Pile foundation pore-forming quality control method under complex karst geology

Abstract

The invention provides a pile foundation pore-forming quality control method under complex karst geology, which comprises the following steps: generating a karst cave distribution feature set and a stratum abnormal point set based on karst geology survey data; then, determining a karst cave development trajectory sequence set, carrying out risk partitioning, and identifying a high-risk karst cave development region set; according to the method, a high-risk hole site set and a high-risk stratum node set are determined, and then a karst geological risk map is constructed. And based on the map, generating a pore-forming process path set and a pore-forming risk thermodynamic diagram, and superposing the pore-forming process path set and the pore-forming risk thermodynamic diagram to the pile foundation construction plan to form a pile foundation pore-forming risk thermodynamic diagram for visually displaying construction risks and guiding pile foundation construction. According to the method, the quality and safety of pile foundation hole forming under the complex karst geological condition can be improved, and the construction risk is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of karst geological pile foundation construction, and more particularly to a pile foundation hole forming quality control method under complex karst geology. Background Art

[0002] Carrying out pile foundation construction under complex karst geological conditions is an extremely challenging task. Karst geology is generally characterized by widespread distribution of caves, complex stratigraphic structure, and frequent changes in rock interface. These characteristics pose numerous risks during the pile foundation drilling process, such as hole collapse, slurry leakage, and drill sticking. Traditional pile foundation construction methods rely primarily on geological survey reports and the experience of construction personnel, but this approach has many limitations. First, geological survey reports can usually only provide limited geological information and cannot accurately reflect the distribution and development of caves. Second, while the experience of construction personnel is important, it is often difficult to accurately judge construction risks when faced with complex karst geological conditions, leading to various problems during the construction process and affecting construction progress and quality.

[0003] There are some problems with the application of existing pile foundation construction technology under karst geological conditions. On the one hand, the traditional hole-making process path planning is relatively simple and does not fully consider the development trajectory of caves and abnormal points in the formation. As a result, the drill bit is prone to encountering caves or abnormal areas in the formation during the construction process, increasing the construction risk. On the other hand, the existing risk assessment method is mainly based on historical data and empirical formulas, lacking the analysis and processing of real-time geological data, making it difficult to accurately predict the risks during the construction process. In addition, when faced with complex karst geological conditions, existing construction equipment lacks effective automatic correction and risk warning functions. Once construction deviations or risks occur, manual intervention is often required, which not only reduces construction efficiency but may also cause safety accidents.

[0004] There are at least the following problems or defects in the existing technology: First, it is impossible to accurately obtain the distribution characteristics and development trajectory of caves under karst geological conditions, which makes it difficult to effectively assess construction risks; second, there is a lack of hole-forming process path planning methods for complex karst geological conditions, and it is easy to encounter caves or abnormal strata areas during construction; third, the existing risk assessment and early warning methods are not accurate and timely enough, and it is difficult to meet the needs of pile foundation construction under complex karst geological conditions; fourth, the existing construction equipment lacks effective automatic correction and risk warning functions. Once deviations or risks occur during the construction process, manual intervention is required, which reduces construction efficiency and increases safety risks. Summary of the Invention

[0005] The present invention provides a method for controlling the quality of pile foundation drilling in complex karst geology, comprising: Based on multi-source karst geological survey data, a three-dimensional geological anomaly model is constructed to characterize the spatial distribution characteristics of caves and stratigraphic anomalies through multi-modal data collaborative optimization processing; Based on the three-dimensional geological anomaly model, dynamically analyzing and reconstructing the development trajectory sequence of the karst cave to form a karst cave development trajectory sequence set containing information on the dissolution zone and rock layer interface; Based on the cave development trajectory sequence set, combined with the cave density in the depth dimension and the distribution characteristics of stratigraphic anomalies, a three-dimensional dynamic risk assessment model is established to identify and divide high-risk cave development areas; Locating and determining a high-risk hole location set based on the spatial distribution characteristics of the high-risk cave development area set and the preset cave development conditions; Based on the high-risk cave development area set, determine the stratigraphic structure nodes within its influence range through spatial correlation analysis to form a high-risk stratigraphic node set; The multi-dimensional karst geological risk map is constructed by integrating the cave distribution feature set, the cave development trajectory sequence set, the high-risk cave development area set, the high-risk well location set, and the high-risk formation node set; Based on the karst geological risk map, an optimized set of drilling process paths is intelligently generated, and based on this, a drilling risk heat map is generated to characterize the spatial distribution of construction risks; The hole formation risk heat map is spatially superimposed and fused with the pile foundation construction plan to generate a pile foundation hole formation risk heat map, which is then visually presented.

[0006] Furthermore, based on multi-source karst geological survey data, a three-dimensional geological anomaly model characterizing the spatial distribution characteristics of caves and stratigraphic anomalies is constructed through multi-modal data collaborative optimization processing, including: Acquire karst geological survey data including borehole coring data, geological radar detection data, and cross-hole CT imaging data; Standardizing the karst geological survey data, unifying the spatial reference and data scale, and obtaining standardized drilling data, radar data, and CT data; Performing spatial registration and data fusion on the standardized drilling data, radar data, and CT data to generate effective geological anomaly data; Extracting and generating a cave distribution feature set based on the effective geological anomaly data; Identify geological anomaly points associated with each cave feature in the cave distribution feature set to generate a stratigraphic anomaly point set.

[0007] Furthermore, the standardized drilling data, radar data and CT data are spatially registered and fused to generate effective geological anomaly data, including: Performing spatial coordinate matching on the standardized drilling data, radar data, and CT data to generate overlapping anomaly data sets and independent anomaly data sets; Performing a three-dimensional data fusion calculation on each overlapping anomaly data group in the overlapping anomaly data group set based on a preset spatial weight allocation strategy to generate a fused geological anomaly data set; The fused geological anomaly data set and the independent anomaly data set are jointly determined as effective geological anomaly data.

[0008] Furthermore, the three-dimensional geological anomaly model is used to dynamically analyze and reconstruct the development trajectory sequence of the cave to form a cave development trajectory sequence set containing information on the dissolution zone and rock layer interface, including: Applying drilling trajectory positioning technology to determine the original spatial development point sequence of each cave in the cave distribution characteristic concentration; Performing interpolation optimization on the original spatial development point sequence under geological constraints to fill in the missing segments of the trajectory caused by sparse survey points, and generating a continuous sequence of completed development points; Performing geological noise filtering on the completed development point sequence to remove interference points caused by measurement errors or local geological disturbances, thereby generating an accurate development point sequence; Based on the precise development point sequence, a cave development trajectory sequence set containing information on the boundary of the dissolution zone and the rock layer interface is constructed through geological law modeling and spatial curve fitting.

[0009] Furthermore, based on the cave development trajectory sequence set, combined with the cave density in the depth dimension and the distribution characteristics of abnormal stratigraphic points, a three-dimensional dynamic risk assessment model is established to identify and divide high-risk cave development area sets, including: Performing layered slicing on the cave development trajectory sequence set according to preset depth intervals to extract the target dissolution zone set corresponding to each depth level; For the target dissolution zone set at each depth level: Perform spatial grid risk zoning to form a cave development risk zone set containing multiple target dissolution zones; Calculate the number of target dissolution zones within the unit spatial volume of each risk zone to obtain the regional cave density; Based on the set of formation anomaly points, counting the number of formation anomaly points associated with each risk zone to obtain the number of regional anomaly points; Combining the density of regional caves and the number of abnormal points in the region, applying preset multi-dimensional high-risk judgment conditions, identifying and marking high-risk cave development areas at the depth level; The high-risk cave development areas at each depth level are summarized to form a set of high-risk cave development areas in three-dimensional space.

