Adaptive grouting water-conducting fracture precision sealing device

Abstract

This invention belongs to the technical field, specifically relating to an adaptive grouting device for precise sealing of water-conducting fractures. The device includes: a data detection unit, a path construction unit, and a multi-domain co-directional frequency conversion grouting unit. The data detection unit is used to perform multi-directional scanning of the area to be sealed using an acoustic pulse detector, forming a node mapping list. The path construction unit is used to read the node mapping list, trigger symmetrical hierarchical reconstruction, output a second-level sealing tree path, and use this second-level sealing tree path as the final grouting channel. The multi-domain co-directional frequency conversion grouting unit is used to lock the grouting operation section corresponding to the current domain combination after each cycle, and perform the grouting operation through a grouting material composite.

Description

Technical Field

[0001] This invention belongs to the field of automatic control technology, specifically relating to a precise sealing device for water-conducting fissures using adaptive grouting. Background Technology

[0002] In the construction and operation of underground engineering projects, such as tunnel excavation, mine roadway support, reservoir dam seepage prevention, and groundwater resource management, water-conducting fractures in fractured rock masses are consistently a key factor causing engineering hazards such as water inrush, leakage, and instability. Grouting technology is widely used to address these water-conducting channels. Its basic principle is to inject a solidifiable material into the fracture system to seal the channels, improve overall impermeability, and thus achieve water stoppage and reinforcement. However, in actual engineering projects, the complex geometry of fracture spaces, the varied interconnected structures, and the difficulty in visualizing the grouting process mean that existing grouting technologies still face many technical bottlenecks in terms of "accurate identification, reasonable path planning, and quantitative control of grouting parameters."

[0003] Currently widely used grouting technologies largely rely on initial manual surveys to determine fracture distribution, followed by grouting operations through the establishment of fixed grouting holes. These traditional methods depend on preliminary drilling, ground-penetrating radar, and resistivity imaging to obtain fracture information. However, these methods are mostly indirect inferences, suffering from low resolution, susceptibility to interference, and significant deviations between the identified results and the actual fracture structure. For example, ground-penetrating radar is ineffective at detecting water-saturated fractures, and electrical resistivity methods are greatly affected by soil moisture content and conductivity, often resulting in blurry images that fail to provide a clear fracture distribution map. While drilling methods are direct, their coverage is limited, the number of points is sparse, and the cost is high, easily overlooking important channels. Furthermore, traditional grouting often uses "equal pressure grouting" or "equal volume grouting" as control methods. Without a real-time response mechanism, it is difficult to adjust the grouting path and parameters promptly based on the fracture spatial structure and fluid diffusion, often leading to over-grouting, under-grouting, or ineffective grout diffusion.

[0004] In terms of grouting path planning, existing methods typically design grouting layouts based on two-dimensional profiles or static fracture distribution maps, lacking connectivity analysis of complex three-dimensional fracture networks and failing to construct grouting paths with hierarchical logic and dynamic adaptability. Fracture structures generally exhibit characteristics of "multiple branches, clear distinction between primary and secondary branches, and strong connectivity." Fractures in different regions vary significantly in scale, aperture, and orientation. If the grouting path cannot adapt to these variations, grout can easily traverse rapidly along the main channels, while secondary channels fail to be effectively sealed due to insufficient pressure distribution. Furthermore, numerous discontinuous, closed, or blind-end structures exist within the fracture network. If the grouting path cannot effectively identify and avoid these structures, it can also lead to grout accumulation, blockage, or even backflow. Summary of the Invention

[0005] Therefore, the main objective of this invention is to provide a precise sealing device for water-conducting fractures using adaptive grouting. It achieves real-time multi-directional identification of fracture structures through acoustic pulse detection, constructs a sealing tree-like path based on node density and connectivity, and employs a multi-domain co-directional frequency conversion strategy to segmentally control the grouting channel. This allows parameters such as pressure, viscosity, rate, and time to alternate and interact during the grouting process, ultimately achieving efficient, precise, and stable sealing of water-conducting fractures. This invention offers advantages such as high identification accuracy, intelligent path construction, strong controllability of the grouting process, and reliable sealing effect, significantly improving grouting quality and engineering safety under complex geological conditions.

[0006] The technical solution adopted in this invention is as follows:

[0007] An adaptive grouting device for precise sealing of water-conducting fractures includes: a data detection unit, a path construction unit, and a multi-domain co-directional frequency conversion grouting unit. The data detection unit uses an acoustic pulse detector to perform multi-directional scanning of the area to be sealed, obtaining raw acoustic pulse data of the fractures. It then generates a fracture segment sequence through layered splicing, marking grouting candidate nodes in the fracture segment sequence to form a node mapping list. The path construction unit reads the node mapping list, calls a three-way recursive connector to connect all grouting candidate nodes to form a sealing tree path, performs incremental hierarchical judgment on the sealing tree path, and divides the path based on node density and connectivity to obtain the first-level sealing tree path. Based on the traversal results of the first-level blocking tree path, a symmetrical hierarchical reconstruction is triggered, outputting the second-level blocking tree path, which is then used as the final grouting channel. The multi-domain co-directional frequency conversion grouting unit is used to divide each second-level blocking tree path into four domains: pressure domain, viscosity domain, velocity domain, and time domain. The second-level blocking tree path is linearly sorted from the beginning to the end to generate a linear path index sequence. The four domains are alternated according to the linear path index sequence, so that the pressure domain, viscosity domain, velocity domain, and time domain appear continuously in a cycle. After each cycle is completed, the grouting operation section corresponding to the current domain combination is locked, and the grouting operation is performed through the grouting material composite.

[0008] Furthermore, the data detection unit initializes the acoustic pulse detector based on the reference time value, drift compensation value, and environmental background value of the acoustic pulse detector. Starting from zero azimuth, the acoustic pulse detector rotates sequentially with fixed azimuth increments according to the preset scanning azimuth sequence number, forming a complete azimuth cycle. At each azimuth angle, the acoustic pulse detector continuously emits multiple acoustic pulses, and records in parallel the pulse emission sequence number, pulse echo peak value sequence, pulse echo duration sequence, and pulse echo arrival time sequence to obtain the original acoustic pulse data of the fracture. All azimuth angles form an azimuth sequence. A scanning batch identifier is generated for each azimuth angle, and this identifier is written into the corresponding sequence of the original acoustic pulse data of the fracture.

[0009] Furthermore, the data detection unit sorts the original acoustic pulse data of the fracture according to the scanning batch identifier to form a root order sorting sequence; the root order sorting sequence is divided into layers for the first time according to the pulse echo duration sequence to generate the first fracture echo layer; within the first fracture echo layer, a second layer is divided according to the pulse echo peak sequence to generate the second fracture echo sub-layer; the layer splicing process assigns a segment index number to each fracture echo sub-layer and associates this number with the original pulse transmission sequence number, finally outputting a continuously increasing fracture segment sequence.

[0010] Furthermore, the data detection unit traverses each segment index number according to the fracture segment sequence and calls the candidate node determination process. The candidate node determination process adopts a dual-threshold differential judgment strategy: first, it compares the pulse echo peak sequence with the dynamic peak threshold sequence; second, it compares the pulse echo duration sequence with the dynamic duration threshold sequence. When both comparisons are satisfied, the current segment index number is marked as a grouting candidate node. A node labeling status sequence is written for each grouting candidate node, and a node sequence number is generated. All grouting candidate nodes are screened from the node labeling status sequence, and the corresponding segment index number, scanning batch identifier, pulse echo peak sequence, pulse echo duration sequence, and azimuth sequence are combined into a node mapping list item. The node mapping list items are reordered according to the ascending order of the azimuth sequence to generate a node sequence list. Based on the node sequence list, a node mapping list is created, and each node mapping list item is written as an independent item in the node mapping list.

[0011] Furthermore, the path construction unit extracts the node sequence number, scan batch identifier sequence, azimuth sequence, pulse echo peak value sequence, and pulse echo duration sequence for each node mapping list entry and writes them into the path construction buffer; it generates a path starting index sequence according to the ascending order of the node sequence number and records the initialization completion flag and initialization completion time flag in the path construction buffer; it selects the first node of the node sequence number as the root node in the path construction buffer; the three-way connection recursive unit synchronously reads the azimuth sequence value of the root node and generates three sets of direction mappings: forward direction mapping, lateral direction mapping, and longitudinal direction mapping; the three-way connection recursive unit searches the path construction buffer for the first node that satisfies the condition of continuously ascending node sequence numbers and azimuth sequence differences being in the same order according to the forward direction mapping. Write a positive pointer to each node in the interval; the three-way recursive recursive recursively retrieves the first node that satisfies the condition that the azimuth sequence difference is in the lateral interval and the pulse echo peak sequence is higher than the root node's pulse echo peak sequence, and writes a lateral pointer to it; the three-way recursive recursively retrieves the first node that satisfies the condition that the scan batch identifier sequence is equal to the root node's scan batch identifier sequence and the pulse echo duration sequence is higher than the root node's pulse echo duration sequence, and writes a vertical pointer to it; the three-way recursive recursively calls itself on the positive, lateral, and vertical pointers respectively, until all node sequence numbers have been written to at least one arbitrary pointer, and the entire connection result is written to the blocking tree path sequence; record the generation time of the blocking tree path sequence and the total number of nodes in the blocking tree path sequence.

