict weight for the target area is calculated based on the conflict weights and adjustment directions of each data packet. The change in conflict weight is compared with the dynamic conflict threshold, and according to preset grading rules, all data packets corresponding to the target area are classified into different conflict handling levels to obtain the adjustment methods corresponding to different levels.

[0209] Step 205: Adjust the target area according to the corresponding adjustment method to obtain the adjusted overall model.

[0210] In practice, the target region is adjusted according to the determined adjustment method: if the data packet corresponding to the target region does not contain non-directional actions such as adding, correcting, or freezing updates, the adjustment action is directly executed to update the geometric boundary of the target region in the overall model. If the data packet contains directional actions such as increasing or decreasing probability, differentiated adjustments are performed based on the conflict classification results. After the adjustment is completed, information such as the data packet number, target region number, attribute values ​​before and after the adjustment, and geometric control points involved in this adjustment are written into the version incremental log to form a traceable model update record, completing the local precise incremental update of the overall model.

[0211] The above scheme enables the cloud to obtain at least one data packet transmitted from the device, which includes adjustment actions. For each data packet, a target anchor point is determined within the overall model. The target anchor point is then combined with the data packet to identify the corresponding target region, and the data packet is bound to the target region. However, this bound target region may be associated with multiple data packets simultaneously, and the adjustment directions between the data packets may be inconsistent. Therefore, it is necessary to construct conflict vectors and determine conflict weights for each data packet corresponding to the target region. Then, based on the adjustment actions, conflict vectors, and conflict weights of each data packet, the adjustment method for the target region is comprehensively determined. The target region is then locally adjusted according to this adjustment method. This yields precise and independent incremental adjustment results for each target region in the overall model, facilitating refined updates of the overall model. The entire process is completed automatically, providing great convenience to users.

[0212] In some embodiments, the data packet includes location data and a first confidence level. Step 202 includes: Step 2021: The location data is transformed into the target anchor point corresponding to the overall model through the transformation matrix.

[0213] In practice, location data and a first confidence level are extracted from the data packet. The location data includes tunnel mileage. Local coordinates of cross section Borehole coordinates and hole orientation parameters Then, the transformation matrix between the preset construction coordinate system and the overall model coordinate system is used. The location data is converted into three-dimensional spatial coordinates in the overall model coordinate system to obtain the target anchor point. The conversion formula is: .in, This is the coordinate correction vector, which includes measurement adjustment and construction cumulative error correction.

[0214] Step 2022: Determine the second influence radius based on the first confidence level and the preset first influence radius.

[0215] In practical implementation, the second radius of influence The calculation formula is as follows: .in, The first radius of influence is defined as follows: face exposure = 3 meters, advanced drilling = 5 meters, geophysical interpretation = 15 meters, monitoring = 5 meters, and manual verification = 3 meters. The lower the first confidence level, the larger the second radius of influence should be to increase the range of candidate inspections.

[0216] Step 2023: Using the target anchor point as the center and the second influence radius as the retrieval radius, retrieve the corresponding candidate region in the overall model.

[0217] In practical implementation, the target anchor point Centered on the second radius of influence To retrieve the radius, a query is performed in the spatial index structure of the overall model to retrieve all candidate voxels or octree nodes within that radius, thus obtaining the candidate region.

[0218] Step 2024: Determine the distance from each candidate region to the target anchor point. If the distance is less than the second influence radius, the candidate region corresponding to that distance is the target region.

[0219] In practice, for each candidate region, the spatial Euclidean distance to the target anchor point is calculated. .like If the candidate region falls within the scope of the event's influence, it is determined that the candidate region is the target region corresponding to the data packet in the overall model.

[0220] The above scheme allows the target area to include one or more model blocks, each of which is directly spatially associated with the geological events carried by the data packet, providing a precise spatial positioning basis for subsequently binding the data packet to the target area and performing local incremental updates.

[0221] In some embodiments, step 204 is preceded by: First of all, Each model block in Calculate adaptive matching score : .in, The formula for calculating spatial overlap is: ; For action adaptation, if the adjustment instruction type of the data packet matches the attribute type of the model block, the value is 1.0; otherwise, it is 0.3. For geological topology consistency, if the data packet is consistent with the adjacency relationship and connectivity rules of the geological body, the value is 1.0; for partial conflict, the value is 0.6; and for complete conflict, the value is 0.2. , , , , and These are the corresponding weighting coefficients.

[0222] Secondly, the binding threshold is determined based on the current state and historical characteristics of the model block. : .in, The base threshold for adjustment; Let j be the confidence level of model block j; This represents the number of days since the last update of the model block. , , These are the corresponding weighting coefficients; This is the time decay factor. If it has not been updated for more than 30 days, this value is 1, indicating "severely outdated".

[0223] Specifically, the higher the confidence level of the model block, the lower the binding threshold; the higher the risk level, the lower the binding threshold; and the longer the time since the last update, the higher the binding threshold.

[0224] Furthermore, if The event enters the update queue of that model block; if Insufficient but This model block is a high-risk model block. Mark this model block as an "observation impact block", only record references, and do not update attributes.

[0225] Additionally, if the event action is "increase probability" or "decrease probability" but If the action is not adapted, the cloud will not perform the update, but will instead send an indirect candidate update to the associated attribute.

[0226] The above scheme calculates an adaptive matching score for candidate model blocks based on spatial overlap, semantic fit of actions, and geological topological consistency. Binding thresholds are determined dynamically based on the model block's confidence level, risk level, and the number of days since the last update. Only update events with matching scores exceeding the threshold are allowed to execute. High-risk blocks that do not meet the criteria are converted to observation references, and incompatible probabilistic actions are converted to indirect candidate updates. This achieves adaptive and refined matching and control of data packet update commands and local model states across geometric, semantic, and temporal dimensions. This allows the dynamic update of the 3D geological model to automatically accept highly relevant new discoveries while effectively avoiding erroneous modifications caused by data conflicts, model obsolescence, or action mismatches through threshold adjustment and hierarchical processing, thus ensuring the accuracy of model evolution.

[0227] In some embodiments, prior to step 204, a three-step initial inspection is performed after the data packet arrives at the cloud: The first step is field integrity verification: check that the data packet number is not duplicated, required fields are not empty, and the signature is valid. If a field is missing, return a "Needs to be supplemented" receipt and list the missing items.

[0228] The second step is to check the legality of the target anchor point: verify whether the mileage station number is within the construction section, whether the drilling parameters are within a reasonable range, and whether the coordinate transformation is unusual. If it is not legal, a "Target Anchor Point Illegal" receipt will be returned.

[0229] Step 3: Version compatibility check: If the data package depends on a version Compared to the current version If they match, proceed directly to space binding. If they don't match, calculate the version suitability score. : .in, This is the version decay coefficient. To affect the self-version within the scope of the model block The number of conflict incidents that have occurred since then The confidence level for the current version. Confidence level for the old version.

[0230] Specifically, if the current version is more definitive, it indicates that the old event may be outdated, resulting in a lower applicability score. If the event applies, proceed with the mapping process: after mapping the old version model block number to the corresponding number in the current version, proceed with space binding. If The event enters the version remapping queue, but does not directly enter the model update queue.

[0231] By implementing the above scheme, the system sequentially performs data packet field integrity verification, target anchor point spatial legality check, and version applicability scoring based on the number of conflict events between versions and changes in confidence level. This achieves a progressive filtering of multi-source update events, from data compliance and spatial rationality to version timeliness. Only high-quality data packets with complete format, valid anchor points, and compatibility with the current model state can enter the subsequent binding and update process. This effectively shields erroneous updates caused by missing fields, abnormal coordinates, or outdated versions, ensuring the accuracy and stability of the dynamic evolution process of the 3D geological model.