[0010] Furthermore, the karst geological survey data further includes a drilling pressure dynamic change sequence, a mud loss sequence, and a core breakage rate sequence; and the method further includes: The drilling pressure dynamic change sequence, mud loss sequence and core breakage rate sequence are defined as the first geological risk time series data; Extracting time series features from the first geological risk time series data to obtain geological risk time series features that characterize the dynamic risk of the drilling process; Based on the geological risk time series characteristics, predict the potential risk probability of hole collapse during the hole drilling process; generating second geological risk spatial data based on the high-risk cave development area set, the high-risk hole location set, and real-time drilling parameters; Performing multi-scale spatial feature extraction on the second geological risk spatial data to obtain multi-scale geological risk features reflecting local and regional risks; Performing multi-scale analysis on the geological risk time series characteristics to obtain the geological risk time series multi-scale characteristics; Performing residual connection on the geological risk time series feature and the geological risk time series multi-scale feature, retaining the original time series information and enhancing the feature expression capability, to obtain the geological risk residual connection feature; fusing the geological risk multi-scale feature with the geological risk residual connection feature to form a fused geological risk feature that comprehensively characterizes the geological risk; Based on the integrated geological risk characteristics, intelligent decisions are made and drilling control instructions are generated, which are transmitted to the pile foundation construction equipment for execution. The drilling control instructions include dynamic drilling speed adjustment instructions, active wall reinforcement instructions, and grouting adaptive compensation instructions.

[0011] Furthermore, the intelligent generation of an optimized drilling process path set based on the karst geological risk map includes: Planning a spatial node sequence for grouting reinforcement based on a set of high-risk stratum nodes in the karst geological risk map; generating a set of avoidance paths for a drill bit according to the spatial morphological characteristics of the cave development trajectory sequence set; The grouting reinforcement node sequence and the drill bit avoidance path set are matched with time-space collaborative optimization to generate a hole-forming process path set.

[0012] Furthermore, after visually presenting the pile foundation hole risk heat map, the method further includes: Real-time collection of drilling sensor data, including drill bit vibration spectrum characteristics, mud density dynamics, and hole inclination angle changes; Calculating a dynamic deviation value between the real-time drilling sensor data and the predicted state of the pile foundation hole risk heat map; generating a hole quality deviation alarm based on the dynamic deviation value; When the hole quality deviation alarm exceeds a preset safety threshold, the closed-loop control correction control protocol is automatically triggered.

[0013] Furthermore, the three-dimensional data fusion calculation based on the preset spatial weight allocation strategy includes: Different spatial weight coefficients are assigned to the borehole data, radar data and CT data in the overlapping anomaly data group, among which the borehole data is assigned the highest weight, followed by the radar data, and the CT data has the lowest weight; Based on the assigned weight coefficients, weighted fusion calculation is performed on the three-dimensional data of the overlapping area to generate fused geological anomaly data.

[0014] Furthermore, an area is determined to be a high-risk karst cave development area if any of the following conditions are met: The regional cave density reaches a first preset high density threshold and the number of regional outliers reaches a first preset high outlier threshold; or the regional cave density reaches a second preset high density threshold and the number of regional outlier points reaches a second preset high outlier point threshold; Or the regional cave density reaches a third preset high density threshold and the number of regional outlier points reaches a third preset high outlier point threshold.

[0015] The above embodiments of the present invention have at least the following beneficial effects: 1. This invention accurately identifies high-risk areas in complex karst geological conditions by generating a set of cave distribution characteristics and a set of stratigraphic anomaly points, determining a sequence set of cave development trajectories, and performing risk zoning. This allows construction personnel to understand the distribution and development of caves in advance, enabling them to take targeted risk prevention measures during pile foundation construction, effectively reducing construction risks such as cave collapse, grout leakage, and drill sticking, and improving construction safety and reliability.

[0016] 2. Based on the set of high-risk karst cave development areas, the present invention identifies a set of high-risk borehole locations and high-risk formation nodes, generates a karst geological risk map, and then generates a set of drilling process paths and a drilling risk heat map. This series of steps optimizes the drilling process path planning, enabling the drill bit to effectively avoid karst caves and abnormal formation areas, reducing obstacles and risks during construction, improving drilling efficiency, and ensuring drilling quality, thereby shortening the construction period and reducing construction costs.

[0017] 3. This method overlays a borehole risk heat map onto a pile foundation construction plan for display. It also captures real-time drilling sensor data during construction, generates a borehole quality deviation alert based on the deviation value, and triggers an automatic deviation correction control protocol when the alert exceeds a preset threshold. This real-time monitoring and automatic deviation correction mechanism promptly detects and corrects deviations during construction, ensuring that the construction process remains under control. This further improves the accuracy and quality of pile foundation boreholes and enhances the stability and reliability of the construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which: Figure 1 A schematic flow chart of a method for controlling pile foundation hole quality in complex karst geology provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0019] The technical solutions of this application will be described clearly and completely below, in conjunction with the accompanying drawings. It should be understood that the described embodiments represent only a portion of the embodiments of this application, and not all of them. The components of this application, generally described and illustrated in the drawings herein, may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but rather merely represents selected embodiments of this application. All other embodiments derived by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. Furthermore, in the description of this application, the terms "first," "second," etc., are used solely to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0020] During traditional pile foundation construction, the three-dimensional spatial distribution characteristics of caves in karst geological conditions are difficult to accurately model. Spatial coordinate deviations exist between borehole core sampling data and geological radar detection data, resulting in inaccurate reconstruction of cave development trajectories. The spatial correlation of stratigraphic anomalies is not effectively quantified, and the demarcation of abnormal area boundaries relies on empirical thresholds, resulting in missed identification of high-risk borehole locations. The drilling process path planning fails to incorporate the dynamic evolution of the dissolution zone, and there is a lack of coordinated optimization between the drill bit trajectory and the changes in the rock interface, leading to the risk of unexpected cave penetration during drilling.

[0021] During implementation, in the construction of limestone layers at a depth of 30 to 50 meters, the traditional method used independent analysis of borehole data and cross-hole CT data, and did not establish a spatial weight fusion mechanism for multi-source data, resulting in an error of more than 15% in the calculation of the cave roof thickness. Stratigraphic anomalies were divided only according to the density threshold, without considering their aggregation characteristics along the direction of the dissolution zone, resulting in hidden caves with a diameter of more than 2 meters not being marked as high-risk areas. The drill path planning was based on two-dimensional plane projection and did not take into account the inclination change of the cave development trajectory, causing the drill bit to get stuck at the interface of the 75-degree inclined dissolution zone, and the mud loss instantly increased to 3 times the normal value. If the above problems are not solved, the error in the spatial distribution model of the cave will lead to insufficient grouting reinforcement range, and the pile foundation bearing layer will not be able to effectively avoid the broken rock mass.

[0022] In this regard, this application proposes a method for controlling the quality of pile foundation drilling in complex karst geology, such as Figure 1 As shown, including: S1. Based on multi-source karst geological survey data, a three-dimensional geological anomaly model is constructed to characterize the spatial distribution characteristics of caves and stratigraphic anomalies through multi-modal data collaborative optimization processing; S2. Based on the three-dimensional geological anomaly model, dynamically analyze and reconstruct the development trajectory sequence of the karst cave to form a karst cave development trajectory sequence set containing information on the dissolution zone and rock layer interface; S3. Based on the cave development trajectory sequence set, combined with the cave density in the depth dimension and the distribution characteristics of abnormal stratigraphic points, a three-dimensional dynamic risk assessment model is established to identify and divide high-risk cave development areas; S4. Locating and determining a high-risk hole location set based on the spatial distribution characteristics of the high-risk cave development area set and the preset cave development conditions; S5. Based on the high-risk cave development area set, determine the stratigraphic structure nodes within its influence range through spatial correlation analysis to form a high-risk stratigraphic node set; S6. Construct a multi-dimensional karst geological risk map by integrating the cave distribution feature set, the cave development trajectory sequence set, the high-risk cave development area set, the high-risk well location set, and the high-risk formation node set; S7. Based on the karst geological risk map, intelligently generate an optimized set of drilling process paths, and accordingly generate a drilling risk heat map representing the spatial distribution of construction risks; S8. Spatially superimpose and fuse the borehole risk heat map with the pile foundation construction plan to generate a pile foundation borehole risk heat map, and visualize the map.

[0023] Constructing a three-dimensional geological anomaly model involves integrating borehole coring data, geological radar detection data, and cross-hole CT imaging data, and employing multimodal data collaborative optimization processing to form a three-dimensional geological anomaly model that characterizes the spatial distribution characteristics of karst caves and stratigraphic anomalies. This can be achieved using methods such as spatial coordinate matching and three-dimensional voxel fusion calculations. Dynamically analyzing and reconstructing the development trajectory sequence of karst caves involves utilizing borehole trajectory positioning technology combined with geological interpolation optimization to restore the dynamic development process of karst caves. This can be achieved through geological noise filtering and the generation of a complete development point sequence, thus overcoming the limitations of static geological analysis in understanding the evolution of karst zones.