[0012] Furthermore, the path construction unit traverses the blocking tree path sequence, counts the number of nodes for each connected segment, and writes it into the node count sequence; it generates a node density value sequence in the path construction buffer according to the node count sequence and the node number sequence; during the traversal of the blocking tree path sequence, it records the sum of the number of forward and backward connections for each node, generating a connectivity value sequence; the incremental level determination process adopts a dual threshold progressive strategy: first, it compares the node density value sequence with the density threshold sequence, and second, it compares the connectivity value sequence with the connectivity threshold sequence. When both comparisons are satisfied, the current connected segment is written into the first-level candidate sequence table; the first-level candidate sequence table is reordered in ascending order of the node number sequence to generate the first-level blocking tree path, and the first-level blocking tree path sequence and the first-level blocking tree path generation flag are written into it.

[0013] Furthermore, the path construction unit performs a full traversal of the first-level blocking tree path, recording the traversal sequence, branch depth sequence, and branch width sequence during the process. After the traversal is completed, a symmetrical hierarchical reconstruction is performed, specifically including: copying the traversal sequence from the beginning to the end in reverse and writing it into the reverse traversal sequence; performing odd-even grouping on the branch depth sequence and branch width sequence respectively, maintaining the order of the odd array and reversing the order of the even array; realigning the reverse traversal sequence, the branch depth sequence after odd-even grouping, and the branch width sequence according to the node number sequence to generate a reconstruction instruction sequence; re-arranging the first-level blocking tree path one by one according to the reconstruction instruction sequence, outputting the final second-level blocking tree path, and writing the second-level blocking tree path sequence and the second-level blocking tree path generation time marker; reading the first node and the last node of the second-level blocking tree path sequence and writing them into the grouting channel boundary sequence; generating the grouting channel index sequence between the grouting channel boundary sequences according to the order of the second-level blocking tree path sequence.

[0014] Furthermore, the multi-domain co-directional frequency conversion grouting unit locates the second-level blocking tree path generation time marker, reads all nodes of the second-level blocking tree path sequence corresponding to the marker; extracts the node sequence number, azimuth sequence, scan batch identifier sequence, pulse echo peak value sequence, and pulse echo duration sequence for each node, and writes them into the grouting unit's operating buffer; records the path import completion flag and path import completion time flag in the grouting unit's operating buffer; sorts the grouting unit's operating buffer according to the ascending order of the node sequence number to generate a linear node sequence; writes the linear node sequence into the linear path index sequence, and appends the linear path index sequence generation time marker; records the first node index and the last node index of the linear path index sequence in the grouting unit's operating buffer, and writes the path... Boundary record sequence; read the total number of nodes in the linear path index sequence, and call the quartile cyclic segmentation process, which specifically includes: performing a node mean scan on the linear path index sequence, recording the coordinates of each quartile node in the node mean sequence; starting from the first node, using the quartile node coordinates recorded in the node mean sequence as the domain segment boundary points, and sequentially labeling the pressure domain, viscosity domain, rate domain, and time domain; immediately jumping back to the first node after segmenting to the last node, realizing multiple loops until all nodes are labeled with the domain segment type; writing the segmentation results into the domain segment labeling sequence, where the pressure domain is labeled as pressure domain, the viscosity domain as viscosity domain, the rate domain as rate domain, and the time domain as time domain, ensuring that the order of the four domain segments remains unchanged in each loop; recording the generation time of the domain segment labeling sequence and the total number of domain segment alternation loops.

[0015] Further, for the multi-domain co-directional frequency conversion grouting unit, the process of reading the linear path index sequence and the domain segment annotation sequence and performing domain segment alternation specifically includes: verifying whether the pressure domain, viscosity domain, rate domain, and time domain appear in sequence in the line order among every four consecutive nodes; if the verification fails, the domain segment alternation process exchanges the domain segment annotation values of adjacent nodes within the same section in a moving window manner until a complete sequence appears; after each complete sequence is achieved, write a domain segment alternation completion flag once; after each domain segment alternation is completed, read the start node index and end node index of the current four consecutive nodes and write them into the current domain segment group boundary sequence; perform an immediate locking operation on the current domain segment group boundary sequence, which specifically includes: copying the current domain segment group boundary sequence to the grouting operation section locking sequence; recording the grouting operation section locking time flag and the grouting operation section locking completion flag.

[0016] Adopting the above technical solutions, the present invention has the following beneficial effects: The accurate plugging device for water-conducting fissures with adaptive grouting provided by the present invention can achieve the full-process integration of fissure identification, path construction, and multi-parameter controlled grouting in a complex rock mass environment, with remarkable beneficial effects. First, through the multi-directional and multi-batch active scanning by the acoustic pulse detector, it can obtain the internal structural change information of the fissures in real time, significantly improving the detection accuracy and spatial recognition ability of water-conducting fissures, and avoiding the misjudgment and omission of traditional static exploration methods. Second, by constructing a plugging tree-shaped path and performing incremental level determination and symmetric reconstruction based on the node density and connectivity, it effectively improves the matching degree between the grouting path and the actual fissure geometric characteristics, realizes the layer-by-layer construction of the grouting channel from thick to thin and from shallow to deep, and improves the uniformity of slurry distribution and the plugging coverage rate. In addition, the grouting process divides the grouting path into a pressure domain, a viscosity domain, a rate domain, and a time domain based on the multi-domain co-directional frequency conversion strategy, achieving multi-parameter linkage control, enabling the grouting operation to dynamically adjust the physical control strategy according to the fissure characteristics of each section, and improving the grouting efficiency and slurry utilization rate. Through the domain segment alternation, path index sorting, and operation section locking mechanisms, the grouting system realizes continuous, multi-loop, and accurate grouting actions on a single plugging path, enhancing the continuity, density, and stability of the plugging body, and is especially suitable for engineering geological environments with developed rock mass fissures and variable structures. Finally, by forming an adaptive closed-loop process of "detection - identification - construction - grouting - feedback", the device not only improves the construction automation level but also significantly reduces the uncertainty of the plugging effect caused by human intervention, and has good engineering practicability and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic structural diagram of the accurate plugging device for water-conducting fissures with adaptive grouting provided by an embodiment of the present invention;

[0018] Figure 2A schematic diagram illustrating the experimental effect of acquiring raw acoustic pulse data of a crack using a multi-directional acoustic pulse detector provided in an embodiment of the present invention.

[0019] Figure 3 This is a schematic diagram illustrating the experimental effect of alternating grouting in different domain segments provided in an embodiment of the present invention. Detailed Implementation

[0020] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.

[0021] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0022] refer to Figure 1 The data detection unit establishes an instantaneous wave field within the underground rock mass using an acoustic pulse detector. The reflection amplitude and duration of the acoustic pulse at different fracture interfaces depend on the abrupt change in medium impedance and the geometric scale of the fracture. As the detector rotates incrementally in a fixed azimuth, the echo time series generated at each azimuth angle constitute a multidimensional data volume. The system uses the reference time value of the acoustic pulse detector as the time synchronization reference for all samples, corrects the time base drift caused by temperature and aging of electronic components with drift compensation values, and subtracts the inherent background noise of the rock mass with environmental background values, thereby ensuring that the echo peak value and duration truly correspond to the energy scattering characteristics of the fracture interface. The acquired raw acoustic pulse data of the fracture is coherent in the time dimension and continuous in the azimuth dimension, laying a signal quality foundation for subsequent layered stitching.