[0232] In some embodiments, the adjustment action includes at least one of the following: adding an action, correcting an action, increasing probability an action, decreasing probability an action, requesting review an action, and freezing update an action, wherein increasing probability and decreasing probability actions include adjusting direction; the data packet also includes multiple conflict parameters, each of which can affect the conflict weight, and step 204 includes: Each of the probability-increasing and probability-decreasing actions carries an adjustment direction indicator, which is determined based on the sign of the attribute bias. If the attribute bias is positive, meaning the actual observed value is greater than the model's predicted value, the adjustment direction is positive, corresponding to an probability-increasing action, indicating that the probability value of the corresponding attribute of the target model block should be adjusted in the direction of increasing. If the attribute bias is negative, meaning the actual observed value is less than the model's predicted value, the adjustment direction is negative, corresponding to a probability-decreasing action, indicating that the probability value of the corresponding attribute of the target model block should be adjusted in the direction of decreasing.

[0233] Step 2041: Construct a conflict vector based on all data packets corresponding to the target area. The conflict vector includes a conflict degree parameter, which is used to characterize the conflict between all adjustment directions corresponding to the target area.

[0234] In practice, a collision vector is constructed based on all data packets corresponding to the target area: .

[0235] in, The parameter for spatial conflict level is calculated using the following formula: Specifically, The set of model blocks affected by data packet a; The set of model blocks affected by data packet b, when the scope of influence of the two events does not overlap at all. When completely overlapping .

[0236] in, The collision level parameter refers to the degree of conflict between data packets a and b, where data packets a and b have the same attribute. If the signs of the adjustment directions are consistent, then If the direction sign is reversed, then Iterate through all attributes and take the maximum value. As the final result. Specifically, It is a very small positive number to prevent the calculation from overflowing due to a zero denominator.

[0237] in, The parameter for version conflict level is calculated using the following formula: , specifically The difference between the data packet time and the current version time. For synchronization tolerance window, This is the maximum tolerance window.

[0238] in, This is a parameter representing the degree of conflict in the time sequence of events. Arrival time later than the event However, if the version dependency is older, then Specifically, The time when data packet b arrives at the cloud. The time when data packet a arrives at the cloud. The time tolerance window is set to 1 hour by default and normalized to... If the chronological order is reasonable, then .

[0239] in, For the parameter indicating the degree of conflict of geological rules, list the rule items that should be checked. Taking the water-rich rule as an example: if two data packages respectively include model blocks... and The probability of water abundance increased to over 0.7, but in the overall model... and If they do not belong to the same connected unit and cannot be connected by faults or fracture zones, then the rule conflict count is incremented by 1. , specifically The number of rules violated; The total number of geological rules activated.

[0240] Step 2042: Sum the conflict parameters and their corresponding first weights to obtain the conflict weights.

[0241] In practice, The formula for calculating conflict weight is: .

[0242] Specifically, For source reliability parameters; For feature sharpness parameters; For the quality parameters of the evidence summary; This is a parameter representing the degree of multi-source support. This is a parameter representing the degree of time lag.

[0243] Each conflict parameter has a dynamic weight coefficient. It is not a fixed value, but changes over time. The calculation formula is: .

[0244] Specifically, This is the initial base value for the coefficient; . Based on the current state The adjustment factor.

[0245] Specifically, The first adjustment factor is: if the historical verification efficiency of this type of event is high. , ;like , .

[0246] Specifically, The second adjustment factor: This represents the total number of times this source has been verified as valid throughout history. This is the time since the most recent verification; Tdecay is the decay period, which is 30 days by default. If the verification is older, the coefficient will decay.

[0247] Specifically, The third adjustment factor: the number of other source events that support the current event direction. , .

[0248] Specifically, Fourth adjustment factor: If a conflict has recently occurred within the scope of the event's impact, , This refers to the number of recent conflicts that have occurred within the area affected by this event.

[0249] Specifically, Fifth adjustment factor: If the source has historically triggered multiple conflicts of geological rules, . This represents the proportion of geological rule conflicts that have historically been triggered by this source.

[0250] Step 2043: Obtain historical conflict data, model attribute data, and conflict vectors corresponding to the target area. Based on the historical conflict data, model attribute data, conflict vectors, and their respective second weights, determine the conflict threshold for each target area.

[0251] In practice, the formula for calculating the conflict threshold is as follows: Value range: , , , , , , .

[0252] in, The geological risk level of this model block; This represents the frequency of collisions that have occurred in the history of this model block; To support the number of multi-source evidence sources in the current data packet direction; Topographic sensitivity reflects the region's susceptibility to geological changes; is a coefficient.

[0253] Step 2044: Filter out data packets that include those for adjusting direction from all data packets corresponding to the target area.

[0254] In practice, all data packets bound to the target area are traversed, and each data packet is read one by one to determine whether the data packet includes an adjustment direction identifier. If it does, the data packet is marked as a data packet that includes an adjustment direction and added to the filtered data packet set. If not, it is determined as a data packet that does not include an adjustment direction and does not enter the process of conflict vector construction and conflict weight calculation. Instead, the corresponding geometry or attribute update is directly performed according to the preset adjustment method.

[0255] For all filtered data packets: Step 2045: Determine the change value of the conflict weight corresponding to the target area based on the conflict weight and adjustment direction of all the filtered data packets.

[0256] In practice, each data packet in the filtered data packet set is traversed, and its adjustment direction is read. The filtered data packets are divided into two categories: one is the positive adjustment direction data packet corresponding to "increasing probability", and the other is the negative adjustment direction data packet corresponding to "decreasing probability".

[0257] Then, the collision weights of the two types of packets are summed separately. Positive weight sum: The collision weights of all packets adjusted in the direction of "increasing probability" are summed. Negative weight sum: The collision weights of all packets adjusted in the direction of "decreasing probability" are summed.

[0258] Finally, the sum of the positive weights is subtracted from the sum of the negative weights, and the absolute value is taken to obtain the change in the conflict weights of the target region.

[0259] Step 2046: In response to the number of packets with a collision level parameter of zero being greater than or equal to the fourth threshold, all packets with a collision level parameter of zero are merged into first-level packets; or...

[0260] In specific implementation, if And the number of parameters where the conflict level parameter is equal to zero. If multiple sources of evidence pointing in the same direction are found to support the current target area, then all data packets with a conflict level parameter of zero are fused together to form a first-level data packet. Data packets with the same adjustment direction are fused using a confidence-weighted Bayesian fusion method.

[0261] Specifically, the probability values ​​of the merged attributes The calculation formula is: .

[0262] in, For prior weights, ; These are the prior values ​​of the current model. Update the attribute direction value given for the i-th data packet.

[0263] Specifically, the formula for calculating the confidence improvement of the fused model blocks is as follows: .

[0264] in, The updated confidence level for model block j; The confidence level of model block j before update. To obtain The smaller of 5.

[0265] Step 2047: In response to a conflict level parameter greater than zero and a conflict weight change value greater than the conflict threshold, select the data packet with the largest conflict weight from the data packets with a conflict level parameter greater than zero as the second-level data packet; or.

[0266] In practice, for data packets within the target area with a direction conflict-conflict degree parameter greater than zero, the data packet with the highest conflict weight is selected as the dominant data packet; simultaneously, the data packet with the second highest conflict weight is selected from the remaining data packets and selected as the substandard data packet. The conflict weight of the dominant data packet is subtracted from the conflict weight of the substandard data packet to obtain the change in conflict weight. If the change in conflict weight is greater than the conflict threshold, it indicates that the evidence strength of the dominant data packet is significantly higher than that of the substandard data packet. The data packet with the highest conflict weight is then selected as the second-level data packet, and the update is performed according to the adjustment direction of the dominant data packet. and Update according to the direction of the dominant event, but limit the update magnitude. If the change in conflict weight exceeds the conflict threshold, reduce the update step size. .when At that time, 63% was taken; when The rate was 97%.

[0267] in, This represents the actual amount of updates performed. The original update amount proposed for the superior data packet; This is a truncation function that ensures the constraint coefficient does not exceed 1, meaning the actual update amount will not exceed the proposed value.

