Tunnel structure health processing method and system based on multi-modal data fusion

Abstract

The application provides a tunnel structure health treatment method and system based on multi-modal data fusion, relates to data processing technology, controls a cooperative collection device to collect a tunnel structure, obtains point cloud data and thermal imaging parameters of a collection position; inputs the point cloud data into a crack identification model for identification, obtains a crack type and crack parameters of a collection graph, filters the crack parameters and thermal imaging parameters based on environmental parameters of the collection position, obtains fusion credible parameters; classifies and analyzes cracks in the collection graph based on the crack type, obtains new grouting holes and grouting sequences of each crack; updates the fusion credible parameters, new grouting holes and grouting sequences to a tunnel twin model at the collection position, generates a tunnel display model, can determine grouting holes adapted to the actual state of cracks, and improves the efficiency of tunnel structure repair.

Description

Technical Field

[0001] This invention relates to data processing technology, and more particularly to a method and system for tunnel structure health processing based on multimodal data fusion. Background Technology

[0002] As a core component of transportation infrastructure, tunnels can currently be monitored for structural health by collecting various types of data within them. Due to the long-term effects of vehicle loads, geological subsidence, temperature and humidity changes, and groundwater erosion, tunnel structures are prone to gradual deterioration. Cracks are the most common problem in tunnel structures; if not repaired with grouting in a timely manner, they can gradually expand, leading to leaks, steel corrosion, and even threatening traffic safety.

[0003] In existing technologies, even when crack information is accurately obtained through multimodal data, standardized fixed patterns are still commonly used when formulating grouting repair plans. For example, the arrangement of grouting holes is often based on uniform spacing or relies on manual experience to select locations. For instance, Chinese patent application CN201911056049.8 discloses a method for reinforcing shallow overburden shield tunnels based on weak strata. This method uses hollow grouting anchors to perform deep-hole grouting reinforcement along the hoisting holes or grouting holes. The grouting holes are located on both sides of the hoisting holes, and the grouting holes on the tunnel segments are evenly distributed across the cross-section of the segments at fixed intervals. This approach fails to consider the actual condition of the cracks, resulting in grouting holes failing to cover critical leakage points or excessive drilling in unnecessary locations that damages the structure. Furthermore, improper grouting sequence can lead to uneven grout distribution and incomplete crack sealing, failing to meet the needs of tunnel structural repair.

[0004] Therefore, how to determine the appropriate grouting holes based on the actual condition of the cracks and improve the efficiency of tunnel structure repair has become an urgent problem to be solved. Summary of the Invention

[0005] This invention provides a method and system for tunnel structure health management based on multimodal data fusion, which can determine the appropriate grouting holes according to the actual state of cracks, thereby improving the efficiency of tunnel structure repair.

[0006] A first aspect of the present invention provides a method for tunnel structure health processing based on multimodal data fusion, comprising:

[0007] The control and collaborative acquisition device acquires data on the tunnel structure, obtaining point cloud data and thermal imaging parameters of the acquisition location.

[0008] The point cloud data is input into the crack identification model for identification, and the crack type and crack parameters of the acquired image are obtained. The crack parameters and thermal imaging parameters are filtered based on the environmental parameters of the acquisition location to obtain fused reliable parameters.

[0009] Based on the crack type, the cracks in the acquired images are classified and customized to obtain new grouting holes and grouting sequences for each crack.

[0010] Based on the data collection location, the fusion reliability parameters, newly added grouting holes, and grouting sequence are updated to the tunnel twin model to generate a tunnel display model.

[0011] Optionally, in one possible implementation of the first aspect, the filtering of crack parameters and thermal imaging parameters based on environmental parameters at the acquisition location to obtain fused reliable parameters includes:

[0012] The environmental sensor is controlled to acquire environmental parameters at the acquisition location, including dust concentration and water mist density;

[0013] The environmental interference degree is calculated based on the dust concentration and water mist density, and the laser weight and infrared weight are determined based on the environmental interference degree.

[0014] Crack parameters are screened based on laser weights to obtain reliable crack parameters, and thermal imaging parameters are screened based on infrared weights to obtain reliable infrared parameters.

[0015] The crack confidence parameter and the infrared confidence parameter are weighted and fused to obtain the fused confidence parameter.

[0016] Optionally, in one possible implementation of the first aspect, the step of calculating the environmental interference degree based on dust concentration and water mist density, and determining the laser weight and infrared weight based on the environmental interference degree, includes:

[0017] The environmental disturbance level can be obtained using the following formula.

[0018]

[0019] in, For environmental interference, Dust concentration, The density of the water mist;

[0020] The laser weight and infrared weight are determined based on the degree of environmental interference.

[0021] The laser weight and infrared weight are obtained using the following formulas.

[0022]

[0023] in, For laser weights, For infrared weights.

[0024] Optionally, in one possible implementation of the first aspect, the weighted fusion of the crack confidence parameter and the infrared confidence parameter to obtain the fused confidence parameter includes:

[0025] The fusion confidence parameter is obtained using the following formula.