[0023] The system treats the pulse echo duration sequence as a first-order index of the fracture scale and the pulse echo peak sequence as a second-order index of the filling state. During layered stitching, threshold partitioning is first performed along the duration dimension, with echoes of similar durations aggregated into the first fracture echo layer. Then, further subdivision is performed along the peak dimension, with segments of similar energy labeled as the second fracture echo sub-layer. This "two-level homogeneous clustering" deconstructs the mixed wavefield observations into a traceable fracture segment sequence. The fracture segment sequence represents the pixel rows in electron transmission scanning, while the segment index number plays the role of the pixel column. Together, they establish an addressable coordinate system for the fracture space. The candidate node determination process adopts a dual-threshold differential judgment strategy because relying solely on peak value or duration is insufficient to eliminate false echoes caused by rock joints, micropores, or equipment noise. First, a dynamic peak threshold sequence is used to filter low-energy scatterers, and then a dynamic duration threshold sequence is used to eliminate short-period reflections, achieving accurate screening of true fracture endpoints. By jointly sorting the node label state sequence and the azimuth sequence, the system unifies the time order of discrete candidate nodes falling in the same azimuth and the angular order of falling in different azimuths into the node mapping list, logically constructing a sparse yet ordered 3D point cloud.

[0024] The primary task of the path building unit is to organize this point cloud into a tree-like path with feasible fluid pathways. A three-way recursive network implements a connectivity exploration similar to depth-first traversal: forward mapping ensures the path maintains spatial sequential consistency with node indices, lateral mapping guarantees that branches can expand laterally without geometric discontinuities due to sharp angles, and longitudinal mapping locks in possible vertical connections within the same scan batch. The introduction of the recursive structure means that once any node is confirmed, it triggers the deep diffusion of the ternary tree until all nodes are integrated into a single-root, multi-branch tree network. This network is not a simple geometric connection but a connectivity graph containing physical flow feasibility, because lateral mapping must satisfy pulse echo peak enhancement, and longitudinal mapping must satisfy pulse echo duration increase, reflecting the hierarchical connectivity evolution of the fracture in the dimensions of energy reflection and aperture.

[0025] After the blocking tree path is generated, the incremental hierarchical judgment mechanism uses the node density value sequence and connectivity value sequence as dual evaluation metrics. Node density corresponds to the "degree of point accumulation" in a certain segment of the path, and connectivity corresponds to the "branch complexity of fluid escape." Through a dual threshold progressive strategy, the path is divided into alternating dense-sparse-dense segments. The first-level blocking tree path is actually a coarse-grained flow distribution diagram. Symmetrical hierarchical reconstruction is not a simple flip, but rather uses the reverse traversal sequence and the branch depth and branch width sequences after rearranging odd and even groups to perform a self-mirror mapping of the first-level result in space. The result is that the second-level blocking tree path has high geometric symmetry and streamline balance. It consolidates the multi-branch structure into several main branches, and then combines the main branches in a symmetrical inverted manner, reducing local high pressure and dead zones during grouting. This path is fixed as the only grouting channel because it still satisfies the node density-connectivity dual thresholds after progressive judgment and reconstruction, and the traversal sequence can be closed between forward and reverse directions.

[0026] The second-level sealing tree path faced by the multi-domain co-directional frequency conversion grouting unit is a linear curve with unevenly distributed nodes. The system first generates a linear path index sequence according to the node number sequence to ensure that the topological order is consistent with the physical order. The fundamental principle of quartile cyclic segmentation is based on the mean scan of statistical distribution: the nodes are evenly divided into four parts, and the quartile node coordinates are used as the boundary points of the pressure domain, viscosity domain, velocity domain, and time domain. This allows different domain segments to spatially enclose an equal number of nodes as much as possible, and makes the grout action range of different domain segments correspond to the same crack length, thereby avoiding "over-grouting" or "under-grouting" within a single domain segment. The domain segment alternation process is equivalent to sliding a window of length four on the linear chain. Each time the window is inspected, if the order is found to be disordered, the domain segment label values ​​of adjacent nodes are swapped. This moving window swapping ensures that the global correct order can be restored with minimal local adjustment cost, reducing the impact of domain segment label mismatch on grouting continuity.

[0027] After the alternation of grouting segments is completed, the boundary sequence of the current grouting segment group is written into the grouting operation segment locking sequence through an instant locking operation. At this time, pressure is applied to the grouting material composite. The pressure domain stage adopts a stepped loading rule, gradually increasing the static pressure head to help the grout overcome the initial resistance of the pore-fracture. The viscosity domain stage switches to a high viscosity-low flow characteristic grout, where the grout slowly expands a short distance at the fracture tip and locally dams due to higher viscosity. The rate domain stage emphasizes flow pulses, creating dynamic disturbances through intermittent control of high and low flow under the same static pressure, pushing the grout agglomerates left in the viscosity domain back into the micro-cracks. The time domain stage maintains a constant flow-constant viscosity combination, using time control to ensure that the grout is fully hydrated and solidified inside the fracture. Here, "unidirectional" means that the four grouting segments advance in one direction along a continuous path, and "variable frequency" means that parameters such as pressure, flow rate, and stirring frequency change periodically with the alternation of grouting segments. After comparing the pressure and flow response curves generated in real time in the grouting section with the system's preset target curve, if a deviation occurs, the data detection unit will immediately perform incremental scanning near the deviation section to update the original acoustic pulse data of the fracture. The path construction unit then determines whether the section needs to add branch deviation lines or rearrange the node density. If it is determined to be necessary, it will superimpose corrected branches on the original path. The multi-domain co-directional frequency conversion grouting unit re-executes the quartile cycle and domain alternation according to the new path nodes, thereby completing one closed-loop iteration. After multiple iterations, the underground fracture network is filled in stages and the branches are sealed one by one until the peak value of the acoustic pulse echo drops to below the continuous threshold and the duration is shortened to the background level, indicating that the water-conducting fractures have lost detectable connectivity, and the system stops grouting.

[0028] Because it uses acoustic pulses as the detection basis, the device is naturally sensitive to the distribution of pressure wave velocity, and can quickly identify newly formed solidified zones and residual unblocked channels after the grout solidifies; the blocking tree path output by the three-way recursive device has an embedded node number increment attribute, which allows the system to quickly trace back to the location index on the path and locate the fault segment when any parameter drift occurs; the multi-domain co-directional frequency conversion grouting frame decouples the four physical processes of pressure, viscosity, rate and time through alternating domain segments, and then recouples them to a single path through co-directional sequence, ensuring that the grout propagation always follows the preset direction and does not cause backflow. The deep integration of the three major units endows the device with adaptive capabilities: when lithology changes, porosity increases suddenly, or in-situ water pressure fluctuates, acoustic pulse feedback immediately alters the dynamic threshold sequence, and the number of candidate nodes changes accordingly; the blocking tree path is automatically re-evaluated in the next incremental level judgment cycle, and whether to reconstruct depends on the threshold judgment result; the alternating domain segments are automatically translated and reorganized due to node reordering, and combined with the switchable different grout components in the material composite, the system can make matching responses to complex and ever-changing water-conducting fracture networks without human intervention. Ultimately, this closed loop achieves graded, high-density, and dynamic sealing of fractures, avoiding the risks of rock mass uplift due to "over-grouting" or water leakage due to "under-grouting" in traditional single-parameter grouting, and also improving material utilization to near the theoretical limit.

[0029] Furthermore, upon starting operation, the data detection unit first retrieves the acoustic pulse detector's reference time value, acoustic pulse detector drift compensation value, and acoustic pulse detector's environmental background value. These three baseline information items correspond to time synchronization, hardware stability correction, and environmental noise subtraction, respectively. The system uses the acoustic pulse detector's reference time value to establish a unified timing starting point, mapping the occurrence time of all subsequent pulse events to the same reference frame. The acoustic pulse detector's drift compensation value is used to dynamically correct the minute frequency offsets generated by the clock, amplifier, and analog front end during long-term operation, ensuring that data from multiple scans maintains time consistency across batches. The acoustic pulse detector's environmental background value is used to sample the background sound field of the rock mass where the detector is currently located in real time, and the background value is subtracted before each pulse echo signal is written, ensuring that the subsequent analysis of the pulse echo peak sequence and pulse echo duration sequence relies more on actual fracture reflections rather than irrelevant noise.

[0030] After initialization, the system initiates azimuth looping based on the preset scanning azimuth sequence number. Starting from zero azimuth, the loop rotates with a fixed azimuth increment until all discrete angles of the azimuth sequence have been traversed. At each azimuth position, the acoustic pulse detector continuously emits multiple acoustic pulses, recording the pulse emission sequence number immediately for each emission to distinguish multiple emission events within the same azimuth. Along with the echo return, the system simultaneously acquires the pulse echo peak sequence, pulse echo duration sequence, and pulse echo arrival time sequence, writing these four sets of information in parallel into the original acoustic pulse data entry for the fracture. This is done to simultaneously save three representative observations of the fracture state: energy amplitude, duration, and propagation delay, facilitating subsequent layered stitching and dynamic adjustment of the threshold strategy.