[0268] in, As a limiting factor, when Take 30% at the time; when At that time, 63% was taken; when When 97% is taken; When it approaches 1.

[0269] Step 2048: In response to a conflict level parameter greater than zero and a conflict weight change value greater than the product of the first conflict coefficient and the conflict threshold but less than the conflict threshold, then data packets with a conflict level parameter greater than zero are classified as third-level data packets; or...

[0270] In practical implementation, when data packets within the target area have directional conflicts—the conflict severity parameter is greater than zero, and This indicates that while the evidence strength of the dominant data packet exists, the advantage is not significant enough to fully adopt the dominant direction, nor should it be completely ignored. In this case, all data packets with a conflict level parameter greater than zero are classified as Level 3 data packets and undergo review and adjustment.

[0271] Step 2049: In response to a conflict level parameter being less than the product of the first conflict coefficient and the conflict threshold, data packets whose conflict level parameter is less than the product of the first conflict coefficient and the conflict threshold are designated as Level 4 data packets.

[0272] In practical implementation, when data packets within the target area have directional conflicts—the conflict severity parameter is greater than zero, and This indicates that the difference in evidence strength between the superior and inferior data packets is extremely small, and neither side possesses sufficient credibility to drive model updates. In this case, all data packets with conflict level parameters greater than zero are classified into Level 4 data packets, and observation freezing is performed.

[0273] The above scheme constructs a multi-dimensional conflict vector and dynamically calculates the conflict weight based on the reliability of the source, the efficiency of historical verification, and real-time conflict feedback. Then, it adaptively determines the conflict threshold by combining the regional risk level, conflict frequency, and the amount of multi-source evidence. Finally, the conflict data packets are divided into four levels of differentiated processing according to the conflict degree parameter and the weight change value. This achieves accurate quantification, dynamic arbitration, and hierarchical resolution of directional conflicts between multi-source geological update events. In this way, while effectively integrating consistent evidence to enhance the confidence of the model, it automatically suppresses contradictory updates with insufficient evidence or evenly matched evidence, significantly improving the stability of the dynamic evolution of the three-dimensional geological model and the reliability of decision-making in complex construction environments.

[0274] In some embodiments, step 205 includes: Step 2051: For data packets in the target area that do not include the direction adjustment, the corresponding adjustment method is to directly perform the adjustment action and update the geometry of the overall model; or...

[0275] In practice, for data packets in the target area that do not include direction adjustment, the adjustment action is executed directly. If the adjustment action is a new action, a new geological object is created at the corresponding spatial location in the overall model, and an initial geometric boundary is generated based on the control point set. If the adjustment action is a correction action, the existing geometric boundary of the target model block is locally adjusted based on the control point set, and the geometry is updated using local surface fitting or boundary line correction methods. If the adjustment action is a request for review or a freeze update action, it is processed according to the corresponding preset process and does not trigger a geometric update. After the geometric boundary update is completed, the system records the update content in the version increment, including the modified model block number, the change in geometric control points, a summary of geometric differences before and after the update, and a timestamp, and returns an "adopted" receipt to the device to ensure the traceability of model updates and closed-loop confirmation at the field end.

[0276] Specifically, for the geometric update of lithological interface events, a control point set is generated. Candidate boundaries were then obtained using the local surface fitting method. The formula for the computational geometric update criterion is as follows: .

[0277] in, A comprehensive criterion score for geometric updates; It is a geometric change; This represents the fitting error; For geological and topological consistency; The degree of conflict; These are the weighting coefficients. .like curing The new boundary is the geometric boundary of the model; if it is credible, reasonable, and supported, it will be formally adopted. The generation probability anomalies do not have fixed boundaries; there is some basis for this, but it is not sufficient, and it is temporarily labeled as "soft information." Once the geometric update is frozen and enters the review queue, if the evidence is insufficient or there are obvious conflicts, manual intervention is necessary.

[0278] Step 2052: For data packets in the target area that include orientation adjustment, the corresponding data packet for the target area is a first-level data packet. The corresponding adjustment method is to directly execute the adjustment action of the first-level data packet and update the geometric part and / or attribute part of the overall model; or...

[0279] In practice, if the target area is determined to be a Level 1 data packet, the adjustment action for the Level 1 data packet is executed directly without any amplitude limiting or review. If the adjustment action is to increase or decrease the probability, the system reads the fused attribute update value and writes it into the corresponding attribute field of the target model block, completing the attribute update.

[0280] Simultaneously, if the data packet also carries a set of geometric control points (such as coordinates of lithological interface control points or anomaly boundary control points), the system uses a local surface fitting method to correct the geometric boundaries of the target model block, completing the geometric update. If only attribute adjustments are involved without geometric modifications, only the attribute part is updated. After the update is completed, the system records the execution results in the version increment, including the modified model block number, attribute change amount, geometric difference summary, and timestamp, and returns an "adopted" receipt to the device. The receipt contains an update result summary and version number information, ensuring the traceability of the model update process and closed-loop confirmation at the field end.

[0281] For example, updates to the attribute section include: If, model block Original water abundance probability First confidence level Matching score .

[0282] Step 1: Determine the likelihood probability of the data packets Based on the proportion of watermark area Location importance coefficient and adjacent geophysical anomaly indications calculate:

[0283] Step 2: Calculate the update probability:

[0284] in, For model blocks Updated probability of new abundant water. Substitute:

[0285] Step 3: Limit the single change to no more than 0.25. .

[0286] Step 4: Write the model version increment, recording the original value 0.30, the new value 0.52, the event number, the evidence summary, and the timestamp. Update the model confidence. .

[0287] Step 2053: For a data packet in the target area that includes an adjustment direction, the response to the target area is a second-level data packet. The corresponding adjustment method is to execute the adjustment action of the second-level data packet according to the execution coefficient, and update the geometric part and / or attribute part of the overall model. The execution coefficient is used to limit the execution speed of the adjustment action; or.

[0288] In practice, the response to the target area corresponds to the second-level data packet. If the adjustment action is to increase or decrease the probability, the system adjusts the current value of the target attribute in the direction suggested by the dominant data packet. If it's to increase the probability, the value increases positively; if it's to decrease the probability, the value decreases negatively. If the adjustment action also involves correcting geometric boundaries, the set of geometric control points is scaled proportionally, meaning the geometric offset is multiplied by the execution coefficient, and the model boundary is updated using a local surface fitting method. After the update is complete, the system records the execution result in the version increment and returns a receipt to the device.

[0289] Step 2054: For data packets in the target area that include direction adjustment, the corresponding data packet in the target area is a third-level data packet. The corresponding adjustment method is to periodically execute the adjustment action of the third-level data packet and update the attribute part of the overall model; or...

[0290] In practice, the system responds to the third-level data packet corresponding to the target region. It does not modify any geometric boundaries of the model block corresponding to the target region, but only periodically adjusts the attribute part. first step: .

[0291] Step Two: .

[0292] in, These are the temporarily adjusted attribute probability values; This represents the current attribute probability value of model block j. Symbols for adjusting direction; The proposed change in attribute; This is the weighting ratio factor.

[0293] Step 2055: For data packets in the target area that include direction adjustment, the corresponding target area is a level 4 data packet, and the corresponding adjustment method is not to perform the adjustment action of the level 4 data packet.

[0294] In practice, the response to the target area is a Level 4 data packet. This means that no geometric boundaries of the model block corresponding to the target area are modified, and no attribute probability values ​​are adjusted; all state parameters of the model block remain unchanged. Simultaneously, the system marks this Level 4 data packet as a "frozen observation," recording only the conflict vector, conflict weight, and decision time of the event in the event reference list for subsequent tracing and statistical analysis. Furthermore, the system automatically generates a supplementary evidence task receipt, returning a suggestion to the device to supplement data. This suggestion recommends obtaining more independent evidence through methods such as adding advanced drilling, supplementary geophysical exploration, or manual verification to break the current conflict deadlock.