[0024] Establishing a three-dimensional dynamic risk assessment model involves identifying and demarcating high-risk cave development zones based on a sequence of cave development trajectories, combined with depth-dimensional cave density and the distribution of stratigraphic anomalies. Comprehensive assessments are conducted using multi-dimensional indicators such as regional cave density and the number of regional anomalies, overcoming the one-sidedness of single-metric risk assessments. A karst geological risk map is a visual topological network formed by spatially correlating cave distribution, development trajectories, hole location risks, and stratigraphic nodes. This can be generated through data fusion and spatial path matching, providing dynamic geological support for process path planning.

[0025] Generating a borehole risk heat map involves intelligently generating an optimized set of borehole process paths based on the karst geological risk map. This process then generates a borehole risk heat map that characterizes the spatial distribution of construction risks. By spatially overlaying and integrating the borehole risk heat map with the pile foundation construction plan, a visualization of construction risks is achieved, providing real-time dynamic warnings and guidance for the construction process.

[0026] As a preferred embodiment, the solution of this application is specifically implemented as follows: First, karst geological survey data, including borehole coring data, geological radar data, and cross-hole CT imaging data, are acquired. These data are standardized to produce standardized borehole, radar, and CT data. During data integration, spatial coordinate matching is performed on the standardized data to generate overlapping anomaly data sets and independent anomaly data sets. The overlapping anomaly data sets are then fused in three dimensions to generate a fused geological anomaly data set. The fused geological anomaly data set and the independent anomaly data sets are collectively identified as valid geological anomaly data. This data is then used to construct a three-dimensional geological anomaly model that characterizes the spatial distribution characteristics of karst caves and stratigraphic anomaly points.

[0027] Secondly, based on borehole trajectory positioning technology, the original spatial development point sequence of each cave in the 3D geological anomaly model was determined. Interpolation optimization was performed on the original spatial development point sequence under geological constraints to complete missing segments of the trajectory due to sparse exploration points, generating a continuous completed development point sequence. The completed development point sequence was further subjected to geological noise filtering to remove interference points caused by measurement errors or local geological disturbances, thereby generating a precise development point sequence. Based on the precise development point sequence, a set of cave development trajectory sequences containing information on dissolution zone boundaries and rock layer interfaces was constructed through geological law modeling and spatial curve fitting.

[0028] Then, based on the cave development trajectory sequence set, combined with the depth-dimensional cave density and stratigraphic anomaly point distribution characteristics, a three-dimensional dynamic risk assessment model was established. The cave development trajectory sequence set was sliced ​​and layered at preset depth intervals to extract the target dissolution zone set corresponding to each depth layer. For each target dissolution zone set at each depth layer, spatial grid risk zoning was performed to form a cave development risk zone set containing multiple target dissolution zones. The number of target dissolution zones per unit spatial volume in each risk zone was calculated to obtain the regional cave density. Based on the stratigraphic anomaly point set, the number of associated stratigraphic anomaly points within each risk zone was counted to obtain the regional anomaly point count. Combining the regional cave density and the number of regional anomaly points, preset multi-dimensional high-risk judgment criteria were applied to identify and mark high-risk cave development areas at that depth layer. The high-risk cave development areas at each depth layer were summarized to form a three-dimensional high-risk cave development zone set.

[0029] Next, based on the spatial distribution characteristics of the high-risk cave development area cluster and the pre-set cave development conditions, the high-risk well location cluster was located and determined. Based on the high-risk cave development area cluster, spatial correlation analysis was performed to identify the stratigraphic structural nodes within its influence range to form a high-risk stratigraphic node cluster.

[0030] Furthermore, a multi-dimensional karst geological risk map is constructed by integrating a set of cave distribution characteristics, a set of cave development trajectory sequences, a set of high-risk cave development areas, a set of high-risk borehole locations, and a set of high-risk stratigraphic nodes. Based on this karst geological risk map, an optimized set of drilling process paths is intelligently generated, and based on this, a drilling risk heat map is generated to characterize the spatial distribution of construction risks.

[0031] Finally, the borehole risk heat map is spatially superimposed and fused with the pile foundation construction plan to generate a pile foundation borehole risk heat map, which is then visualized.

[0032] This application further proposes a method for constructing a three-dimensional geological anomaly model that characterizes the spatial distribution characteristics of caves and stratigraphic anomalies based on multi-source karst geological survey data through multi-modal data collaborative optimization processing, including: Acquire karst geological survey data, including borehole coring data, geological radar data, and cross-hole CT imaging data. Standardize this data to unify the spatial reference and data scale, resulting in standardized borehole data, radar data, and CT data. Standardization can include steps such as coordinate system unification, data format conversion, and scale normalization.

[0033] Spatial registration and data fusion are performed on standardized borehole data, radar data, and CT data to generate valid geological anomaly data. Specifically, spatial coordinate matching is first performed to map data from different sources to the same 3D spatial grid. Data in overlapping areas are then fused using a weighted average method, with a weight of 0.5 for borehole data, 0.3 for radar data, and 0.2 for CT data. For non-overlapping independent data areas, the original data is retained. This method generates valid geological anomaly data.

[0034] Based on valid geological anomaly data, a cave distribution feature set is extracted and generated. Clustering algorithms can be used to identify the spatial distribution patterns of anomalous data points and extract characteristic parameters such as the location, morphology, and volume of the caves. In practice, the DBSCAN algorithm can be used to cluster the anomalous points and obtain the spatial outlines of each cave.

[0035] Identify geological anomalies associated with each cave feature within the cave distribution feature set to generate a stratigraphic anomaly point set. A buffer zone of a certain radius can be set around each cave feature, and anomalies falling within the buffer zone are associated with that cave feature, thus forming a stratigraphic anomaly point group. In implementation, a buffer zone is established centered on the cave boundary, and rock fracture points within a 5-meter radius are extracted as associated anomalies.

[0036] Specifically, the core fragmentation rate in the borehole coring data was converted to a standardized value between 0 and 1, the electromagnetic wave reflection time difference in the geological radar data was converted to a spatial coordinate offset, and the density difference value in the cross-hole CT imaging data was converted into three-dimensional gridded data. During the data integration process, when overlapping anomalies exist at the same spatial location from different data sources, a weighted fusion algorithm is used to generate higher-confidence anomaly data. In practice, when an anomaly is detected simultaneously in the borehole data and the CT data at a depth of 15 meters, that location is marked as a high-confidence anomaly area. Cave distribution characteristics are generated by extracting continuous anomaly areas from the fused data. In practice, anomaly areas with a volume greater than 2 cubic meters and a connectivity index greater than 0.7 are defined as independent caves. Stratigraphic anomaly point groups are generated by analyzing the geological parameters of the cave boundaries. Drill points with a core fragmentation rate exceeding 60% within 3 meters of the cave edge are selected as associated anomalies.

[0037] As a preferred embodiment, the solution of this application is specifically implemented as follows: Acquire karst geological survey data, including borehole coring data, geological radar data, and cross-hole CT imaging data. Standardize this data to obtain standard borehole data, standard radar data, and standard CT data. Standardization may include steps such as coordinate system unification, data format conversion, and scale normalization.

[0038] The standardized data is then integrated. First, spatial coordinate matching is performed to map data from different sources to the same 3D spatial grid. Data from overlapping areas are then fused using a weighted average method, with a weight of 0.5 for borehole data, 0.3 for radar data, and 0.2 for CT data. For areas with non-overlapping independent data, the original data is retained. This method yields valid geological anomaly data.

[0039] More specifically, clustering algorithms can be used to identify the spatial distribution patterns of abnormal data points and extract characteristic parameters such as the location, morphology, and volume of the caves. In practice, the DBSCAN algorithm can be used to cluster the abnormal points and obtain the spatial outlines of each cave.

[0040] Determine the stratigraphic anomaly points associated with each cave feature in the cave distribution feature set to generate a stratigraphic anomaly point set. A buffer zone of a certain radius can be set around each cave feature, and the anomaly points falling within the buffer zone can be associated with the cave feature, thereby forming a stratigraphic anomaly point group.