[0031] All azimuth angles together form an azimuth angle sequence, and the order of the azimuth angle sequence is the physical order of detector rotation. To avoid data cross-contamination between different azimuth angles, the system generates a scan batch identifier for each azimuth angle. The scan batch identifier is independent of the pulse emission sequence number in format; it embodies the concept of "spatial location index," while the pulse emission sequence number embodies the concept of "temporal emission index." The cross-referencing of the two makes the raw acoustic pulse data of the fracture traceable in both spatial and temporal dimensions. Once the scan batch identifier is generated, it is immediately written into the corresponding sequence of raw acoustic pulse data of the fracture and serves as the first key sorting key in the subsequent layered stitching process. This ensures that multiple sets of pulse data from the same azimuth angle remain aggregated and synchronized in the primary processing stage, providing a stable input starting point for the candidate node determination process.

[0032] Furthermore, after completing the azimuth scan and batch identifier writing, the data detection unit first calls the sorting engine to perform a full rearrangement of the original acoustic pulse data of the fracture according to the scan batch identifier. The rearrangement result is written into the root order sorting sequence. The fundamental purpose of this operation is to maintain the natural clustering of pulse echo entries from the same azimuth angle, while arranging data entries generated at different times on the time axis according to the spatial scan order. Once the root order sorting sequence is established, the system immediately performs threshold grouping using the pulse echo duration sequence as the first dividing dimension. The specific strategy is to divide the duration into several intervals from short to long, and group pulse entries whose durations fall into the same interval into the same layer, generating the first fracture echo layer. Since the duration is positively correlated with the fracture aperture, this division effectively distinguishes between wider and narrower fractures, laying the spatial foundation for subsequent energy difference analysis.

[0033] Within the first fracture echo layer, the system further subdivides the data using the pulse echo peak sequence as a second dividing dimension. Echo energies within the same duration interval are not necessarily consistent, and energy levels directly reflect the reflection intensity and filling state of the fracture wall. By setting incremental thresholds for peak values ​​from low to high, the detection algorithm further subdivides entries with similar energies to generate a second fracture echo sublayer. This two-level stratification not only maintains the homogeneity of duration but also explicitly reveals the differences in energy distribution, allowing subsequent candidate node determination to focus on energy dispersion within the same geometric scale, thereby reducing erroneous labeling.

[0034] Once the second layer of fracture echo sublayers is confirmed by the system, the layer-by-layer splicing process assigns an independent segment index number to each fracture echo sublayer and immediately establishes a one-to-one association with the corresponding original pulse emission sequence number. The segment index number provides a continuously addressable spatial label, while the original pulse emission sequence number preserves the temporal order information. The combination of the two ensures that spatial segments are traceable and preserves the true temporal sequence of events. The layer splicer then reassembles all the numbers in the order of generation, outputting a continuously increasing sequence of fracture segments. This sequence remains monotonically increasing throughout, providing a clear and traversable main chain for the subsequent candidate node determination process. It also lays a strict index-increasing prior for the forward direction mapping of the three-way connection recursive device, ensuring that the data faced by the back-end path construction is neither gap-filled nor overlapping, thus creating stable and reliable front-end data support for precise blocking.

[0035] Furthermore, after obtaining the continuously increasing sequence of fracture segments, the data detection unit immediately activates the traversal engine to read each segment index number sequentially according to the fracture segment sequence, and dynamically invokes the candidate node determination process for each read. The candidate node determination process uses a dual-threshold differential judgment strategy as its core logic. First, it references the corresponding entry of the pulse echo peak sequence and compares the difference with the dynamic peak threshold sequence calculated in real time in the same batch. The dynamic peak threshold sequence is not a fixed constant, but rather a sliding window method that statistically analyzes the mean and dispersion of the peak sequences in the same batch, and then floats the mean up according to a preset ratio to form an energy threshold that is updated over time and with orientation. If the instantaneous amplitude of the current pulse echo peak sequence exceeds the dynamic peak threshold sequence, the system determines that the first-level energy condition is met; if it does not exceed, it directly terminates the current round of determination, marks the segment index number as non-candidate, and continues to read the next segment index number.

[0036] Once the energy condition is met, the process enters the second-level comparison, comparing the corresponding entries of the pulse echo duration sequence with the dynamic duration threshold sequence. The dynamic duration threshold sequence also employs a sliding window strategy, but it focuses on the median and upper quartile trend of the duration sequence, applying an adjustable gain to the statistical results to capture the relative peak value of the fracture width within a local segment. If the current pulse echo duration sequence is higher than the dynamic duration threshold sequence, the second-level geometric condition is met; if it is lower than the threshold, the system also abandons the candidate status of the current segment index number. Only when both the energy and geometric conditions are met simultaneously, the data detection unit marks the current segment index number as a grouting candidate node, writes the corresponding bit in the node labeling status sequence into the "candidate" state, and appends the segment index number to the node sequence number.

[0037] After traversing the node-annotated state sequence, it is scanned all at once, and the system extracts all entries marked as "candidates". For these entries, the detection module simultaneously reads the scan batch identifier, pulse-echo peak value sequence, pulse-echo duration sequence, and azimuth sequence associated with the segment index number, and combines them into a node mapping list according to the entry aggregation rules. Each entry contains five elements: the segment index number provides the fracture segment number, the scan batch identifier indicates the spatial azimuth, the pulse-echo peak value sequence and pulse-echo duration sequence describe the local energy and geometric state, and the azimuth sequence points to the angular position in the global coordinate system. Subsequently, the system reorders all node mapping list entries in ascending order of the azimuth sequence to generate a node order list. The sorting adopts a stable sorting algorithm to ensure that if the azimuth values ​​are the same, the original order of the batch scans is maintained, thus taking into account both spatial continuity and temporal coherence.

[0038] Logically, the node sequence list is equivalent to a chain of candidate nodes arranged angularly. The data exploration unit uses this list as input and calls the write routine to create an independent entry space for the node mapping list. The system copies each entry of the node mapping list to its corresponding position until all candidate nodes have been written. At this point, the node mapping list is finalized during the data exploration phase, and the subsequent path construction unit can directly read the list to perform three-way recursive connection, incremental hierarchical judgment, and symmetrical hierarchical reconstruction. The entire process effectively eliminates noise, false echoes, and non-target reflections through a dual-threshold differential judgment strategy, retaining only the true fracture endpoints that exhibit local high values ​​in both the energy and geometric dimensions. This ensures high accuracy and high confidence of the grouting candidate nodes, laying a reliable source data foundation for the generation of precise blocking tree paths.

[0039] Furthermore, after the path construction unit takes over the node mapping list output by the data detection unit, it first opens a path construction buffer internally. It then extracts and sequentially writes the node sequence number, scan batch identifier, azimuth sequence, pulse-echo peak value sequence, and pulse-echo duration sequence from each entry in the node mapping list, field by field. At this point, the path construction buffer forms a five-dimensional data table: the node sequence number acts as the primary key index, the scan batch identifier and azimuth sequence jointly describe the spatial location, and the pulse-echo peak value and pulse-echo duration sequences retain energy and geometric information. The system generates a path starting index sequence in ascending order of the node sequence number, using it as the row pointer to access the buffer, ensuring that any subsequent traversal follows a monotonically increasing physical continuity. Once all entries have been written, the path construction unit immediately records the initialization completion flag in the buffer and simultaneously writes the initialization completion time flag, providing an anchor point for time backtracking and incremental updates.

[0040] Subsequently, the system extracts the first node of the node sequence from the path construction buffer and sets it as the root node. The root node is the starting trigger point of the three-way connection recursive and the source of the blocking tree path sequence. The three-way connection recursive first synchronously reads the azimuth sequence value corresponding to the root node and generates three sets of direction mappings based on this: the forward direction mapping sets the same direction interval on the azimuth sequence to keep the angle deviation within a preset threshold; the lateral direction mapping sets the lateral interval on the azimuth sequence to capture the fracture branches that have a limited angle deflection from the root node but still have the same scattering fluctuation characteristics; the longitudinal direction mapping does not focus on the azimuth difference value, but limits the scan batch identifier sequence to be consistent with the root node to emphasize the geometric extension of the depth direction on the same detection plane.