[0026]

[0027] in, To integrate trusted parameters, For the crack reliability parameters, These are the infrared reliable parameters.

[0028] Optionally, in one possible implementation of the first aspect, the step of classifying and customizing the cracks in the acquired image based on the crack type to obtain new grouting holes for each crack and a grouting sequence includes:

[0029] Cracks in the acquired images are classified based on crack type, resulting in linear cracks and mesh cracks.

[0030] By performing a box-selection hole location analysis on each of the aforementioned linear cracks, the newly added grouting holes and grouting sequence for each linear crack are obtained;

[0031] The connection points of the network cracks were analyzed to obtain the newly added grouting holes for each network crack, as well as the grouting sequence.

[0032] Optionally, in one possible implementation of the first aspect, the step of performing a box-selection hole location analysis on each of the linear cracks to obtain new grouting holes for each linear crack and a grouting sequence includes:

[0033] The acquired image is processed into coordinates, and the extreme values ​​of the crack coordinates of each line are calculated based on the set movement value to obtain the selected extreme values;

[0034] Based on the selected extreme values, a box selection rectangle corresponding to each line crack is constructed, and a new grouting hole for each line crack is constructed at the midpoint of each rectangle side in the box selection rectangle.

[0035] The newly added grouting holes corresponding to both ends of the linear crack are taken as core holes, and the remaining newly added grouting holes are taken as boundary holes.

[0036] The core holes and boundary holes are sorted according to the crack locations of each line crack to obtain the grouting sequence of each line crack. The crack locations include the top wall location and the side wall location.

[0037] Optionally, in one possible implementation of the first aspect, the step of sorting the core holes and boundary holes according to the crack positions of each line crack to obtain the grouting sequence of each line crack includes:

[0038] By placing the boundary hole in front of the core hole, the initial sequence of cracks in each line is obtained;

[0039] When the crack location is determined to be the top wall location, the corresponding linear crack is taken as the top wall crack, and the initial sequence is taken as the grouting sequence of the top wall crack;

[0040] When the crack location is determined to be a sidewall location, the corresponding line crack is taken as a sidewall crack. If the crack direction of the sidewall crack is transverse, the boundary hole located at the bottom in the initial sequence is adjusted to the first position to obtain the grouting sequence of the transverse sidewall crack.

[0041] If the crack direction of the sidewall crack is longitudinal, the core hole located at the bottom in the initial sequence is adjusted to the front of the core hole located at the top to obtain the grouting sequence of the longitudinal sidewall crack.

[0042] Optionally, in one possible implementation of the first aspect, the analysis of the connection positions of the network cracks to obtain the newly added grouting holes for each network crack and the grouting sequence includes:

[0043] Obtain the two main endpoints of the main cracks and the secondary endpoints of the secondary cracks in the network crack;

[0044] Starting from a main endpoint, connect it to an adjacent but unconnected secondary endpoint on one side of the main gap. Use this secondary endpoint as the current main endpoint and repeat the above steps until all other secondary endpoints on the same side are connected. Then, establish a connection with another main endpoint to generate a connection polygon.

[0045] New grouting holes are constructed at the center points of each side of the connected polygon to create each mesh-like crack;

[0046] The newly added grouting holes are enlarged based on the grouting threshold to obtain grouting zones. The newly added grouting holes are sorted in descending order based on the number of crack pixels in each grouting zone to obtain the grouting sequence of each network crack.

[0047] Optionally, in one possible implementation of the first aspect, it also includes:

[0048] When it is determined that there is an obstruction at the location of the newly added grouting hole, the corresponding newly added grouting hole is taken as the hole to be moved. The outline points of the obstruction are moved based on the hole position radius of the hole to be moved to obtain the offset outline. The obstruction includes the main reinforcement and bolt holes.

[0049] Use the outline of the connecting polygon or the selected rectangle as the wrapping outline, use the intersection of the offset outline and the wrapping outline as the offset position, and move the center point of the hole to be moved to the offset position.

[0050] A second aspect of the present invention provides a tunnel structure health processing system based on multimodal data fusion, comprising:

[0051] The control module is used to control the collaborative acquisition device to acquire data on the tunnel structure, and obtain point cloud data and thermal imaging parameters of the acquisition location.

[0052] The input module is used to input the point cloud data into the crack recognition model for recognition, obtain the crack type and crack parameters of the acquired image, and filter the crack parameters and thermal imaging parameters based on the environmental parameters of the acquisition location to obtain fused reliable parameters.

[0053] The analysis module is used to classify and customize the cracks in the acquired image based on the crack type, and obtain the new grouting holes and grouting sequence for each crack.

[0054] The update module is used to update the fused reliable parameters, newly added grouting holes and grouting sequence to the tunnel twin model according to the acquisition location, and generate a tunnel display model.

[0055] A third aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, is used to implement the first aspect of the present invention and various methods possibly involved in the first aspect.