[0041] This application further proposes a method for spatially registering and fusing standardized borehole data, radar data, and CT data to generate effective geological anomaly data, including: The standardized borehole data, radar data and CT data are spatially matched to generate overlapping anomaly data sets and independent anomaly data sets.

[0042] Spatial coordinate matching uses an iterative closest point algorithm to unify the spatial reference of multi-source data, with the point cloud registration error controlled within ±0.05 meters. Specifically, a unified 3D coordinate system is first established, and then the spatial coordinates of each data collection point are determined using GPS positioning technology. Furthermore, a coordinate transformation algorithm is used to convert the coordinate systems of the different data sources into a unified coordinate system. In implementation, a seven-parameter coordinate transformation method can be used for coordinate system integration.

[0043] For each overlapping anomaly data set, a 3D data fusion calculation was performed based on a preset spatial weighting strategy to generate a fused geological anomaly dataset. During the 3D fusion process, borehole data, radar data, and CT data were assigned weights of 0.5, 0.3, and 0.2, respectively, and fused voxel data were generated using a weighted averaging algorithm. Independent anomaly datasets were screened using a threshold for determining isolated anomalies. Anomalous regions were retained when the signal-to-noise ratio (SNR) of a single data source exceeded 8dB. During the data integration process, the spatial correlation between multi-source data established by spatial coordinate matching synergized with the complementary mechanism of 3D fusion. The high-resolution characteristics of the borehole data combined with the deep-depth detection capabilities of the CT data effectively eliminated multipath interference from the radar data in the shallow layers.

[0044] The fused geological anomaly dataset and the independent anomaly dataset are jointly determined as valid geological anomaly data. This not only retains the highly reliable anomaly information cross-validated by multi-source data, but also does not miss special anomalies detected by a single method, ensuring the integrity and accuracy of the anomaly data.

[0045] As a preferred embodiment, spatial coordinate matching is performed on standard drilling data, standard radar data, and standard CT data. Specifically, a unified three-dimensional coordinate system is first established, and then the spatial coordinates of each data collection point are determined using GPS positioning technology. Furthermore, a coordinate transformation algorithm is used to transform the coordinate systems of the different data sources into a unified coordinate system. In implementation, a seven-parameter coordinate transformation method can be used for coordinate system matching.

[0046] After spatial coordinate matching is completed, overlapping anomaly data sets and independent anomaly data sets are generated. The overlapping anomaly data set contains anomaly data detected simultaneously by multiple survey methods at the same spatial location, while the independent anomaly data set contains anomaly data detected by only a single survey method.

[0047] Each overlapping anomaly data set in the set is fused in three dimensions to generate fused geological anomaly data. A weighted average approach can be used to perform the fusion process, assigning weights of 0.5, 0.3, and 0.2 to the borehole data, radar data, and CT data, respectively. Using a three-dimensional voxel fusion algorithm, information from different data sources is integrated into the same spatial unit, resulting in a fused geological anomaly dataset.

[0048] The fused geological anomaly dataset and the independent anomaly dataset are jointly identified as valid geological anomaly data. This not only retains the highly reliable anomaly information cross-validated by multi-source data, but also does not miss special anomalies detected by a single method, ensuring the integrity and accuracy of the anomaly data.

[0049] This application further proposes a method for dynamically analyzing and reconstructing a cave development trajectory sequence based on the three-dimensional geological anomaly model to form a cave development trajectory sequence set containing information on the dissolution zone and rock layer interface, including: Drilling trajectory positioning technology is applied to determine the original spatial development point sequence of each cave within the cave distribution feature set. This technology can be implemented using multi-source sensor fusion, combining gyroscope azimuth data with borehole inclinometer inclination data to generate three-dimensional spatial coordinate positioning. More specifically, when obtaining the original development points through drilling trajectory positioning, a high-precision inertial navigation unit is used to collect borehole azimuth data, ensuring that the spatial coordinate error of each development point is controlled within ±5 cm.

[0050] The original spatial development point sequence is interpolated and optimized under geological constraints to fill in the missing segments of the trajectory caused by sparse exploration points, generating a continuous sequence of completed development points. Geological interpolation optimization can be implemented based on a geostatistical model, using the Kriging interpolation method to predict the cave development trend between adjacent boreholes, and the interpolation spacing can be set to within a range of 0.5 to 2 meters. When geological interpolation optimization is used to fill in missing points, an interpolation model is established based on the rock stratum dip data revealed by two adjacent boreholes. During implementation, when the rock stratum strikes 30 degrees north-east and the dip is 15 degrees, development points are inserted along this direction at a spacing of 1 meter.

[0051] The completed development point sequence is subjected to geonoise filtering to remove interference points caused by measurement errors or local geological disturbances, thereby generating an accurate development point sequence. Geonoise filtering can be combined with a morphological filtering algorithm. During implementation, a dissolution rate threshold is set. Points with a dissolution volume change rate exceeding 0.01 cubic meters per second between adjacent development points are identified as outliers and removed. During geonoise filtering, a density anomaly threshold is set. Data points with resistivity values ​​exceeding three standard deviations of the surrounding rock are identified as instrument interference and removed.

[0052] Based on the precise development point sequence, a set of karst cave development trajectory sequences containing information on dissolution zone boundaries and rock layer interfaces is constructed through geological law modeling and spatial curve fitting. During the generation of the karst cave development trajectory sequence, the filtered development points can be smoothly connected based on the Bezier curve algorithm to form a continuous trajectory. In the final generated karst cave development trajectory sequence, the dissolution zone set is automatically divided by calculating the dissolution volume change rate between adjacent development points, and the rock layer interface set is determined by identifying resistivity mutation points and core fragmentation rate peak points. The trajectory sequence thus formed can accurately reflect the extension direction of the cave in three-dimensional space. During implementation, a certain cave development trajectory showed that it extended in a northwest-southeast direction and was accompanied by three dissolution zones, providing an accurate basis for subsequent judgment of the collapse risk in the area.

[0053] As a preferred embodiment, the development trajectory sequence of each cave in the cave distribution feature set is determined to obtain a cave development trajectory sequence set. First, based on drilling trajectory positioning technology, the original development point sequence of each cave distribution feature in the cave distribution feature set is determined to obtain the original development point sequence set. During implementation, a magnetic field positioning system is used to track the drilling trajectory in real time, record the drill bit position coordinates, and spatially match the detected cave feature points with the drilling trajectory to generate the original development point sequence.

[0054] Secondly, geological interpolation optimization is performed on each original development point sequence in the original development point sequence set to generate a supplementary development point sequence, resulting in a supplementary development point sequence set. During implementation, the Kriging interpolation method is used to interpolate and estimate the missing areas based on the spatial distribution characteristics of the known development points, generating a continuous supplementary development point sequence.

[0055] Then, each of the completed development point sequences in the set was subjected to geonoise filtering to generate a development point sequence, resulting in a development point sequence set. During implementation, a median filtering algorithm was applied to remove outliers caused by equipment errors or local geological anomalies, while retaining development points that conform to the laws of cave development.

[0056] Finally, the corresponding cave development trajectory sequence for each development point sequence in the development point sequence set is determined, resulting in a cave development trajectory sequence set. Each cave development trajectory sequence in the cave development trajectory sequence set contains a set of dissolution zones and a set of rock formation interfaces. During implementation, a three-dimensional curve fitting algorithm is used to convert the development point sequence into a continuous cave development trajectory, and the spatial distribution of dissolution zones and rock formation interfaces is identified based on geological laws.

[0057] This application further proposes a method for establishing a three-dimensional dynamic risk assessment model based on the cave development trajectory sequence set, combined with the cave density in the depth dimension and the distribution characteristics of abnormal stratigraphic points, to identify and divide high-risk cave development areas, including: The cave development trajectory sequence set is sliced ​​and layered at preset depth intervals to extract the target dissolution zone set corresponding to each depth layer. During implementation, vertical division is performed at intervals of one meter in depth, and dissolution zones at the same depth are grouped into the corresponding target depth dissolution zone set.