[0041] The three-way recursive algorithm prioritizes forward direction mapping. It searches the path buffer for the first node that simultaneously meets two conditions: its node number must be consecutively increasing after the root node, and its azimuth sequence difference must fall within the same direction interval. Once found, this node is written to the forward pointer, and the relationship between the root node and this node in the buffer is recorded as parent-child. If no node meeting the conditions is found at the end of the buffer, the forward pointer remains empty. Next, the recursive algorithm performs a lateral direction mapping search: it searches the buffer for the first node whose azimuth sequence difference is within the lateral interval and whose pulse-echo peak sequence is higher than that of the root node. The energy condition exists to ensure that lateral branches have stronger reflections than the main branch, conforming to the natural law of fracture branches gradually decreasing in width, and preventing incorrect layer jumps when the path alternates between wide and narrow sections. Nodes meeting the conditions are written to the lateral pointer; if no suitable node exists, it remains empty. The recursive then searches for the first node that simultaneously meets the criteria of having a scan batch identifier sequence equal to the root node and a pulse-echo duration sequence higher than the root node's pulse-echo duration sequence, based on the vertical direction mapping, and writes it to the vertical pointer. The duration sequence threshold constraint ensures that vertical extension will inevitably enter a wider fissure; otherwise, the fluid will find it difficult to continue moving forward.

[0042] After writing the three pointers at the root node, the three-way recursive recursive calls itself sequentially for the forward, lateral, and vertical pointers, following a depth-first principle. Each time a recursion occurs, the current node is treated as the new root node, and three sets of direction mappings are regenerated. The search for successor nodes continues in the path construction buffer. The recursion terminates when all nodes in the node sequence have been written to any pointer at least once, meaning all candidate nodes are included in the pointer network. During the backtracking phase, the recursive writes the parent-child relationships of each level into the blocking tree path sequence, causing the tree structure to converge from bottom to top into a complete hierarchical connected graph. The blocking tree path sequence not only records the connection order of node numbers but also retains the direction attribute corresponding to each pointer creation, providing a basis for subsequent incremental level determination. When the tree paths finally converge, the path construction unit records the generation time of the blocking tree path sequence in the buffer and counts the total number of nodes in the blocking tree path sequence to verify whether it matches the number of candidate nodes in the node mapping list. If they match, it means there were no lost steps in the path construction process; if they don't match, the system backtracks to the corresponding recursive layer based on the list of missing nodes, re-triggers the direction mapping retrieval, until it is completely closed. At this point, the blocking tree path sequence is formally established as the geometric skeleton for subsequent operations of the multi-domain co-directional frequency conversion grouting unit, and the path construction unit completes its phased task.

[0043] Furthermore, after obtaining the blockade tree path sequence, the path construction unit first proceeds along the connecting segments using a depth-first traversal approach. At each parent-child node connection segment, it immediately counts the number of nodes contained within that segment and writes the results sequentially into the node count sequence. The node count sequence is essentially a discretized description of the local clustering degree of the blockade tree path, mapping branches of different lengths to comparable integer indices. Next, the system aligns the node count sequence with the node number sequence line by line, generating a node density value sequence in the path construction buffer. The node density value sequence does not directly take the node count sequence value; instead, it divides the node count by the span of the same connecting segment in the node number sequence, allowing for comparison of node filling at different segment lengths using a unified dimension. Therefore, the same number of nodes distributed in short segments will result in high density values, while those distributed in long segments will result in low density values, facilitating subsequent decisions on whether to retain the main branch or discard redundant branches based on the gap-filling requirements.

[0044] During the traversal of the blocked tree path sequence, the system simultaneously establishes a local counter for each visited node to accumulate the number of forward connections from the current node to subsequent nodes and the number of backward connections pointed to by the predecessor node. The sum of these two counts forms the connectivity value, which is sequentially written into the connectivity value sequence. The connectivity value sequence reflects the number of "walkable" paths for the fluid in the network: a high value indicates that the node is in a network intersection area, while a low value indicates that the node is at the end of a single channel. This index, along with the node density value sequence, is used in the incremental hierarchy determination process.

[0045] The incremental grouting process employs a dual-threshold progressive strategy. First, the system compares each node density value sequence with a density threshold sequence, which is a density lower limit curve that adjusts in real-time for different segment lengths and lithological conditions. This threshold sequence is used to eliminate segments that are statistically too sparsely filled and contribute little to plugging. Only segments whose node density values ​​are not lower than the corresponding density threshold are retained for the next comparison. Next, the system compares segments that have passed the first stage with the connectivity threshold sequence in terms of connectivity. The connectivity threshold sequence defines the minimum allowed intersection complexity under the current grouting strategy. Segments with excessively low connectivity mean that even if the density is acceptable, it will be difficult to form multi-directional grout flow, and therefore they will also be eliminated. Only connected segments that simultaneously satisfy both the node density value sequence and the connectivity value sequence being not lower than the connectivity threshold sequence are written into the first-level candidate sequence table.

[0046] After the dual screening is completed, the entries in the first-level candidate sequence list are then sorted in ascending order according to the position of their first node in the node sequence. This ensures that the path generated in the next stage is consistent with the topological order of the original tree structure, avoiding disruption of the continuous propagation from the root node due to rearrangement. Once sorting is complete, the system stream-assembles the ordered candidate fragments to generate the first-level blocking tree path, simultaneously writing it into the first-level blocking tree path sequence and setting a first-level blocking tree path generation flag in the buffer. This flag not only records the completion time of the path but also serves as a trigger signal for subsequent symmetric hierarchical reconstruction. Through the parallel determination of density and connectivity, and the dynamic progressive threshold method described above, the path construction unit compresses a massive number of primary branches into a compact, highly fluid first-order backbone network, laying a stable and concise geometric framework for subsequent symmetric reconstruction.

[0047] Furthermore, the path construction unit immediately initiates a full traversal of the path after the first-level blocking tree path generation flag is set. The traversal process unfolds along the physical connection sequence of the node index sequence. For each node visited, the node index is written into the traversal sequence list, and the relative level and lateral expansion of the node in the tree structure are detected in real time: the branch depth is obtained by counting the number of connection segments traversed along the path from the root node to the current node, and the cumulative length of nodes within the same level in the horizontal direction is quantized as the branch width. These two sets of data are written into the branch depth sequence and the branch width sequence, respectively, and kept synchronized with the traversal sequence list. Thus, the traversal sequence list records the node order from the root to the end, the branch depth sequence characterizes the depth of the vertical hierarchy, and the branch width sequence reflects the lateral bifurcation tension.

[0048] Once the traversal reaches the final node, the system enters the symmetric hierarchical reconstruction phase. First, a reverse copy operation is performed from the beginning to the end of the traversal sequence, writing the result into the reverse traversal sequence. This sequence is logically a mirror image of the original traversal sequence, and its subsequent combination with the forward sequence creates a spatial correspondence. Next, the system groups the branch depth sequence and branch width sequence into odd and even groups: elements with odd indices are grouped into the odd array, and elements with even indices into the even array. The odd array is preserved in its original order; the even array is reversed, rearranging its elements in the opposite direction. This ensures that the vertical and horizontal geometric features are staggered and symmetrical in the mirror mapping, avoiding branch overlap caused by simple mirroring.

[0049] Next, the path construction unit realigns the three sequences—the reverse traversal sequence, the branch depth sequence after odd-even grouping rearrangement, and the branch width sequence—according to the node index sequence to generate a reconstruction instruction sequence. The reconstruction instruction sequence corresponds to the node selection order in the row attribute and carries three types of instructions in the column attribute: reverse index, correction depth, and correction width. Based on this, the system rearranges the first-level blocking tree path line by line: for each connecting segment, the node order is first mirrored according to the reverse index, and then the relative positions of the branches in the depth and lateral directions are adjusted using the correction depth and correction width, ultimately forming a second-level blocking tree path with balanced depth, symmetrical branches, and staggered intersection points. Once this path is completed, it is written into the second-level blocking tree path sequence and simultaneously written with a second-level blocking tree path generation time stamp, providing a timestamp for subsequent grouting units.

[0050] At this point, the path construction unit reads the first and last nodes according to the second-level blocking tree path sequence and writes them into the grouting channel boundary sequence, using the two end nodes as the geometric boundaries of the grouting channel. Finally, the system generates a grouting channel index sequence between the grouting channel boundary sequences according to the second-level blocking tree path sequence, ensuring that each node obtains a unique index position in the linear channel. This guarantees that the subsequent linear sorting, quartile cyclic segmentation, and domain segment alternation operations of the multi-domain co-directional frequency conversion grouting unit can strictly proceed from the first end to the last end of the channel, achieving continuous grouting of the entire fracture network according to the symmetrically optimized topology. This eliminates the risk of pressure concentration and material stagnation that may be caused by irregular geometric deformation, thereby improving the overall blocking efficiency and the consistency of grouting quality.