[0295] The above scheme directly performs geometric updates on data packets without direction adjustments, and adopts a differentiated processing strategy for data packets with direction adjustments based on the conflict classification results—directly updating geometry and attributes for Level 1 same-direction evidence, updating Level 2 dominant evidence with limited amplitude according to the execution coefficient, making only periodic attribute fine-tuning and freezing geometry for Level 3 weak evidence, and completely freezing Level 4 deadlocked evidence and generating supplementary evidence suggestions—achieving refined update control from complete acceptance to limited acceptance and then to waiting for review. This enables the model to efficiently integrate high-confidence multi-source consistent information while effectively preventing erroneous updates caused by insufficient evidence or direction conflicts, significantly improving the accuracy and robustness of the dynamic evolution of the 3D geological model.

[0296] In some embodiments, regarding event one: First, the initial conditions are: Specifically, the overall geological region is a zone of interbedded limestone and argillaceous limestone. The tunnel traverses this entire geological region, and an overall model is established based on this region and stored in the cloud. Within this model, 10-meter mileage segments are organized, and octatree voxels are used to represent lithology, faults, water-bearing probability, and fracturing probability within each segment.

[0297] Specifically, the target geological area is from K23+420 to K23+520, and the corresponding 3D geological model is stored on the device. The current version of the 3D geological model is [version number missing]. The current working face is located at K23+456.

[0298] Specifically, three-dimensional geological model The predicted object information stored in the interval from K23+420 to K23+520 includes: Optional, predicted attribute value: Lithology type is intact limestone, mean water-bearing probability is The average breakage probability is Prediction confidence: Model confidence Environmental parameters of the wireless relay network inside the tunnel at the construction site: historical average bandwidth. Mbps, currently available bandwidth Mbps, network status is normal.

[0299] Secondly, the specific processing procedure: Device-side handling of Event 1: Step 1: When the tunnel face advances to K23+456, on-site personnel use survey equipment (tablet computer and industrial camera) to obtain geological information of the target geological area, including photos of the tunnel face, manually recorded surrounding rock grades and joint development.

[0300] Step 2: Step 2.1: Tag all the data to be processed (photos, cataloging forms) according to the preset data source identifier to obtain the first data with the data source identifier, and then determine... The face of the working face is exposed.

[0301] Step 2.2: Feature extraction is performed on the first data to obtain multiple feature data and classify them. The extracted features include the lithological zone boundary coordinates minus the positional deviation of 3.2 meters and the area percentage of the watermark region. Multiple geological object information was obtained through classification. This geological object information includes location data: mileage K23+456, cross-sectional coordinates (-2.1, 3.5); and multiple attribute values: water-rich probability increment +0.42, fracturing probability increment +0.38; geological object type. Water-rich fractured zone.

[0302] Step 2.3: Obtain the geological object information in the three-dimensional geological model on the device. The corresponding prediction object information: Specifically, the predicted attribute values ​​are: lithology type: intact limestone; water-rich probability: 0.18; fractured probability: 0.22; prediction confidence level: 0.72). The deviation between the geological object information and the predicted object information is determined (location deviation 3.2 meters, attribute deviations +0.42 and +0.38 respectively). Adjustment actions are then determined based on these deviations. Because the location deviation exceeds the interface offset tolerance (1.5 meters) and the attribute deviations are positive and the attribute values ​​are greater than the predicted attribute values, the third condition is met, and adjustment actions are initiated. Increase the probability.

[0303] Step 2.4, based on the deviation between the geological object information and the predicted object information: geometric position deviation. Attribute probability bias Initial confidence bias .

[0304] Then, determine the difference vector: .

[0305] Step 2.5: Determine whether the difference vector exceeds a preset first threshold. : .

[0306] Step 2.6: In response to the difference vector exceeding the first threshold, generate a recognition event based on the data source identifier (face exposure), geological object information, adjustment action (probability-increasing action), and difference vector. .

[0307] Step 3: Step 3.1: Identify that the event includes multiple confidence parameters, each of which affects the first confidence level. The first confidence level is obtained by weighted summation of each confidence parameter and its corresponding confidence weight. .

[0308] The calculation of each confidence parameter is as follows: Specifically, the reliability of the underlying sources revealed at the working face. Equipment calibration status 1.0, Coordinate system calibration accuracy The reliability of the data source .

[0309] Specifically, =0.18, then .

[0310] Specifically, for face events, 4 / 5 required spatial fields are provided: mileage, cross-section x, cross-section y, and calibration point. If the exact face normal value is missing, then... .

[0311] Specifically, there is only one historical event: the weak anomaly of TSP at K23+465. .

[0312] Specifically, Hours, just collected, then .

[0313] Specifically, there is manual review and verification. ,but .

[0314] Specifically, if there are currently no conflict events, then .

[0315] Wherein, the confidence weight corresponding to each confidence parameter is: to : .

[0316] In this approach, each confidence level parameter is linearly weighted with its corresponding confidence level weight: Mapped to the (0,1) interval by the Sigmoid function: .

[0317] Step 3.2: First, determine the average value corresponding to the first confidence level of all recognition events in the current batch. 0.668 and standard deviation .

[0318] Next, obtain the first standard deviation coefficient of 1.5 and the second standard deviation coefficient of 0.5.

[0319] Then, determine , in The interval will be changed from an adjustment action to a request for review action.

[0320] Finally, however, due to the high risk level of this identified event: geological risk level... And manual verification confirmed. If the identified event is matched, it will be uploaded to the queue in real time. It will then be fully verified and bound in the cloud before being adopted.

[0321] Step 3.3: Filter all the identification events to obtain the pending identification events, and convert the pending identification events into data packets. .

[0322] Specifically, identify event types Exposing the working face - a water-rich fracture zone - an action to increase probability.

[0323] Specifically, identify event value vectors .

[0324] Specifically, identifying event space anchor points K23+456, (-2.1, 3.5), face normal (0.98, 0.15, 0.12) .

[0325] Specifically, identifying event time For collection timestamps.

[0326] Specifically, identify the source identifier of the time data. Device ID + Data Acquisition Team Specifically, first confidence level .

[0327] Specifically, the set of numbers that influence the model for candidate event recognition. Pre-binding to to (14 voxels).

[0328] Specifically, adjustment instructions Increase the probability of water abundance Increase the probability of breakage Reduce model confidence .

[0329] Specifically, version dependencies .

[0330] Step 4: Step 4.1, based on the first confidence level corresponding to the data packet. The confidence bias is obtained by comparing the predicted confidence level with the actual confidence level. According to geometric position deviation Attribute probability bias Confidence bias and their corresponding conversion weights ( , , The model difference values ​​are obtained by weighted summation. .

[0331] Step 4.2: Obtain multiple transmission parameters of the data packet.

[0332] Specifically, geological risk level The highest risk level for affected voxels is water abundance = 0.7, but this is not necessarily true when considering muddy interlayers. .

[0333] Specifically, the urgency of construction : (Palm face) .

[0334] Specifically, model block sensitivity : .

[0335] Specifically, the degree of multi-source support .

[0336] Specifically, the length of the event packet upload Evidence compression factor . Estimated size 48KB, normalized. .

[0337] Specifically, version lag penalty (The dependency version is the same as the current one).

[0338] Step 4.3: First, the model discrepancy values, each transmission parameter, and their corresponding transmission weights (initial weights) are analyzed. We then perform a weighted summation to obtain the transmission priority value: .

[0339] Secondly, the three-dimensional distance from the construction site to the nearest known major geological anomaly—the water-rich channel. , rice, This will trigger a weight increase: ; The normalized weights are as follows: Recalculate transmission priority: .

[0340] Step 4.4: Order all transmission priority values ​​in the current batch in descending order. Since the current batch contains only one data packet, boundary correction is applied: the number of first-level data packets is less than the minimum number. The first two are taken as the first level, and the transmission method of this data packet is determined to be the first level (real-time transmission).