[0058] For each target dissolution zone set at each depth level, spatial gridding risk zoning is performed to form a cave development risk zoning set containing multiple target dissolution zones. The dissolution zones at the same depth level are divided into multiple rectangular or hexagonal units based on plane coordinates, and the number of dissolution zones contained in each unit is quantified as the regional cave density.

[0059] Calculate the number of target dissolution zones per unit spatial volume within each risk zone to obtain the regional cave density. Based on the stratigraphic anomaly point set, count the number of stratigraphic anomaly points associated with each risk zone to obtain the regional anomaly point count. Calculating the number of stratigraphic anomaly points requires associating geological anomaly data within the same spatial unit. During implementation, coordinate mapping is used to match the stratigraphic anomaly point set with the cave development risk zone.

[0060] Combining the regional cave density and the number of regional outliers, the pre-set multi-dimensional high-risk determination criteria are applied to identify and mark high-risk cave development areas at that depth level. These determination criteria utilize three fixed threshold combinations: the first threshold is a regional cave density greater than or equal to 5 per cubic meter and a number of stratigraphic outliers greater than or equal to 20; the second threshold is a regional cave density greater than or equal to 3 per cubic meter and a number of stratigraphic outliers greater than or equal to 30; and the third threshold is a regional cave density greater than or equal to 8 per cubic meter and a number of stratigraphic outliers greater than or equal to 10. When the spatial spacing of dissolved zones within a grid cell is detected to be less than a pre-set aggregation threshold, a dynamic grid cell size reduction mechanism is initiated until the spacing between dissolved zones meets the analysis accuracy requirement. If the regional cave density meets the high-risk threshold but the number of stratigraphic outliers does not, the cave connectivity index is introduced for supplementary determination. This cave connectivity index is calculated from the proportion of connected paths between voids using cross-hole CT data. A grid cell with a cave connectivity index greater than or equal to 0.7 is still marked as a high-risk cave development area.

[0061] The high-risk karst cave development areas at each depth level are summarized to form a set of high-risk karst cave development areas in three-dimensional space. During implementation, at a depth of 20 meters, all dissolution zones at this depth are extracted and a plane distribution data set is formed. Then, the dissolution zone of this layer is gridded and partitioned, with each grid unit covering an area of ​​10 square meters. The number of dissolution zones in each unit is counted as the regional karst cave density. At the same time, the data in the concentration of stratigraphic anomaly points are mapped to the corresponding grid units through coordinate matching, and the number of anomaly points in each unit is counted as the number of regional anomaly points. The preset high-risk condition uses a dual-indicator threshold combination. During implementation, when the cave density of a grid unit reaches 8 / cubic meter and the number of anomaly points exceeds 10, it is judged as a high-risk area. By performing this judgment layer by layer, the high-risk areas of each depth layer are finally integrated into a set of high-risk karst cave development areas distributed in three-dimensional space.

[0062] As a preferred embodiment, the solution of this application is specifically implemented as follows: Each dissolution zone corresponding to the same depth in each cave development trajectory sequence in the cave development trajectory sequence set is identified as a target depth dissolution zone, thereby obtaining a target depth dissolution zone set to generate each target depth dissolution zone set. During implementation, corresponding target depth dissolution zone sets are generated for three target depths of 10 meters, 20 meters, and 30 meters.

[0063] For each target depth erosion zone set in each target depth erosion zone set, perform the following steps to generate a target depth high-risk erosion cave development zone set, and obtain each target depth high-risk erosion cave development zone set: First, risk zoning was performed on each target depth erosion zone in the target depth erosion zone set to obtain a cave development risk zone set. Each cave development risk zone in the cave development risk zone set contained multiple target depth erosion zones. During implementation, the target depth erosion zone set was divided into multiple 10m x 10m grids, with each grid serving as a cave development risk zone.

[0064] The number of target depth dissolution zones within each cave development risk zone in the cave development risk zone set is determined as the regional cave density, resulting in a regional cave density set. During implementation, the number of dissolution zones within each 10m x 10m grid is calculated as the regional cave density for that grid.

[0065] Based on the stratigraphic anomaly point set, the number of stratigraphic anomaly points corresponding to each karst cave development risk zone in the karst cave development risk zone set is determined as the regional anomaly point number, resulting in a regional anomaly point number set. During implementation, the number of stratigraphic anomaly points within each 10m x 10m grid is counted as the regional anomaly point number for that grid.

[0066] Based on the regional cave density set and the regional outlier number set, each cave development risk zone within the cave development risk zone set that meets the preset high-risk criteria is identified as a target depth high-risk cave development zone, thereby obtaining a target depth high-risk cave development zone set. During implementation, a grid is identified as a target depth high-risk cave development zone when the regional cave density is greater than or equal to 5 per cubic meter and the number of regional outliers is greater than or equal to 20.

[0067] The high-risk karst cave development area sets at each target depth are identified as high-risk karst cave development area sets. During implementation, the high-risk karst cave development area sets at depths of 10 meters, 20 meters, and 30 meters are combined to form the final high-risk karst cave development area set.

[0068] The present application further proposes to define the dynamic change sequence of drilling pressure, mud loss sequence and core breakage rate sequence contained in the karst geological exploration data as the first geological risk time series data; to extract the time series features of the first geological risk time series data to obtain the geological risk time series features that characterize the dynamic risk of the drilling process; to predict the potential collapse risk probability in the drilling process based on the geological risk time series features; to generate the second geological risk spatial data according to the high-risk cave development area set, the high-risk hole position set and the real-time drilling parameters; to extract the multi-scale spatial features of the second geological risk spatial data to obtain the geological risk multi-scale features that reflect the local and regional risks. characteristics; perform multi-scale analysis on the geological risk time series characteristics to obtain geological risk time series multi-scale characteristics; perform residual connection on the geological risk time series characteristics and the geological risk time series multi-scale characteristics, retain the original time series information and enhance the feature expression capability, and obtain geological risk residual connection characteristics; fuse the geological risk multi-scale characteristics and the geological risk residual connection characteristics to form a fused geological risk characteristic that comprehensively characterizes the geological risk; based on the fused geological risk characteristics, intelligent decision-making and generation of drilling control instructions are transmitted to the pile foundation construction equipment for execution, and the drilling control instructions include drilling speed dynamic adjustment instructions, wall protection active reinforcement instructions and grouting adaptive compensation instructions.

[0069] The dynamic change sequence of drilling pressure is obtained through the pressure sensor at a sampling frequency of once per second, the mud loss sequence is monitored in real time through the flow meter, and the core breakage rate sequence is calculated based on the volume ratio of the broken core in the coring barrel.

[0070] Time series feature extraction uses a sliding window analysis method, with a window size of five or ten seconds, to capture the frequency of parameter fluctuations and the correlation of mutation points. Collapse probability estimation is achieved using a trained LSTM neural network model. The input layer receives the standardized time series feature vector, and the output layer generates a probability value.

[0071] During the generation of the second geological risk data, high-risk cave development zones are spatially matched to the drill bit position coordinates in the real-time drilling parameters. Dynamic data collection is triggered when the drill bit enters a high-risk hole. Multi-scale feature extraction uses convolution kernels of varying sizes to detect local outliers and separate regional trends. A three-by-three convolution kernel is used to capture local details, while a seven-by-seven convolution kernel is used to extract macro-trends. Residual connections use a skip connection structure to add the original time series features to the multi-scale time series features, preserving the high-frequency components of the original signal. Feature fusion employs a channel attention mechanism to weightedly concatenate the multi-scale spatial features with the residual-enhanced time series features.

[0072] The generation of drilling control instructions is based on the comparison of fusion features with preset risk thresholds. When the probability of hole collapse exceeds 70%, the drilling speed adjustment instruction is triggered. The wall reinforcement instruction dynamically adjusts the grouting pressure according to the increasing rate of core breakage rate. The flow parameter of the grouting compensation instruction is linearly related to the mud loss.

[0073] Specifically, during pile drilling, a pressure sensor monitors the axial pressure of the drill pipe in real time. When the drill bit contacts the rock formation at the edge of the cave, a positive pulse signal exceeding 20 kilonewtons appears in the drilling pressure mutation sequence. A mud loss sensor detects a mud loss exceeding 0.5 cubic meters per minute, marking it as a high-risk leakage event. The core breakage rate is calculated using an image recognition algorithm, and a feature extraction process is triggered when the breakage rate exceeds 40%. The time series analysis module identifies the simultaneous increase of three parameters within a five-second window, identifying a composite risk pattern and inputting the time series features into the neural network model.