[0051] Furthermore, the multi-domain co-directional frequency conversion grouting unit immediately retrieves the second-level blocking tree path generation time marker after the path construction unit writes it, confirming that the final version of the grouting channel has been frozen. The system then reads all nodes of the second-level blocking tree path sequence corresponding to the marker at once, extracting the node sequence number, azimuth sequence, scan batch identifier sequence, pulse echo peak value sequence, and pulse echo duration sequence field by field, and writes them completely into the grouting unit's running buffer to form a five-dimensional synchronous dataset. After writing all nodes, the system sets the path import completion marker in the buffer and records the path import completion time marker, providing a time breakpoint for any subsequent incremental adjustments.

[0052] Next, the grouting unit performs a stable sorting of the running buffer according to the ascending order of the node sequence, eliminating any local intersections that may have occurred during the previous symmetrical hierarchical reconstruction, thus restoring the strict monotonicity of the path on the one-dimensional index. The sorting output is written to the linear node sequence, and then the system copies the entire linear node sequence to the linear path index sequence, appending a time stamp generated by the linear path index sequence to bind the current round of linearization processing to the time flow. At this time, the grouting unit records the first and last node indices of the linear path index sequence in the running buffer and writes them to the path boundary record sequence, ensuring that the grouting start and end points can be quickly located at any given time.

[0053] The system reads the total number of nodes in the linear path index sequence and calls the quartile cyclic segmentation process. This process first performs a node mean scan across the entire index sequence, recursively calculating the mean of node indices during cumulative traversal to obtain a node mean sequence. Subsequently, the system automatically derives the quartile node coordinates based on the node mean sequence: the first quartile, the second quartile, and the third quartile are considered segmentation markers, which, together with the first and last nodes, constitute the set of boundary points for the four domain types. Next, the algorithm starts from the first node, labeling the segment before the first quartile coordinates as the pressure domain; the segment from the first quartile to the second quartile as the viscosity domain; the segment from the second quartile to the third quartile as the velocity domain; and the segment from the third quartile to the last node as the time domain. After the first round of segmentation, the algorithm immediately jumps back to the first node and resumes the same sequence, continuing to label nodes that have not yet been assigned to a domain type. After each cycle is completed, the system automatically accumulates the total number of alternating domain segments and checks whether the order of the four domain segments within the same cycle maintains the strict cycle of the pressure domain, viscosity domain, rate domain, and time domain. If a misalignment is detected, the algorithm will use a moving window strategy to exchange the domain segment label values ​​of adjacent nodes within the current cycle until the order is restored.

[0054] After all nodes are labeled with their domain segment types, the system writes the final domain segment assignment into the domain segment labeling sequence. The pressure domain is labeled as the pressure domain, the viscosity domain as the viscosity domain, the rate domain as the rate domain, and the time domain as the time domain, ensuring that the different domain segment labels are unambiguous in data structure. The generation time of the domain segment labeling sequence is written simultaneously with its generation, and the cumulative total number of domain segment alternation cycles is recorded in a separate field, providing a cycle counting benchmark for subsequent dynamic frequency conversion logic.

[0055] The design principle of the above process lies in automatically approximating the node density peaks and valleys of the fracture network using the quartile node coordinates under median statistics. This ensures that each domain segment is approximately equal in spatial length, thereby enabling the four physical control actions—pressure loading, viscosity adjustment, flow rate variation, and time extension—to have equidistant effects on the fracture channel. The existence of multiple cycles ensures that even with highly uneven node density, residual uncovered nodes can still be captured by subsequent cycles. The sequence-maintaining strategy minimizes the rearrangement cost through a moving window, avoiding the tearing of established domain segment continuity under high alternation counts. As the domain segment labeling sequence and alternation count are locked, the multi-domain co-directional frequency conversion grouting unit can call the corresponding pressure loading curve, grout viscosity formula, flow pulse parameters, and curing time window for different domain segments in subsequent stages. This achieves serialized and rhythmic multi-domain co-directional frequency conversion grouting control, ensuring unidirectional propagation while periodically changing the grouting physical parameters to adapt to differences in fracture geometry and wall energy absorption characteristics, thus achieving overall balanced sealing of water-conducting fractures.

[0056] Furthermore, the multi-domain co-directional frequency conversion grouting unit first reads the linear path index sequence and the domain segment label sequence from the running buffer, and couples and maps them in memory so that subsequent node-by-node and domain-by-domain segment alternation operations can be performed. Specifically, the system organizes the linear path index sequence into a one-dimensional array with increasing index values, and maps the domain segment label sequences to the same index one-to-one, treating every four consecutive nodes as a basic window. For each window, the grouting unit reads the domain segment label values ​​of the four nodes in the window in the line order and verifies whether they sequentially represent the four types: pressure domain, viscosity domain, rate domain, and time domain. If the verification passes, it is considered that the current window has completed a complete domain segment alternation; if the verification fails, a moving window exchange mechanism is initiated: within the same window range, the system first attempts to swap the domain segment label values ​​of the first and second nodes in the window, and performs the sequential verification again. If it still does not meet the requirements, it continues to swap the second and third nodes until a complete sequence appears or all adjacent node pairs in the window have been tried to be swapped. This process ensures that the correct domain segment cycle order is restored while minimizing the adjustment range, and avoids cross-influence on adjacent windows.

[0057] Whenever the sequence of four consecutive node segments in a certain window is confirmed or meets the requirements of the pressure, viscosity, rate, and time domains after being corrected through exchange, the system immediately writes a segment alternation completion flag and reads the start and end node indices of the four nodes to form the current segment group boundary sequence entry. The segment group boundary sequence is stored in the grouting unit operation buffer in the form of an ordered list. Each boundary sequence records the start and end positions of the corresponding four-node segment in the entire grouting channel. After writing, the system performs an immediate locking operation on the boundary sequence: first, it copies the current segment group boundary sequence to the grouting operation segment locking sequence and writes a grouting operation segment locking time flag for the entry in the grouting operation segment locking sequence; then, the system checks whether the locking operation is successful and writes a grouting operation segment locking completion flag in the corresponding entry, indicating that the current four-node segment has officially switched to the locked state of performing grouting operations.

[0058] During the parallel advancement of window alternation and instant locking, the multi-domain co-directional frequency conversion grouting unit achieves full coverage labeling and continuous locking of the entire linear node sequence through a moving window strategy and boundary sequence management. The moving window strategy uses a sliding step size of one to sequentially cover the node index interval, ensuring that even in cases of extremely uneven node distribution or severe initial misalignment of domain segment labels, the complete sequence can be repaired within a limited range of attempts through local exchanges. The instant locking operation switches each successfully alternating window segment to the grouting operation state, thereby ensuring that subsequent grouting material composites only perform pressure, viscosity, rate, and time domain control actions on the corrected and confirmed areas when reading the path boundaries, avoiding grouting short circuits or material loss on window segments that have not completed alternation or are not locked. Throughout the process, the alternation completion flag, the grouting operation section locking time flag, and the grouting operation section locking completion flag together constitute the closed-loop control chain of the multi-domain co-directional frequency conversion grouting unit: after each complete sequence confirmation, the flag system immediately feeds back to the grouting execution module, driving the corresponding physical parameter switching, providing accurate start and end coordinates and timestamps for grouting actions in different domains, ensuring the traceability and controllability of grouting operations, and ultimately achieving efficient, balanced, and precise sealing along the second-level sealing tree path.

[0059] Figure 2 The working principle and experimental results of the data detection unit in the adaptive grouting precision sealing device for water-conducting fractures are shown. Figure 2As shown, the acoustic pulse detector is located at the center of the rock mass in the area to be sealed, and comprehensively probes the entire area through multi-azimuth scanning. A complex network of fractures exists within the rock mass, including major horizontal water-conducting fractures and vertically connecting fractures. These fractures intersect to form a three-dimensional network structure, providing channels for groundwater flow. The acoustic pulse detector rotates sequentially according to a preset scanning azimuth sequence number, starting from zero azimuth and increasing in increments at fixed azimuths to form a complete azimuth cycle. The eight main scanning azimuths shown are 0° azimuth 4a, 45° azimuth, 90° azimuth, 135° azimuth, 180° azimuth, 225° azimuth, 270° azimuth, and 315° azimuth. At each azimuth angle, the acoustic pulse detector continuously emits multiple acoustic pulses, which propagate within the rock mass and are reflected at the fracture interfaces. The pulse echo reflection points mark the interaction positions between the acoustic pulses and the fracture structure; the distribution of these reflection points directly reflects the geometric characteristics and spatial distribution of the fracture network. The data detection unit records four key data sequences in parallel: pulse emission sequence number, pulse echo peak value sequence, pulse echo duration sequence, and pulse echo arrival time sequence. The pulse echo peak value sequence shows the variation in echo signal intensity at different reflection points, reflecting the fracture opening and filling state. The pulse echo duration sequence displays the duration distribution of each reflected signal in a bar chart format, used to determine the fracture's extension and connectivity. The pulse echo arrival time sequence records the arrival time of each echo signal, providing a precise time reference for fracture depth localization. Each scanning azimuth generates an independent scan batch identifier, ensuring that the raw acoustic pulse data of the fracture acquired at different azimuth angles can be accurately classified and subsequently processed. Through this multi-azimuth integrated scanning method, the data detection unit can obtain complete three-dimensional information of the fracture network within the area to be sealed, providing a reliable data foundation for subsequent layered stitching and candidate node annotation.