[0341] Step 4.5, transfer the data packet The data is transmitted to the cloud in real time, and the overall model in the cloud is adjusted according to the corresponding adjustment action - the probability increase action.

[0342] Furthermore, the cloud-based processing of Event 1: Step 5: Step 5.1: Obtain the data packets transmitted from the device in the cloud. The data packet includes adjustment actions—specifically, actions to increase probability. The cloud processes the data packet. Perform a three-step initial inspection: Step 1: Field integrity verification passed.

[0343] Step 2: Target anchor point legality check passed: K23+456 is within the construction section, and the cross-sectional coordinates are legal.

[0344] Step 3: Version compatibility check passed: If the dependency version is consistent with the current version, proceed directly to space binding.

[0345] Step 5.2: The location data (mileage K23+456, cross-section coordinates (-2.1, 3.5), and face normal) in the data packet are transformed into three-dimensional spatial coordinates in the overall model coordinate system using a transformation matrix to obtain the target anchor point.

[0346] Step 5.3, based on the first confidence level The second influence radius is determined based on the preset first influence radius, where the first influence radius exposed at the working face is 3 meters. .

[0347] Step 5.4: First, with the target anchor point as the center and the second influence radius of 3.50 meters as the retrieval radius, the corresponding candidate regions (14 voxels, 5 meters before and after K23+456) are retrieved in the spatial index of the overall model. The distance from each candidate region to the target anchor point is determined one by one, and the candidate regions with a distance less than the second influence radius are determined as the target regions: 14 voxels.

[0348] Secondly, before binding, an adaptive matching score is calculated for each model block in the target region, in voxel format. For example: Matching score .

[0349] Next, the binding threshold is determined based on the current state and historical features of the model block. .

[0350] at last, Match successful. Enter The update queue. Similarly, check the remaining 13 voxels.

[0351] Step 5.5, transfer the data packet Bind to the corresponding target area.

[0352] Step 6: Step 6.1: First, a collision vector is constructed based on all data packets corresponding to the target area. The system discovered a weak reflection anomaly at K23+465 in historical data, representing the source of geophysical interpretation data for the target area, with a first confidence level... No model geometry update was triggered.

[0353] Secondly, data packets Spatial conflict parameter with the geophysical interpretation data .

[0354] Secondly, since the attribute direction is the same—to increase the probability of water abundance / fragmentation—the parameter for the degree of conflict in attribute directions... .

[0355] Then, the version conflict level parameter, the time sequence conflict level parameter, and the geological rule conflict level parameter are all 0.

[0356] Finally, the conflict vector is: .

[0357] Step 6.2: For each conflict parameter and its corresponding first weight, perform a weighted sum to obtain the conflict weight. (Data packet) The conflict weight for the data is 0.668, while the conflict weight for the geophysical interpretation data is 0.54.

[0358] Step 6.3: Obtain historical conflict data, model attribute data, and conflict vectors corresponding to the target area. Determine the conflict threshold based on these three data points and their respective second weights. .

[0359] Step 6.4: Filter out the data packets and data including those for direction adjustment from all data packets and data corresponding to the target area: This includes actions to increase probability, and geophysical interpretation data containing such actions will be processed further.

[0360] Step 6.5: Determine the change value of the conflict weight based on the conflict weight and adjustment direction corresponding to the filtered data packets.

[0361] Step 6.6, Conflict Level Parameter The number of data packets is 2 (data packets) The number of geophysical interpretation data If the condition is greater than or equal to the fourth threshold, then the two data packets are merged into a first-level data packet.

[0362] Specifically, enhanced fusion of multiple events in the same direction is performed through confidence-weighted Bayesian fusion: First, obtain the prior weights. The current model prior values .

[0363] Then, calculate the attribute probability values ​​after fusion: .

[0364] Finally, 0.27 was taken as the increment of the water-rich probability after fusion.

[0365] Step 7: Step 7.1: The target area corresponds to the first-level data packet. The corresponding adjustment method is to directly execute the adjustment action of the first-level data packet and update the attribute part of the overall model.

[0366] Specifically, using voxels For example: First, obtain the water abundance probability attribute value before the update. .

[0367] Secondly, based on the watermark area ratio of 0.35, location importance of 0.80, and adjacent geophysical anomaly indication of 0.60, the likelihood probability of the data packet was calculated and determined. .

[0368] Next, calculate the updated probability of abundant water: .

[0369] Then, because the limit for a single change is 0.25, the actual ( Upper limit).

[0370] Then, calculate the updated breakage probability. And update the model confidence. .

[0371] Finally, at this point, the version increment... The system records the attribute changes and confidence level changes of 14 voxels. The cloud sends a "received" confirmation to the field and issues a version increment.

[0372] In some embodiments, regarding event two, based on event one: It should be noted that Event 1 is a new discovery from scratch (the mudstone interlayer revealed at the working face), and its multi-source support comes entirely from historical data. Event 2, on the other hand, relies on Event 1 as a multi-source support source and is another triggering scenario for the "first-level data packet": the fusion of two real-time data packets in the same direction.

[0373] First, the initial conditions are: Specifically, after anomalies were revealed at the tunnel face, a pre-drilling borehole was initiated in the upper left section (K23+456, cross-sectional coordinates (-2.2, 3.6)). The borehole azimuth angle is... ,inclination The designed hole depth is 30 meters.

[0374] Furthermore, anomalies occurred when drilling reached 18.5-22.0 meters: the drilling speed increased from 0.35 m / min to 0.82 m / min, the torque decreased from 12.5 kN·m to 7.8 kN·m, and the turbidity index of the return water increased from 0.1 to 0.75.

[0375] Specifically, the current version of the 3D geological model is Fragmentation probability .

[0376] Secondly, the specific processing procedure: Device-side handling of Incident 2: Step a, at this time, Advanced drilling, Fracturing zone, Increase probability. Difference vector Determine the generation of recognition events .

[0377] Among them, the difference between the fracturing probability attribute value in the current geological object information and the fracturing probability attribute value in the predicted object information. This indicates that a sudden change in drilling speed was observed in the advanced borehole near K23+465 (immediately behind the location of Event 1), which shows that the drilling evidence independently and in the same direction confirms the judgment given in Event 1 that "there is a fracture zone in this area".

[0378] Step b: First, determine the confidence level parameters: Specifically, .

[0379] Specifically, the signal-to-noise ratio (SNR) is 9.0. .

[0380] Specifically, all five required fields for drilling have been provided. : Specifically, High spatial support, then .

[0381] Specifically, Hour, ,but .

[0382] Specifically, there is cross-checking. 0.75, then .

[0383] Specifically, .

[0384] Secondly The linear weighted sum is 0.766, and the first confidence level is determined by the Sigmoid function. .

[0385] Secondly, after screening, Convert into data packets.

[0386] Step c, based on the data packet corresponding , Geological risk level (exist (After upgrade, the maximum risk within the voxel collection). Construction urgency. (Same working face), sensitivity Multi-source support Event packet upload length Wait, and obtain the transmission priority value:

[0387] At this point, the transmission method is set to Level 1, real-time transmission.

[0388] Cloud-based processing of Event Two: Step d: Obtain data packets from the cloud. For data packets Initial screening: Step 1: Field integrity check passes. Step 2: Target anchor point validity check passes. Step 3: Version applicability check passes. If the dependency version is consistent with the current version, proceed directly to space binding.

[0389] Step e: Determine the target anchor point using the transformation matrix, and determine the second radius of influence based on the first confidence level. Candidate regions are searched centered on the target anchor point, and the target region is determined along the 18.5-22.0 meter segment of the borehole axis. The target region is the tubular influence zone. to A total of 13 voxels.

[0390] Furthermore, with For example, matching score .

[0391] Furthermore, binding threshold .

[0392] Furthermore, , through matching.

[0393] Furthermore, the data packet Bind to the corresponding target area.