[0074] After the model outputs the probability of hole collapse, it combines it with the distribution data of high-risk karst cave development zones at the drill bit's location to generate secondary geological risk data containing spatial coordinates. Multi-scale convolutional layers extract local dissolution characteristics within a three-meter radius around the current drilling point and regional leakage trends within a ten-meter radius. Residual connections retain information about sudden fluctuations in the original time series data. The feature fusion module combines spatial multi-scale features with temporal residual features through adaptive weight allocation to form a four-dimensional risk tensor. Based on the relationship between the maximum value and threshold value of each dimension in the tensor, the control instruction generator coordinates the issuance of composite instructions to reduce the drilling speed to 0.5 meters per minute, increase the retaining wall grouting pressure to 2 MPa, and increase the mud compensation flow rate to 0.8 cubic meters per minute, thereby achieving dynamic risk control.

[0075] This application further proposes a method for intelligently generating an optimized set of drilling process paths based on the karst geological risk map, including: Based on the high-risk stratum node set in the karst geological risk map, a spatial node sequence for grouting reinforcement is planned. The grouting reinforcement node sequence can be determined by performing density analysis on the high-risk stratum node set using a spatial clustering algorithm. During implementation, the DBSCAN algorithm is used to mark dense areas with node spacing less than 0.5 meters as reinforcement core areas. Within the high-risk stratum node set in the karst geological risk map, the geological risk quantification results are converted into specific coordinate points for grouting reinforcement operations through spatial coordinate mapping technology. During implementation, areas with a dissolution zone density exceeding 5 per cubic meter are automatically marked as reinforcement nodes.

[0076] Based on the spatial morphological characteristics of the cave trajectory sequence, a drill avoidance path set is generated. This avoidance path set is generated by incorporating the three-dimensional curvature characteristics of the cave trajectory. A circular avoidance path is generated when the curvature radius of the cave trajectory is less than 3 meters, and a linear detour path is generated when the curvature radius is greater than or equal to 3 meters. Obstacle avoidance rules are generated from the cave trajectory sequence using a trajectory parsing algorithm. The boundary points of the cave cavity are converted into a three-dimensional obstacle avoidance surface, and the drill path is constrained to a safety range of 0.5 meters outside of this surface.

[0077] The grouting reinforcement node sequence and the drill avoidance path set are matched through spatiotemporal collaborative optimization to generate a set of drilling process paths. The spatial path matching process utilizes a three-dimensional collision detection algorithm, where the reinforcement radius of the grouting reinforcement node sequence is set to 1.2 times the drill diameter, and the minimum safe distance between the avoidance path and the reinforced area is set to 0.8 meters. During the spatial path matching phase, the grouting reinforcement node sequence and the drill avoidance path set are dynamically coupled through vector superposition in a three-dimensional coordinate system. The spatial coverage of the reinforced area and the obstacle avoidance area of ​​the drill path form a complementary layout, ensuring that the drill bit remains within the protection range of the reinforced area when traversing high-risk formations, while avoiding spatial interference between the drill bit trajectory and unreinforced cave areas. This matching process ensures that the grouting material has reached its designed strength by the time the drill bit reaches the reinforced area by calculating the time difference between the solidification time of the reinforced area and the drill bit's travel speed in real time. In practice, when the drill bit travels at a speed of 0.3 meters per minute, the reinforcement operation must be completed 15 minutes in advance.

[0078] As a preferred embodiment, the solution of this application is specifically implemented as follows: In the high-risk stratum node cluster of the karst geological risk map, the geological risk quantification results are converted into specific coordinate points for grouting reinforcement operations through spatial coordinate mapping technology. During implementation, areas with a dissolution zone density exceeding 5 per cubic meter are automatically marked as reinforcement nodes. The cave development trajectory sequence set generates obstacle avoidance rules through a trajectory parsing algorithm. The boundary point set of the cave cavity is converted into a three-dimensional obstacle avoidance surface, and the drill bit path is constrained to a safety range of 0.5 meters outside the surface. In the spatial path matching stage, the grouting reinforcement node sequence and the drill bit avoidance path set are dynamically coupled through vector superposition in a three-dimensional coordinate system. The spatial coverage of the reinforced area and the obstacle avoidance area of ​​the drill bit path form a complementary layout, ensuring that the drill bit is always within the protection range of the reinforced area when traversing high-risk strata, while avoiding spatial interference between the drill bit trajectory and the unreinforced cave area. The matching process calculates the time difference between the solidification time of the reinforced area and the drill bit travel speed in real time to ensure that the grouting material has reached the designed strength when the drill bit reaches the reinforced area. During implementation, when the drill bit travel speed is 0.3 meters per minute, the reinforcement work must be completed 15 minutes in advance.

[0079] The present application further proposes obtaining real-time drilling sensor data, wherein the real-time drilling sensor data includes the drill bit vibration spectrum characteristics, the dynamic value of the mud density and the change of the hole inclination angle; calculating the dynamic deviation value between the real-time drilling sensor data and the predicted state of the pile foundation hole risk heat map; generating a hole quality deviation alarm based on the dynamic deviation value; and automatically triggering a closed-loop control correction control protocol when the hole quality deviation alarm exceeds a preset safety threshold.

[0080] The sampling frequency of the drill bit vibration spectrum is set to a range of 2000Hz to 5000Hz. Fast Fourier transform is used to extract the energy distribution characteristics in the 0.5kHz to 2kHz frequency range, thereby capturing the impact vibration signal when the drill bit contacts the cave. Mud density is monitored using an online densitometer with an accuracy of ±0.01g / cm³. A cave leakage event is identified when the mud density decreases by more than 5% within 30 seconds. Hole inclination is collected in real time using a dual-axis inclination sensor with an angular resolution of 0.01 degrees. Trajectory correction is triggered when the horizontal offset exceeds 0.3% of the designed trajectory.

[0081] Within the coordinate system of the pile foundation construction plan, real-time drilling sensor data is converted into a three-dimensional spatial coordinate point set and spatially matched to the predicted risk areas in the karst geological risk map. The Euclidean distance deviation between the real-time data points and the thermal map grid cells is calculated. A level 1 alarm is generated when the average deviation over three consecutive sampling periods exceeds 0.25, and automatic correction is triggered when it exceeds 0.3.

[0082] The deviation correction control protocol consists of three parallel execution modules: the drill pressure adjustment module dynamically adjusts the rotational speed based on the vibration spectrum characteristics, controlling the impact vibration energy within the range of 50J to 80J; the mud compensation module increases the mud density to a preset safety value at a rate of 0.5% per second; and the trajectory correction module applies a corrective torque of 0.5kN·m to 1.2kN·m through a hydraulic corrector, restoring the hole angle to within the allowable deviation range within five drilling cycles. This multi-parameter coordinated control mechanism ensures that the actual drilling trajectory dynamically matches the predicted path of the risk heat map, effectively suppressing the accumulation of trajectory deviations caused by cave collapse.

[0083] For example, during the pile foundation drilling construction phase, the drill bit vibration spectrum data is collected in real time through a three-axis acceleration sensor, where the vibration spectrum sampling frequency is set to 2000 Hz. At the same time, a mud density sensor is used to obtain the mud specific gravity value at a sampling rate of twice per second, and the hole inclination angle is monitored in real time through a high-precision inclination sensor with a resolution of 0.1 degrees.

[0084] This sensor data is transmitted to the data processing center via an industrial bus and dynamically compared with a pre-generated risk heat map. Deviations are calculated using a weighted Euclidean distance algorithm, with vibration spectrum deviation weights set to 0.6, mud density deviation weights set to 0.3, and hole angle deviation weights set to 0.1. When the combined deviation exceeds a threshold of 0.85, the system automatically activates the deviation correction control module, which consists of three parallel execution units: the first controls the hydraulic system to adjust the drilling pressure distribution for trajectory correction; the second activates the high-pressure grouting pump to reinforce the wall; and the third uses a frequency converter to adjust the main winch speed to optimize drilling speed.