[0060] Figure 3 The working principle of the multi-domain unidirectional frequency conversion grouting unit and the experimental results of alternating domain grouting in an adaptive grouting precision sealing device for water-conducting fractures are shown. Figure 3As shown, the multi-domain co-directional frequency conversion grouting unit converts the second-level sealing tree path into a linear path index sequence. This sequence is generated according to a linear sorting principle from the first end to the last end, providing the basic data structure for subsequent domain segmentation and grouting operations. The multi-domain co-directional frequency conversion grouting unit calls the quartile cyclic segmentation process to perform node mean scanning on the linear path index sequence. Using the quartile node coordinates recorded in the node mean sequence as the domain segment boundary points, it sequentially labels four domain segments: pressure domain, viscosity domain, rate domain, and time domain. As shown in the figure, the four domain segments appear continuously and cyclically in a fixed order in the linear path index sequence, ensuring that the domain segment order remains unchanged in each cycle. When the last node is segmented, it immediately jumps back to the first node, realizing multiple cycles until all nodes are labeled with the corresponding domain segment type. The grouting material composite, as the core execution unit of the grouting operation, achieves precise control of the grouting pipelines of different domain segments through a built-in control valve. The pressure domain pipeline, viscosity domain pipeline, velocity domain pipeline, and time domain pipeline correspond to four different grouting parameter control modes. Each pipeline is directly connected to the grouting material composite, ensuring that the grouting material can be accurately delivered according to the preset domain characteristics. Inside the grouting operation section, the complex fracture network structure includes major water-conducting fractures and connecting fractures, forming a three-dimensional seepage channel network. The domain alternation process ensures that the pressure domain grouting point, viscosity domain grouting point, velocity domain grouting point, and time domain grouting point are precisely grouted at key locations in the fracture network. The grouting point of each domain is connected to the grouting material composite through the corresponding domain pipeline, realizing parameterized grouting control. The current domain group boundary sequence marks the start node index and end node index of each complete cycle, providing precise operational boundaries for the domain alternation process. After each complete four-domain sequential cycle, the system generates a domain alternation completion flag and then performs an immediate locking operation on the current domain group boundary sequence, generating a grouting operation section locking sequence and a dredging operation section locking completion flag. This alternating domain mechanism ensures the coordination of grouting operations in time and space. Through continuous cycling of the pressure domain, viscosity domain, rate domain, and time domain, it achieves comprehensive coverage and precise sealing of complex fracture networks.

[0061] Taking a practical application as an example, this demonstrates the complete implementation process of the invention from detection to grouting. For simplicity, assume scanning is performed at four azimuth angles, with three acoustic pulses emitted at each angle; therefore, the total number of original acoustic pulse data entries generated is... .

[0062] The launch sequence number is recorded as corresponding batch identifier azimuth sequence .

[0063] Each pulse records the peak value. (Unit), Duration (milliseconds), the specific value is set as:

[0064] ;

[0065] .

[0066] According to the scan batch identifier Sort the original data in root order (i.e., keep the sequence number). (Unchanged), resulting in a root-order sorted sequence. Within the same sequence, sort by pulse echo duration. With threshold The first layer division is performed in milliseconds: .

[0067] Within each echo layer, further based on the pulse echo peak value... With threshold Perform a second level division: .

[0068] Assign segmented index numbers to each sub-level entry. ; and the original launch sequence number The correlation ultimately outputs a continuously increasing sequence of fracture segments. .

[0069] To screen candidate nodes for grouting, a dynamic threshold needs to be calculated:

[0070] ;

[0071] .

[0072] For each segment index Perform dual threshold level difference judgment: .

[0073] This yields the set of candidate node indices. .

[0074] The corresponding batch identifier Azimuth Peak Duration Combine the entries into a node mapping list, and generate a node order list in ascending order of azimuth. .

[0075] The path construction unit writes the above five data elements into a buffer and records an initialization completion flag. It then generates a path start index sequence in ascending order of node numbers. Take the first term from the root node. Its corresponding azimuth angle The three-way recursive generator thus generates three sets of direction mapping intervals: Root node Starting from this point, sequentially search for the first node that satisfies the mapping conditions: find the node within the lateral interval. ( , Write the lateral pointer; for the node Repeat the above process to obtain the node pointed to by the lateral pointer. Then for the nodes Executing the mapping reveals that the forward pointer points to a node. ( ); to node With no successor pointer, the recursion ends. The connection results are sequentially written into the blockade tree path sequence. And record the total number of nodes. .

[0076] Perform a full traversal of the tree path and count the connected segments. The number of nodes is Calculate density ; and during the traversal, for nodes The connectivity is obtained by summing the number of forward and backward connections respectively. .

[0077] Set density threshold , connectivity threshold Then all fragments are retained, and a first-level blocking tree path is generated. .

[0078] Then, a symmetrical hierarchical reconstruction is performed on the first-level path: first, the traversal order is recorded. Branch depth With branch width Then copy the traversal order in reverse to get Group the depth and width into odd and even groups respectively, keeping the odd groups in their original order and reversing the even groups. Finally, align the results with the reverse traversal sequence and output the second-level path. .

[0079] Read its first and last nodes to generate the grouting channel index sequence .

[0080] The multi-domain co-directional frequency conversion grouting unit locates the channel, extracts all nodes and their attributes, writes them to the running buffer, and records the path import completion flag. The buffer contents are then sorted in ascending order of node number to obtain a linear node sequence. Write the linear path index sequence and record the generation flag; read the first and last indices. Write the path boundaries. Set the total number of nodes. Perform quartile cyclic partitioning: .

[0081] The corresponding node indices are respectively the first Item. The domain segment types are sequentially labeled from the first node: .

[0082] A single loop covers all nodes; the number of loops is denoted as . During the segment alternation phase, let the sliding window length be... The first window contains all nodes, which are then verified sequentially to ensure they meet the requirements. Write an alternation completion flag once, and set the window start and end indices. Write the current domain segment group boundary sequence, copy it to the grouting operation segment locking sequence, and record the locking completion flag. Since there is only one set of windows, the alternation and locking are completed in one go. Then, perform grouting on the locked segment. Set the domain sequence number. , for the first Physical parameters are applied sequentially to each window node.

[0083] ;

[0084] ;

[0085] ;

[0086] .

[0087] when At the node injection ; At the node injection And so on until... After grouting is completed, the system monitors the peak value and duration of the section using acoustic echo re-measurement. If the peak value drops by more than [a certain amount], [the system will take action]. Or the duration is shortened by more than If the grouting is successful, the sealing is considered successful; otherwise, grouting is automatically repeated in the same section until the sealing conditions are met. At this point, the adaptive grouting method for precisely sealing water-conducting fractures is complete.

[0088] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. An adaptive grouting device for precise sealing of water-conducting fissures, characterized in that, The device includes: a data detection unit, a path construction unit, and a multi-domain co-directional frequency conversion grouting unit. The data detection unit uses an acoustic pulse detector to perform multi-directional scanning of the area to be sealed, obtaining raw acoustic pulse data of the fractures. It then generates a fracture segment sequence through layered splicing, marking grouting candidate nodes in the fracture segment sequence to form a node mapping list. The path construction unit reads the node mapping list, calls a three-way connection recursive device to connect all grouting candidate nodes to form a sealing tree path, performs incremental hierarchical judgment on the sealing tree path, and divides the path according to node density and connectivity to obtain a first-level sealing tree path. The traversal result of the shape path triggers symmetrical hierarchical reconstruction, outputs the second-level blocking tree path, and uses this second-level blocking tree path as the final grouting channel; the multi-domain co-directional frequency conversion grouting unit is used to divide each second-level blocking tree path into four domains: pressure domain, viscosity domain, velocity domain, and time domain; the second-level blocking tree path is linearly sorted from the beginning to the end to generate a linear path index sequence; the four domains are alternated according to the linear path index sequence, so that the pressure domain, viscosity domain, velocity domain, and time domain appear continuously in a cycle. After each cycle is completed, the grouting operation section corresponding to the current domain combination is locked, and the grouting operation is performed through the grouting material composite.