[0394] Step f: First, data packets After binding with the corresponding target region, a collision vector is constructed based on all data packets corresponding to the target region: Discovery in the cloud and exist to The attributes above are completely consistent, both aiming to increase the probability of breakage and abundant water. , , , .

[0395] Secondly, the conflict weight is obtained by weighting and summing each conflict parameter with its corresponding first weight.

[0396] Secondly, the number of data packets with a conflict level parameter of zero is 2, which is greater than or equal to the fourth threshold. This condition is met, so these two data packets are merged into a first-level data packet. Then, a same-direction enhanced fusion is performed through confidence-weighted Bayesian fusion.

[0397] Finally, the relative prior increment is 0.118, while Based primarily on the indication of the fracture zone, and indirectly inferred from the indication of abundant water, the value is set to 0.10 after modulation.

[0398] Step g: The target area corresponds to the first-level data packet. The corresponding adjustment method is to directly execute the adjustment action of the first-level data packet and update the attribute part of the overall model.

[0399] First, the probability of abundant water is updated: .

[0400] Secondly, the breakage probability is updated: .

[0401] Furthermore, the model confidence is improved due to independent support from multiple sources: .

[0402] Based on the same inventive concept, corresponding to any of the above-described embodiments, this application also provides a data processing apparatus.

[0403] refer to Figure 3 The device is applied to the equipment end, which stores a three-dimensional geological model of the target geological area, including: The acquisition module 301 is configured to acquire geological information of the target geological area; The identification module 302 is configured to perform attribute identification on geological information and determine multiple identification events; The filtering module 303 is configured to determine the first confidence level of each identification event, filter all identification events according to the first confidence level to obtain undetermined identification events, and convert the undetermined identification events into data packets, wherein the data packets include adjustment actions; The determination module 304 is configured to determine the difference value between each data packet and the three-dimensional geological model, and determine the transmission mode corresponding to each data packet based on each data packet and the corresponding difference value. The transmission module 305 is configured to transmit each data packet to the cloud according to the corresponding transmission method, and adjust the overall model located in the cloud according to the adjustment action corresponding to the data packet to obtain the adjusted overall model.

[0404] In some embodiments, the geological information includes multiple data to be processed; the geological information is subjected to attribute identification to determine multiple identification events. The identification module 302 is specifically configured to: All the data to be processed are tagged according to the preset data source identifier, resulting in multiple first data with data source identifiers; Feature extraction is performed on each first data point to obtain multiple feature data points. All feature data points are then classified to obtain multiple geological object information corresponding to each first data point. Information for each geological object: Obtain the corresponding predicted object information in the three-dimensional geological model from the geological object information, determine the deviation between the geological object information and the predicted object information, and determine the adjustment actions; Determine the difference vector based on the deviation between geological object information and predicted object information; Determine whether the difference vector exceeds the first threshold; In response to the difference vector exceeding the first threshold, an identification event is generated based on the data source identifier, geological object information, adjustment action, and difference vector.

[0405] In some embodiments, geological object information includes location data and multiple attribute values; adjustment actions include at least one of the following: adding an action, correcting an action, increasing probability an action, decreasing probability an action, requesting review an action, and freezing update an action; the identification module 302 is specifically configured as follows: Based on all the feature data corresponding to the geological object information, determine the second confidence level corresponding to the geological object information; Based on the location data, obtain the anchor point coordinates, determine whether the 3D geological model includes the anchor point coordinates, and obtain the determination result. If the anchor point coordinates are not included in the 3D geological model, the judgment result is negative, and the second confidence level is greater than the second threshold, indicating that the first condition is met, then the adjustment action is changed to a new action; or... If the anchor point coordinates are included in the 3D geological model, and the determination result is yes, then proceed with the following steps: In a three-dimensional geological model, the information of the predicted object corresponding to the anchor point coordinates is obtained. The information of the predicted object includes the predicted location data and multiple predicted attribute values. The location deviation is determined based on the location data and the predicted location data, and the attribute deviation is determined based on multiple attribute values ​​and multiple predicted attribute values. If the positional deviation and / or attribute deviation are less than the third threshold, and the second condition is determined to be met, then the adjustment action is a corrective action; or, If the attribute deviation is positive and the attribute value is greater than the predicted attribute value, and the third condition is met, then the adjustment action is to increase the probability; or... If the attribute bias is negative and the attribute value is less than the predicted attribute value, and the fourth condition is met, then the adjustment action is to reduce the probability; or... In response to the following: the historical adjustment actions corresponding to the anchor point coordinates in the 3D geological model have different adjustment directions than the probability-increasing actions, and / or the historical adjustment actions corresponding to the anchor point coordinates in the 3D geological model have different adjustment directions than the probability-decreasing actions, and if the fifth condition is met, then the adjustment action is a request for review; or, If any of the first, second, third, fourth, or fifth conditions are not met, the adjustment action is a freeze update action.

[0406] In some embodiments, the identified event includes multiple confidence parameters, each of which can affect the first confidence level. The filtering module 303 is specifically configured as follows: For each identified event: the first confidence level is obtained by weighted summation of each confidence parameter and its corresponding confidence weight; Determine the mean and standard deviation for all first confidence levels; Obtain the coefficient of the first standard deviation and the coefficient of the second standard deviation; If the first confidence level is greater than or equal to the product of the mean minus the standard deviation and the coefficient of the second standard deviation, and the first confidence level is less than the product of the mean plus the standard deviation and the coefficient of the second standard deviation, then the first confidence level is adjusted according to the attenuation coefficient, and the adjustment action corresponding to the first confidence level is changed to a request for review action; or, If the first confidence level is greater than or equal to the product of the mean minus the standard deviation and the coefficient of the first standard deviation, and the first confidence level is less than the product of the mean minus the standard deviation and the coefficient of the second standard deviation, then the identified event corresponding to the first confidence level is filtered out; or, If the first confidence level is less than the product of the mean minus the standard deviation and the coefficient of the first standard deviation, then the identification event corresponding to the first confidence level is filtered out. Filter all identified events to obtain pending identified events, and convert the pending identified events into data packets.

[0407] In some embodiments, the prediction object information further includes prediction confidence; the data packet includes multiple transmission parameters, and the determination module 304 is specifically configured as follows: For each data packet: determine the confidence deviation based on the first confidence level and the predicted confidence level corresponding to the data packet; sum the location deviation, attribute deviation, confidence deviation and their corresponding transformation weights by weight to obtain the model difference value; sum the model difference value, each transmission parameter and its corresponding transmission weight by weight to obtain the transmission priority value. All transmission priority values ​​are sorted in descending order to obtain a transmission priority value sequence; The transmission priority values ​​that fall within the first proportion range of the transmission priority value sequence are designated as the first level, the transmission priority values ​​that fall within the second proportion range of the transmission priority value sequence are designated as the second level, and the transmission priority values ​​that fall within the third proportion range of the transmission priority value sequence are designated as the third level. Data packets belonging to the first-level transmission priority value are transmitted in real time; or, Data packets belonging to the second-level transmission priority value are transmitted periodically; or, Data packets belonging to the third-level transmission priority value are transmitted after a preset delay.

[0408] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0409] The apparatus of the above embodiments is used to implement the corresponding method in any of the foregoing embodiments and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0410] refer to Figure 4 The device is applied in the cloud, where the overall model is stored. The overall model is a three-dimensional geological model corresponding to the entire geological region, including: The second acquisition module 401 is configured to acquire at least one data packet transmitted by the device, wherein the data packet includes an adjustment action; The first determining module 402 is configured to, for each data packet: determine the target anchor point corresponding to the data packet in the overall model, combine the target anchor point with the data packet, and determine the target area corresponding to the data packet in the overall model; Binding module 403 is configured to bind data packets to corresponding target areas; The second determining module 404 is configured to construct conflict vectors and determine conflict weights for at least one data packet corresponding to the target area, and determine the adjustment method corresponding to the target area based on the adjustment action, conflict vector and conflict weight corresponding to each data packet. The adjustment module 405 is configured to adjust the target area according to the corresponding adjustment method to obtain the adjusted overall model.