[0085] The present application further proposes real-time collection of drilling sensor data, wherein the data includes drill bit vibration spectrum characteristics, mud density dynamic value and hole inclination angle change; calculating the dynamic deviation value between the real-time drilling sensor data and the predicted state of the pile foundation hole risk heat map; generating a hole quality deviation alarm based on the dynamic deviation value; and automatically triggering a closed-loop control correction control protocol when the hole quality deviation alarm exceeds a preset safety threshold.

[0086] The sampling frequency of the drill bit vibration spectrum is set to a range of 2000Hz to 5000Hz. Fast Fourier transform is used to extract the energy distribution characteristics in the 0.5kHz to 2kHz frequency range, thereby capturing the impact vibration signal when the drill bit contacts the cave. Mud density is monitored using an online densitometer with an accuracy of ±0.01g / cm³. A cave leakage event is identified when the mud density decreases by more than 5% within 30 seconds. Hole inclination is collected in real time using a dual-axis inclination sensor with an angular resolution of 0.01 degrees. Trajectory correction is triggered when the horizontal offset exceeds 0.3% of the designed trajectory.

[0087] Within the coordinate system of the pile foundation construction plan, real-time drilling sensor data is converted into a three-dimensional spatial coordinate point set and spatially matched to the predicted risk areas in the karst geological risk map. The Euclidean distance deviation between the real-time data points and the thermal map grid cells is calculated. A level 1 alarm is generated when the average deviation over three consecutive sampling periods exceeds 0.25, and automatic correction is triggered when it exceeds 0.3.

[0088] The deviation correction control protocol consists of three parallel execution modules: the drill pressure adjustment module dynamically adjusts the rotational speed based on the vibration spectrum characteristics, controlling the impact vibration energy within the range of 50J to 80J; the mud compensation module increases the mud density to a preset safety value at a rate of 0.5% per second; and the trajectory correction module applies a corrective torque of 0.5kN·m to 1.2kN·m through a hydraulic corrector, restoring the hole angle to within the allowable deviation range within five drilling cycles. This multi-parameter coordinated control mechanism ensures that the actual drilling trajectory dynamically matches the predicted path of the risk heat map, effectively suppressing the accumulation of trajectory deviations caused by cave collapse.

[0089] For example, during the pile foundation drilling construction phase, the drill bit vibration spectrum data is collected in real time through a three-axis acceleration sensor, where the vibration spectrum sampling frequency is set to 2000 Hz. At the same time, a mud density sensor is used to obtain the mud specific gravity value at a sampling rate of twice per second, and the hole inclination angle is monitored in real time through a high-precision inclination sensor with a resolution of 0.1 degrees.

[0090] This sensor data is transmitted to the data processing center via an industrial bus and dynamically compared with a pre-generated risk heat map. Deviations are calculated using a weighted Euclidean distance algorithm, with vibration spectrum deviation weights set to 0.6, mud density deviation weights set to 0.3, and hole angle deviation weights set to 0.1. When the combined deviation exceeds a threshold of 0.85, the system automatically activates the deviation correction control module, which consists of three parallel execution units: the first controls the hydraulic system to adjust the drilling pressure distribution for trajectory correction; the second activates the high-pressure grouting pump to reinforce the wall; and the third uses a frequency converter to adjust the main winch speed to optimize drilling speed.

[0091] This application further proposes a method for three-dimensional data fusion calculation based on a preset spatial weight allocation strategy, including: Different spatial weight coefficients are assigned to the borehole data, radar data, and CT data in the overlapping anomaly data group, with the highest weight assigned to the borehole data, followed by the radar data, and the lowest weight assigned to the CT data. Specifically, after completing the spatial coordinate matching of the standard borehole data, standard radar data, and standard CT data, weight assignment and fusion calculations are performed for each overlapping anomaly data group. The high weight of the borehole data makes it dominate the location of the cave boundary. During implementation, when the borehole shows the existence of a cave at a certain coordinate, even if the radar and CT data do not detect anomalies, the coordinate is still marked as a potential risk area. The second highest weight of the radar data enhances the ability to identify the extension direction of the cave. During implementation, in areas where the borehole spacing exceeds 10 meters, the radar data can supplement the identification of laterally developed dissolution zones. The lowest weight of the CT data is used to correct local anomaly features. During implementation, in areas with dense boreholes, CT data can assist in identifying micro-caves with a diameter of less than 0.5 meters.

[0092] As a preferred embodiment, the borehole data, radar data, and CT data are normalized in the spatial coordinate system, and the spatial sampling rates of the three are unified into a 0.1 m × 0.1 m × 0.1 m voxel grid. In the overlapping anomaly area, the core intensity value of the borehole data, the reflected wave amplitude value of the radar data, and the density inversion value of the CT data are extracted for each voxel unit. A linear weighted algorithm is used to superimpose the borehole data with a weight of 0.5, the radar data with a weight of 0.3, and the CT data with a weight of 0.2 to generate a fused anomaly value. For non-overlapping areas, the original independent anomaly data is retained. An outlier threshold is set during the fusion process. When the weighted calculation result exceeds the threshold, the voxel is determined to be a valid geological anomaly body, and finally a three-dimensional fused geological anomaly model is constructed.

[0093] This application further proposes that a high-risk cave development area shall be determined if any of the following conditions are met: The regional cave density reaches a first preset high density threshold and the number of regional outliers reaches a first preset high outlier threshold; or the regional cave density reaches a second preset high density threshold and the number of regional outlier points reaches a second preset high outlier point threshold; Or the regional cave density reaches a third preset high density threshold and the number of regional outlier points reaches a third preset high outlier point threshold.

[0094] Specifically, during the geological risk zoning process before pile foundation drilling, three-dimensional cave distribution data and stratigraphic anomaly detection results are obtained for the target area. The construction area is then stratified by depth and divided into several cubic units. For each unit, the number of caves per unit volume is counted as the regional cave density. The sum of rock fracture zones, fissure development points, and groundwater seepage points within the unit is also counted as the regional anomaly number.

[0095] For example, in a regional unit with a depth of 15-20 meters, the number of caves measured is 7 / cubic meter and the number of abnormal points is 25. At this time, the first high-risk judgment condition is triggered, the area is automatically marked as a red warning area, and a corresponding grouting reinforcement plan is generated.

[0096] As a preferred embodiment, the solution of this application is specifically implemented as follows: During the geological risk zoning process before pile foundation drilling, three-dimensional cave distribution data and stratigraphic anomaly detection results are obtained for the target area. The construction area is then divided into several cubic units by depth. For each unit, the number of caves per unit volume is counted as the regional cave density. The sum of rock fracture zones, fracture development points, and groundwater seepage points within the unit is also counted as the regional anomaly number.

[0097] The three preset threshold conditions are used for judgment. When a unit meets one of the following conditions, the unit is marked as a high-risk cave development area: The regional cave density is greater than or equal to 5 / cubic meter and the number of regional anomaly points is greater than or equal to 20; Or the regional cave density is greater than or equal to 3 / cubic meter and the number of regional abnormal points is greater than or equal to 30; Or the regional cave density is greater than or equal to 8 / cubic meter and the number of regional anomaly points is greater than or equal to 10.

[0098] The above embodiments of the present invention have at least the following beneficial effects: By generating a set of cave distribution characteristics and stratigraphic anomaly points, determining a sequence set of cave development trajectories, and performing risk zoning, high-risk areas in complex karst geological conditions can be accurately identified. This allows construction personnel to understand the distribution and development of caves in advance, enabling targeted risk prevention measures during pile foundation construction, effectively reducing construction risks such as hole collapse, grout leakage, and drill sticking, and improving construction safety and reliability. Based on the set of high-risk cave development areas, a set of high-risk hole locations and high-risk stratigraphic nodes is determined, and a karst geological risk map is generated. This process then generates a set of drilling process paths and a drilling risk heat map. This series of steps optimizes drilling process path planning, enabling the drill bit to effectively avoid caves and stratigraphic anomalies, reducing obstacles and risks during construction, improving drilling efficiency, and ensuring drilling quality, thereby shortening the construction period and reducing construction costs. The drilling risk heat map is overlaid on the pile foundation construction plan for display. During construction, real-time drilling sensor data is acquired, and drilling quality deviation alarms are generated based on the deviation values. When the alarm exceeds a preset threshold, an automatic correction control protocol is triggered. This real-time monitoring and automatic deviation correction mechanism can promptly detect and correct deviations during the construction process, ensuring that the construction process is always under control, further improving the accuracy and quality of pile foundation drilling, and enhancing the stability and reliability of the construction process.