2. The adaptive grouting water-conducting fracture precision sealing device as described in claim 1, characterized in that, The data detection unit initializes the acoustic pulse detector based on the reference time value, drift compensation value, and background value of the acoustic pulse detector. Then, starting from the zero-degree azimuth position, the acoustic pulse detector is rotated sequentially with a fixed azimuth increment to form a complete azimuth cycle. At each azimuth angle, the acoustic pulse detector continuously emits multiple acoustic pulses, and records in parallel the pulse emission sequence number, pulse echo peak sequence, pulse echo duration sequence, and pulse echo arrival time sequence to obtain the original acoustic pulse data of the fracture. All azimuth angles form an azimuth angle sequence. A scan batch identifier is generated for each azimuth angle, and the scan batch identifier is written into the corresponding sequence of the original acoustic pulse data of the fracture.

3. The adaptive grouting water-conducting fracture precision sealing device as described in claim 2, characterized in that, The data detection unit sorts the raw acoustic pulse data of the fracture according to the scanning batch identifier to form a root order sorting sequence; the root order sorting sequence is divided into layers for the first time according to the pulse echo duration sequence to generate the first fracture echo layer; inside the first fracture echo layer, the second layer is divided according to the pulse echo peak sequence to generate the second fracture echo sub-layer. The layered splicing process assigns a segment index number to each fracture echo sublayer and associates this number with the original pulse emission sequence number, ultimately outputting a continuously increasing fracture segment sequence.

4. The adaptive grouting water-conducting fracture precision sealing device as described in claim 3, characterized in that, The data detection unit traverses each segment index number according to the fracture segment sequence and calls the candidate node determination process; The candidate node determination process employs a dual-threshold differential judgment strategy: first, the pulse echo peak sequence is compared with the dynamic peak threshold sequence; second, the pulse echo duration sequence is compared with the dynamic duration threshold sequence. If both comparisons are satisfied, the current segment index number is marked as a grouting candidate node. A node labeling status sequence is written for each grouting candidate node, and a node sequence number is generated. All grouting candidate nodes are filtered from the node labeling status sequence, and the corresponding segment index number, scanning batch identifier, pulse echo peak sequence, pulse echo duration sequence, and azimuth sequence are combined into a node mapping list item. The node mapping list items are reordered according to the ascending order of the azimuth sequence to generate a node sequence list. Based on the node order list, create a node mapping list and write each node mapping list entry into a separate entry in the node mapping list.

5. The adaptive grouting water-conducting fracture precision sealing device as described in claim 4, characterized in that, The path construction unit extracts the node sequence number, scan batch identifier, azimuth sequence, pulse echo peak value sequence, and pulse echo duration sequence for each node mapping list entry and writes them into the path construction buffer. It generates a path start index sequence based on the ascending order of the node sequence number and records the initialization completion flag and initialization completion time flag in the path construction buffer. It selects the first node of the node sequence number in the path construction buffer as the root node. The three-way recursive reader synchronously reads the azimuth sequence value of the root node and generates three sets of direction mappings: forward direction mapping, lateral direction mapping, and longitudinal direction mapping. The three-way recursive reader searches the path construction buffer for the first node whose sequence number increases consecutively and whose azimuth sequence difference is within the same direction interval, according to the forward direction mapping. The node is written to the positive pointer; the three-way recursive recursive recursively retrieves the first node that satisfies the condition that the azimuth sequence difference is in the lateral interval and the pulse echo peak sequence is higher than the root node's pulse echo peak sequence, and writes it to the lateral pointer; the three-way recursive recursively recursively retrieves the first node that satisfies the condition that the scan batch identifier sequence is equal to the root node's scan batch identifier sequence and the pulse echo duration sequence is higher than the root node's pulse echo duration sequence, and writes it to the vertical pointer; the three-way recursive recursively calls itself on the positive pointer, lateral pointer, and vertical pointer respectively, until all node sequence numbers have been written to at least one arbitrary pointer, and the entire connection result is written to the blocking tree path sequence; the generation time of the blocking tree path sequence and the total number of nodes in the blocking tree path sequence are recorded.

6. The adaptive grouting water-conducting fracture precision sealing device as described in claim 5, characterized in that, The path construction unit traverses the blocked tree path sequence, counts the number of nodes for each connected segment, and writes it into the node count sequence. It generates a node density value sequence in the path construction buffer according to the node count sequence and the node number sequence. During the traversal of the blocked tree path sequence, it records the sum of the forward and backward connections for each node, generating a connectivity value sequence. The incremental level determination process adopts a dual threshold progressive strategy: first, it compares the node density value sequence with the density threshold sequence; second, it compares the connectivity value sequence with the connectivity threshold sequence. If both comparisons are satisfied, the current connected segment is written into the first-level candidate sequence table. The first-level candidate sequence table is reordered in ascending order of the node number sequence to generate the first-level blocked tree path, and the first-level blocked tree path sequence and the first-level blocked tree path generation flag are written into it.

7. The adaptive grouting water-conducting fracture precision sealing device as described in claim 6, characterized in that, The path construction unit performs a full traversal of the first-level blocking tree path, recording the traversal sequence, branch depth sequence, and branch width sequence during the process. After the traversal is completed, a symmetrical hierarchical reconstruction is performed, specifically including: copying the traversal sequence from the beginning to the end in reverse and writing it into the reverse traversal sequence; performing odd-even grouping on the branch depth sequence and branch width sequence respectively, maintaining the order of the odd array and reversing the order of the even array; realigning the reverse traversal sequence, the branch depth sequence after odd-even grouping, and the branch width sequence according to the node number sequence to generate a reconstruction instruction sequence; re-arranging the first-level blocking tree path one by one according to the reconstruction instruction sequence, outputting the final second-level blocking tree path, and writing the second-level blocking tree path sequence and the second-level blocking tree path generation time marker; reading the first node and the last node of the second-level blocking tree path sequence and writing them into the grouting channel boundary sequence; generating the grouting channel index sequence between the grouting channel boundary sequences according to the order of the second-level blocking tree path sequence.

8. The adaptive grouting water-conducting fracture precision sealing device as described in claim 7, characterized in that, The multi-domain co-directional frequency conversion grouting unit locates the second-level blocking tree path generation time marker, reads all nodes of the second-level blocking tree path sequence corresponding to the marker; for each node, it extracts the node sequence number sequence, azimuth sequence, scan batch identifier sequence, pulse echo peak value sequence, and pulse echo duration sequence, and writes them into the grouting unit's operating buffer; it records the path import completion flag and path import completion time flag in the grouting unit's operating buffer; it sorts the grouting unit's operating buffer according to the ascending order of the node sequence number sequence to generate a linear node sequence; it writes the linear node sequence into the linear path index sequence and appends the linear path index sequence generation time marker; it records the first node index and the last node index of the linear path index sequence in the grouting unit's operating buffer and writes them into the path boundary. Record the sequence; read the total number of nodes in the linear path index sequence, and call the quartile cyclic segmentation process, which specifically includes: performing a node mean scan on the linear path index sequence, recording the coordinates of each quartile node in the node mean sequence; starting from the first node, using the quartile node coordinates recorded in the node mean sequence as the domain segment boundary points, sequentially labeling the pressure domain, viscosity domain, rate domain, and time domain; immediately jumping back to the first node after segmenting to the last node, realizing multiple loops until all nodes are labeled with the domain segment type; writing the segmentation results into the domain segment labeling sequence, where the pressure domain is labeled as pressure domain, the viscosity domain as viscosity domain, the rate domain as rate domain, and the time domain as time domain, ensuring that the order of the four domain segments remains unchanged in each loop; recording the generation time of the domain segment labeling sequence and the total number of domain segment alternation loops.

9. The adaptive grouting water-conducting fracture precision sealing device as described in claim 8, characterized in that, The multi-domain co-directional frequency conversion grouting unit reads the linear path index sequence and the domain segment label sequence, and the process of domain segment alternation specifically includes: verifying whether the pressure domain, viscosity domain, rate domain, and time domain appear sequentially in every four consecutive nodes according to the line order; if the verification fails, the domain segment alternation process exchanges the domain segment label values ​​of adjacent nodes within the same segment by moving the window until a complete sequence appears; after each complete sequence is completed, a domain segment alternation completion flag is written; after each domain segment alternation is completed, the start node index and end node index of the current four consecutive nodes are read and written into the current domain segment group boundary sequence; an immediate locking operation is performed on the current domain segment group boundary sequence, specifically including: copying the current domain segment group boundary sequence to the grouting operation section locking sequence; and recording the grouting operation section locking time flag and the grouting operation section locking completion flag.

Images