[0411] In some embodiments, the data packet includes location data and a first confidence level, and the first determining module 402 is specifically configured to: The location data is transformed into target anchor points corresponding to the overall model using a transformation matrix; The second influence radius is determined based on the first confidence level and the preset first influence radius; Using the target anchor point as the center and the second influence radius as the retrieval radius, the corresponding candidate region is retrieved in the overall model; Determine the distance from each candidate region to the target anchor point. If the distance is less than the second influence radius, the candidate region corresponding to that distance is the target region.

[0412] In some embodiments, the adjustment action includes at least one of the following: adding an action, correcting an action, increasing probability an action, decreasing probability an action, requesting review an action, and freezing update an action, wherein increasing probability and decreasing probability actions include adjusting direction; the data packet also includes multiple conflict parameters, each of which can affect the conflict weight, and the second determining module 404 is specifically configured as follows: Based on all the data packets corresponding to the target area, a conflict vector is constructed. The conflict vector includes a conflict degree parameter, which is used to characterize the conflict between all the adjustment directions corresponding to the target area. The conflict weight is obtained by summing the conflict parameters and their corresponding first weights. Obtain historical conflict data, model attribute data, and conflict vectors corresponding to the target area. Based on the historical conflict data, model attribute data, conflict vectors, and their respective second weights, determine the conflict threshold for each target area. Filter out data packets that include those indicating orientation adjustment from all data packets corresponding to the target area; For all filtered data packets: Based on the conflict weights and adjustment directions of all the filtered data packets, determine the change value of the conflict weights for the target area. If the number of packets with a collision level parameter of zero is greater than or equal to the fourth threshold, then all packets with a collision level parameter of zero are merged into first-level packets; or, If the conflict severity parameter is greater than zero and the change in conflict weight is greater than the conflict threshold, then the data packet with the highest conflict weight is selected from the data packets with a conflict severity parameter greater than zero as the second-level data packet; or, If the conflict level parameter is greater than zero, and the change in conflict weight is greater than the product of the first conflict coefficient and the conflict threshold but less than the conflict threshold, then the data packet with the conflict level parameter greater than zero is classified as a level 3 data packet; or, If the collision level parameter is less than the product of the first collision coefficient and the collision threshold, then the data packet whose collision level parameter is less than the product of the first collision coefficient and the collision threshold is regarded as the fourth level data packet.

[0413] In some embodiments, the adjustment module 405 is specifically configured as follows: For data packets in the target region that do not include orientation adjustment, the corresponding adjustment method is to directly perform the adjustment action and update the geometry of the overall model; or, For data packets that include orientation adjustment within the target region, the response to the target region is a first-level data packet. The corresponding adjustment method is to directly execute the adjustment action of the first-level data packet, updating the geometry and / or attribute parts of the overall model; or, For a data packet in the target region that includes orientation adjustment, the response to the target region is a second-level data packet. The corresponding adjustment method is to execute the adjustment action of the second-level data packet according to the execution coefficient, updating the geometric and / or attribute parts of the overall model. The execution coefficient is used to limit the execution speed of the adjustment action; or, For data packets in the target region that include orientation adjustments, the response to the target region is a Level 3 data packet. The corresponding adjustment method is to periodically execute the Level 3 data packet adjustment action and update the attribute part of the overall model; or, For a data packet in the target area that includes a direction adjustment, the response to the target area is a Level 4 data packet, and the corresponding adjustment method is to not perform the adjustment action of the Level 4 data packet.

[0414] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0415] The apparatus of the above embodiments is used to implement the corresponding method in any of the foregoing embodiments and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0416] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the methods described in any of the above embodiments.

[0417] Figure 5 This embodiment illustrates a more specific hardware structure of an electronic device. The device may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0418] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0419] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0420] The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components within the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touchscreens, microphones, various sensors, etc., while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0421] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0422] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0423] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0424] The electronic devices described above are used to implement the corresponding methods in any of the foregoing embodiments and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0425] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides a non-transitory computer-readable storage medium that stores computer instructions for causing the computer to perform the methods described in any of the above embodiments.

[0426] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a determined device.

[0427] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to perform the methods described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0428] Based on the same concept, corresponding to any of the above embodiments, this application also provides a computer program product, including computer program instructions, which, when run on a computer, cause the computer to perform the method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0429] It is understood that before using the technical solutions of the various embodiments in this disclosure, users will be informed of the type, scope of use, and usage scenarios of the personal information involved in an appropriate manner, and user authorization will be obtained.

[0430] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose, based on the prompt message, whether to provide personal information to the software or hardware such as electronic devices, applications, servers, or storage media performing the operations of this disclosed technical solution.

[0431] As an optional but not limited implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0432] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0433] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application is limited to these examples; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0434] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0435] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0436] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the claims of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. A method of processing data, characterized by, Applied to a device that stores a three-dimensional geological model of the target geological area, the method includes: Obtain geological information of the target geological area; The geological information is subjected to attribute identification to determine multiple identification events; A first confidence level is determined for each of the identified events. All the identified events are filtered according to the first confidence level to obtain undetermined identified events. The undetermined identified events are converted into data packets, wherein the data packets include adjustment actions. Determine the difference value between each data packet and the three-dimensional geological model, and determine the transmission mode corresponding to each data packet based on each data packet and the corresponding difference value; Each data packet is transmitted to the cloud according to the corresponding transmission method. The overall model located in the cloud is adjusted according to the adjustment action corresponding to the data packet to obtain the adjusted overall model.

2. The method of claim 1, wherein, The geological information includes multiple data points to be processed; The geological information is subjected to attribute identification, and multiple identification events are determined, including: All the data to be processed are tagged according to the preset data source identifier to obtain multiple first data with the data source identifier; Feature extraction is performed on each first data point to obtain multiple feature data points. All the feature data points are then classified to obtain multiple geological object information corresponding to each first data point. Information for each of the geological objects: Obtain the predicted object information corresponding to the geological object information in the three-dimensional geological model, determine the deviation between the geological object information and the predicted object information, and determine the adjustment action; Based on the deviation between the geological object information and the predicted object information, a difference vector is determined; Determine whether the difference vector exceeds a first threshold; In response to the difference vector exceeding the first threshold, the identification event is generated based on the data source identifier, the geological object information, the adjustment action, and the difference vector.

3. The method according to claim 2, characterized in that, The geological object information includes location data and multiple attribute values; The adjustment action includes at least one of the following: adding an action, correcting an action, increasing the probability action, decreasing the probability action, requesting a review action, or freezing the update action; The steps of obtaining the predicted object information corresponding to the geological object information in the three-dimensional geological model, determining the deviation between the geological object information and the predicted object information, and determining the adjustment action include: Based on all the feature data corresponding to the geological object information, determine the second confidence level corresponding to the geological object information; Based on the location data, obtain the anchor point coordinates, determine whether the three-dimensional geological model includes the anchor point coordinates, and obtain the determination result. If the anchor point coordinates are not included in the three-dimensional geological model, the judgment result is negative, and the second confidence level is greater than the second threshold, and the first condition is satisfied, then the adjustment action is the new action; or... In response to the inclusion of the anchor point coordinates in the three-dimensional geological model, if the determination result is yes, then the following processing is performed: In the three-dimensional geological model, the predicted object information corresponding to the anchor point coordinates is obtained, wherein the predicted object information includes predicted location data and multiple predicted attribute values; The location deviation is determined based on the location data and the predicted location data, and the attribute deviation is determined based on the multiple attribute values ​​and the multiple predicted attribute values. If, in response to the position deviation and / or the attribute deviation being less than a third threshold, it is determined that the second condition is met, then the adjustment action is the correction action; or, If the attribute deviation is positive and the attribute value is greater than the predicted attribute value, and the third condition is satisfied, then the adjustment action is the probability-increasing action; or... If, in response to the attribute deviation being negative and the attribute value being less than the predicted attribute value, the fourth condition is satisfied, then the adjustment action is the probability reduction action; or... In response to the following: the historical adjustment action corresponding to the anchor point coordinates in the 3D geological model has a different adjustment direction than the adjustment direction of the probability-increasing action, and / or the historical adjustment action corresponding to the anchor point coordinates in the 3D geological model has a different adjustment direction than the adjustment direction of the probability-decreasing action, and it is determined that the fifth condition is met, then the adjustment action is the request for review action; or, If any one of the first condition, the second condition, the third condition, the fourth condition, and the fifth condition is not met, then the adjustment action is the freeze update action.