[0099] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for controlling the quality of pile foundation holes in complex karst geology, characterized in that: include: Based on multi-source karst geological survey data, a three-dimensional geological anomaly model is constructed to characterize the spatial distribution characteristics of caves and stratigraphic anomalies through multi-modal data collaborative optimization processing; Based on the three-dimensional geological anomaly model, dynamically analyzing and reconstructing the development trajectory sequence of the karst cave to form a karst cave development trajectory sequence set containing information on the dissolution zone and rock layer interface; Based on the cave development trajectory sequence set, combined with the cave density in the depth dimension and the distribution characteristics of stratigraphic anomalies, a three-dimensional dynamic risk assessment model is established to identify and divide high-risk cave development areas; Locating and determining a high-risk hole location set based on the spatial distribution characteristics of the high-risk cave development area set and the preset cave development conditions; Based on the high-risk cave development area set, determine the stratigraphic structure nodes within its influence range through spatial correlation analysis to form a high-risk stratigraphic node set; The multi-dimensional karst geological risk map is constructed by integrating the cave distribution feature set, the cave development trajectory sequence set, the high-risk cave development area set, the high-risk well location set, and the high-risk formation node set; Based on the karst geological risk map, an optimized set of drilling process paths is intelligently generated, and based on this, a drilling risk heat map is generated to characterize the spatial distribution of construction risks; The hole formation risk heat map is spatially superimposed and fused with the pile foundation construction plan to generate a pile foundation hole formation risk heat map, which is then visually presented.

2. The method according to claim 1, characterized in that The method is based on multi-source karst geological survey data and uses multi-modal data collaborative optimization processing to construct a three-dimensional geological anomaly model that characterizes the spatial distribution characteristics of caves and stratigraphic anomalies, including: Acquire karst geological survey data including borehole coring data, geological radar detection data, and cross-hole CT imaging data; Standardizing the karst geological survey data, unifying the spatial reference and data scale, and obtaining standardized drilling data, radar data, and CT data; Performing spatial registration and data fusion on the standardized drilling data, radar data, and CT data to generate effective geological anomaly data; Extracting and generating a cave distribution feature set based on the effective geological anomaly data; Identify geological anomaly points associated with each cave feature in the cave distribution feature set to generate a stratigraphic anomaly point set.

3. The method according to claim 2, characterized in that The above-mentioned spatial registration and data fusion of standardized drilling data, radar data and CT data to generate effective geological anomaly data includes: Performing spatial coordinate matching on the standardized drilling data, radar data, and CT data to generate overlapping anomaly data sets and independent anomaly data sets; Performing a three-dimensional data fusion calculation on each overlapping anomaly data group in the overlapping anomaly data group set based on a preset spatial weight allocation strategy to generate a fused geological anomaly data set; The fused geological anomaly data set and the independent anomaly data set are jointly determined as effective geological anomaly data.

4. The method according to claim 1, wherein The method of dynamically analyzing and reconstructing the development trajectory sequence of the karst cave based on the three-dimensional geological anomaly model to form a karst cave development trajectory sequence set containing information on the dissolution zone and rock layer interface includes: Applying drilling trajectory positioning technology to determine the original spatial development point sequence of each cave in the cave distribution characteristic concentration; Performing interpolation optimization on the original spatial development point sequence under geological constraints to fill in the missing segments of the trajectory caused by sparse survey points, and generating a continuous sequence of completed development points; Performing geological noise filtering on the completed development point sequence to remove interference points caused by measurement errors or local geological disturbances, thereby generating an accurate development point sequence; Based on the precise development point sequence, a cave development trajectory sequence set containing information on the boundary of the dissolution zone and the rock layer interface is constructed through geological law modeling and spatial curve fitting.

5. The method according to claim 1, wherein Based on the cave development trajectory sequence set, combined with the cave density in the depth dimension and the distribution characteristics of abnormal points in the formation, a three-dimensional dynamic risk assessment model is established to identify and divide the high-risk cave development area set, including: The cave development trajectory sequence set is sliced ​​and layered according to preset depth intervals to extract the target dissolution zone set corresponding to each depth layer; For the target dissolution zone set at each depth level: Perform spatial grid risk zoning to form a cave development risk zone set containing multiple target dissolution zones; Calculate the number of target dissolution zones within the unit spatial volume of each risk zone to obtain the regional cave density; Based on the set of formation anomaly points, counting the number of formation anomaly points associated with each risk zone to obtain the number of regional anomaly points; Combining the density of regional caves and the number of abnormal points in the region, applying preset multi-dimensional high-risk judgment conditions, identifying and marking high-risk cave development areas at the depth level; The high-risk cave development areas at each depth level are summarized to form a set of high-risk cave development areas in three-dimensional space.

6. The method according to claim 1, characterized in that The karst geological survey data further includes a drilling pressure dynamic change sequence, a mud loss sequence, and a core breakage rate sequence; and the method further includes: The drilling pressure dynamic change sequence, mud loss sequence and core breakage rate sequence are defined as the first geological risk time series data; Extracting time series features from the first geological risk time series data to obtain geological risk time series features that characterize the dynamic risk of the drilling process; Based on the geological risk time series characteristics, predict the potential risk probability of hole collapse during the hole drilling process; generating second geological risk spatial data based on the high-risk cave development area set, the high-risk hole location set, and real-time drilling parameters; Performing multi-scale spatial feature extraction on the second geological risk spatial data to obtain multi-scale geological risk features reflecting local and regional risks; Performing multi-scale analysis on the geological risk time series characteristics to obtain the geological risk time series multi-scale characteristics; Performing residual connection on the geological risk time series feature and the geological risk time series multi-scale feature, retaining the original time series information and enhancing the feature expression capability, to obtain the geological risk residual connection feature; fusing the geological risk multi-scale feature with the geological risk residual connection feature to form a fused geological risk feature that comprehensively characterizes the geological risk; Based on the integrated geological risk characteristics, intelligent decisions are made and drilling control instructions are generated, which are transmitted to the pile foundation construction equipment for execution. The drilling control instructions include dynamic drilling speed adjustment instructions, active wall reinforcement instructions, and grouting adaptive compensation instructions.

7. The method according to claim 1, characterized in that The intelligent generation of optimized drilling process path sets based on the karst geological risk map includes: Planning a spatial node sequence for grouting reinforcement based on a set of high-risk stratum nodes in the karst geological risk map; generating a set of avoidance paths for a drill bit according to the spatial morphological characteristics of the cave development trajectory sequence set; The grouting reinforcement node sequence and the drill bit avoidance path set are matched with time-space collaborative optimization to generate a hole-forming process path set.

8. The method according to claim 1, characterized in that After visually presenting the pile foundation hole risk heat map, the method further includes: Real-time collection of drilling sensor data, including drill bit vibration spectrum characteristics, mud density dynamics, and hole inclination angle changes; Calculating a dynamic deviation value between the real-time drilling sensor data and the predicted state of the pile foundation hole risk heat map; generating a hole quality deviation alarm based on the dynamic deviation value; When the hole quality deviation alarm exceeds a preset safety threshold, the closed-loop control correction control protocol is automatically triggered.

9. The method according to claim 3, characterized in that The three-dimensional data fusion calculation based on the preset spatial weight distribution strategy includes: Different spatial weight coefficients are assigned to the borehole data, radar data and CT data in the overlapping anomaly data group, among which the borehole data is assigned the highest weight, followed by the radar data, and the CT data has the lowest weight; Based on the assigned weight coefficients, weighted fusion calculation is performed on the three-dimensional data of the overlapping area to generate fused geological anomaly data.

10. The method according to claim 5, characterized in that An area with high-risk karst cave development is determined to be one that meets any of the following conditions: The regional cave density reaches a first preset high density threshold and the number of regional outliers reaches a first preset high outlier threshold; or the regional cave density reaches a second preset high density threshold and the number of regional outlier points reaches a second preset high outlier point threshold; Or the regional cave density reaches a third preset high density threshold and the number of regional anomaly points reaches a third preset high anomaly point threshold.

Images