4. The method according to claim 3, characterized in that, The identified event includes multiple confidence parameters, each of which can affect the first confidence level; The process of determining a first confidence level for each identified event, filtering all identified events according to the first confidence level to obtain undetermined identified events, and converting the undetermined identified events into data packets includes: For each of the identified events: the first confidence level is obtained by weighted summation of each confidence parameter and the confidence weight corresponding to each confidence parameter; Determine the mean and standard deviation for all of the first confidence levels; Obtain the coefficient of the first standard deviation and the coefficient of the second standard deviation; In response to the first confidence level being greater than or equal to the product of the average value minus the standard deviation and the second standard deviation coefficient, and the first confidence level being less than the product of the average value plus the standard deviation and the second standard deviation coefficient, the first confidence level is adjusted according to the attenuation coefficient, and the adjustment action corresponding to the first confidence level is changed to the request for review action; or... In response to a first confidence level being greater than or equal to the product of the average value minus the standard deviation and the coefficient of the first standard deviation, and a first confidence level being less than the product of the average value minus the standard deviation and the coefficient of the second standard deviation, the identified event corresponding to the first confidence level is filtered out; or, In response to the first confidence level being less than the product of the average value minus the standard deviation and the coefficient of the first standard deviation, the identification event corresponding to the first confidence level is filtered out; Filter all the identified events to obtain the pending identified events, and convert the pending identified events into data packets.

5. The method according to claim 4, characterized in that, The prediction object information also includes prediction confidence level; The data packet includes multiple transmission parameters; The step of determining the difference value between each data packet and the three-dimensional geological model, and determining the transmission mode corresponding to each data packet based on the difference value, includes: For each data packet: determine the confidence deviation based on the first confidence level and the predicted confidence level corresponding to the data packet; sum the location deviation, the attribute deviation, the confidence deviation, and their respective corresponding transformation weights to obtain the model difference value; sum the model difference value, each transmission parameter, and their respective corresponding transmission weights to obtain the transmission priority value. All the transmission priority values ​​are sorted in descending order to obtain a transmission priority value sequence; The transmission priority value located in the first proportion range of the transmission priority value sequence is taken as the first level, the transmission priority value located in the second proportion range of the transmission priority value sequence is taken as the second level, and the transmission priority value located in the third proportion range of the transmission priority value sequence is taken as the third level. The data packet corresponding to the transmission priority value belonging to the first level is transmitted in real time; or, The data packets corresponding to the transmission priority value belonging to the second level are transmitted in a periodic manner; or, The data packet corresponding to the transmission priority value of the third level is transmitted after a preset delay.

6. A data processing method, characterized in that, The method is applied to the cloud, whereby the cloud stores the overall model, which is a three-dimensional geological model corresponding to the entire geological region. Acquire at least one data packet transmitted from the device, wherein the data packet includes an adjustment action; For each data packet: determine the target anchor point corresponding to the data packet in the overall model, combine the target anchor point with the data packet, and determine the target region corresponding to the data packet in the overall model; Bind the data packet to the corresponding target area; For each of the at least one data packets corresponding to the target area, a conflict vector is constructed and a conflict weight is determined. Based on the adjustment action corresponding to each data packet, the conflict vector, and the conflict weight, the adjustment method corresponding to the target area is determined. The target region is adjusted according to the corresponding adjustment method to obtain the adjusted overall model.

7. The method according to claim 6, characterized in that, The data packet includes location data and a first confidence level; The step of determining the target anchor point corresponding to the data packet in the overall model, and combining the target anchor point with the data packet to determine the target region corresponding to the data packet in the overall model includes: The location data is transformed into the target anchor point corresponding to the overall model using a transformation matrix; The second influence radius is determined based on the first confidence level and the preset first influence radius; Using the target anchor point as the center and the second influence radius as the retrieval radius, the corresponding candidate region is retrieved in the overall model; The distance from each candidate region to the target anchor point is determined, and in response to the distance being less than the second influence radius, the candidate region corresponding to the distance is the target region.

8. The method according to claim 7, characterized in that, The adjustment action includes at least one of the following: adding an action, correcting an action, increasing the probability action, decreasing the probability action, requesting a review action, and freezing the update action, wherein the increasing probability action and the decreasing probability action include adjusting the direction; The data packet also includes multiple conflict parameters, each of which can affect the conflict weight; The step of constructing conflict vectors and determining conflict weights for at least one data packet corresponding to the target region, and determining the adjustment method for the target region based on the adjustment action corresponding to each data packet, the conflict vector, and the conflict weight, includes: Based on all the data packets corresponding to the target area, a conflict vector is constructed. The conflict vector includes a conflict degree parameter, which is used to characterize the conflict between all the adjustment directions corresponding to the target area. The conflict weight is obtained by weighted summation of each conflict parameter and its corresponding first weight; Obtain historical conflict data, model attribute data, and conflict vectors corresponding to the target region; and determine the conflict threshold corresponding to each target region based on the historical conflict data, model attribute data, conflict vectors, and their respective second weights. Filter the data packets that include the adjustment direction from all the data packets corresponding to the target area; For all filtered data packets: Based on the conflict weights and adjustment directions corresponding to all the filtered data packets, determine the change value of the conflict weights corresponding to the target area; If the number of packets with a conflict level parameter of zero is greater than or equal to a fourth threshold, then all packets with a conflict level parameter of zero are merged into a first-level packet; or, In response to the conflict level parameter being greater than zero and the change in conflict weight being greater than the conflict threshold, the data packet with the highest conflict weight is selected from the data packets with the conflict level parameter being greater than zero as the second-level data packet; or, In response to the conflict level parameter being greater than zero, and the change in conflict weight being greater than the product of the first conflict coefficient and the conflict threshold but less than the conflict threshold, the data packet with the conflict level parameter greater than zero is designated as a Level 3 data packet; or, In response to the conflict level parameter being less than the product of the first conflict coefficient and the conflict threshold, the data packet whose conflict level parameter is less than the product of the first conflict coefficient and the conflict threshold is designated as a fourth-level data packet.

9. The method according to claim 8, characterized in that, The adjustment of the target area according to the corresponding adjustment method includes: For data packets in the target region that do not include the adjustment direction, the corresponding adjustment method is to directly execute the adjustment action and update the geometry of the overall model; or, For the data packet that includes the adjustment direction in the target region, in response to the target region corresponding to a first-level data packet, the corresponding adjustment method is to directly execute the adjustment action of the first-level data packet, updating the geometric part and / or attribute part of the overall model; or, For the data packet that includes the adjustment direction in the target region, in response to the target region corresponding to a second-level data packet, the corresponding adjustment method is to execute the adjustment action of the second-level data packet according to the execution coefficient, updating the geometric part and / or the attribute part of the overall model, wherein the execution coefficient is used to limit the execution speed of the adjustment action; or, For the data packet that includes the adjustment direction in the target region, in response to the target region corresponding to a third-level data packet, the corresponding adjustment method is to periodically execute the adjustment action of the third-level data packet and update the attribute portion of the overall model; or, For a data packet in the target area that includes the adjustment direction, in response to a fourth-level data packet corresponding to the target area, the corresponding adjustment method is to not perform the adjustment action of the fourth-level data packet.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 9.

Images