[0056] The beneficial effects of this invention are as follows:

[0057] 1. This invention collects point cloud data and thermal imaging parameters collaboratively, and selects reliable parameters by combining environmental parameters to avoid the one-sidedness of single data. Then, according to the crack type, it customizes and adds grouting holes and grouting sequences, automatically determines the grouting holes, and makes the grouting holes better cover the cracks. Finally, all information is updated to the tunnel twin model to generate a display model, which is convenient for managers to view intuitively.

[0058] 2. This invention first acquires dust concentration and water mist density using environmental sensors, calculates environmental interference, and dynamically determines laser and infrared weights. Then, it filters crack parameters and thermal imaging parameters according to these weights, eliminating false data caused by environmental interference (such as false cracks caused by dust obstruction or thermal imaging deviations caused by water mist). Finally, it fuses the data using a weighted formula to obtain reliable parameters. This ensures that the fused data accurately reflects the parameters within the tunnel structure.

[0059] 3. This invention uses a rectangular frame to place holes at the midpoints of the edges, and adjusts the grouting sequence based on the crack location and direction to ensure relatively uniform grout coverage. The network of cracks forms polygons by connecting the endpoints of main and secondary cracks, with holes placed at the midpoints of the edges and grouting ordered according to crack density, avoiding omissions of branch cracks. Simultaneously, if the grouting hole encounters obstacles such as main reinforcement bars or bolt holes, it can automatically shift to the intersection of the covering contour and the offset contour of the obstacle, avoiding structural components. This avoids the inefficiency of manually adjusting hole positions and improves the grouting repair effect. Attached Figure Description

[0060] Figure 1 This is a schematic diagram illustrating the application scenario of the technical solution provided by the present invention;

[0061] Figure 2 A flowchart illustrating a tunnel structure health method based on multimodal data fusion provided by this invention;

[0062] Figure 3 A schematic diagram of a core hole provided by the present invention;

[0063] Figure 4 A schematic diagram of a connected polygon provided by the present invention;

[0064] Figure 5 This is a schematic diagram of a tunnel structure health system based on multimodal data fusion provided by the present invention. Detailed Implementation

[0065] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.

[0067] It should be understood that in the various embodiments of the present invention, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0068] It should be understood that in this invention, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0069] It should be understood that in this invention, "multiple" refers to two or more. "And / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "Contains A, B, and C", "Contains A, B, and C" means that all three A, B, and C are contained; "Contains A, B, or C" means that one of A, B, and C is contained; "Contains A, B, and / or C" means that any one, two, or three of A, B, and C are contained.

[0070] It should be understood that in this invention, "B corresponding to A", "B corresponding to A", "A and B correspond", or "B and A correspond" means that B is associated with A, and B can be determined based on A. Determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information. Matching A and B is defined as a similarity between A and B that is greater than or equal to a preset threshold.

[0071] Depending on the context, "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection."

[0072] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0073] like Figure 1The diagram illustrates a scenario of the technical solution provided by this invention. This application scenario includes a server, a collaborative acquisition device, an edge computing device, and a management terminal. The server is communicatively connected to the collaborative acquisition device, the edge computing device, and the management terminal. When health monitoring is required during tunnel operation, the server controls the collaborative acquisition device within the tunnel to perform multimodal data acquisition of the tunnel structure, obtaining point cloud data and thermal imaging parameters at the acquisition location. The collaborative acquisition device consists of a laser scanner and an infrared thermal imager, both mounted on the same bracket. Specifically, the collaborative acquisition device uses a laser scanner to scan along the tunnel arch, generating three-dimensional point cloud data, while simultaneously using an infrared thermal imager to acquire the surface temperature field distribution of the structure. This data is transmitted in real-time to the edge computing device. The terminal's built-in crack recognition model (e.g., integrating the PointNet++ crack recognition model) processes the images, identifying crack types and corresponding parameters (point cloud data and thermal imaging data). At this time, the system simultaneously collects environmental parameters (such as dust and water mist inside the tunnel) to screen the crack parameters and thermal imaging data for reliability, eliminating interference from dust and water mist.

[0074] Based on this, we will dynamically determine the appropriate grouting holes and grouting sequence according to the real-time status of the cracks. Then, we will summarize all the data, update the reliable parameters, crack locations, and grouting hole locations to the positions of the tunnel twin model, and send the resulting tunnel display model to the management terminal for display. The management terminal can be a user's mobile phone, tablet, computer, etc., without any restrictions.

[0075] This invention provides a method for tunnel structure health processing based on multimodal data fusion, such as... Figure 2 As shown, steps S1-S4 are included:

[0076] S1 controls the collaborative acquisition device to acquire data on the tunnel structure, obtaining point cloud data and thermal imaging parameters at the acquisition location.

[0077] It should be noted that existing technologies mostly rely on independent analysis using a single sensor, lacking a dynamic fusion mechanism for multimodal data, leading to incomplete detection results. For example, laser point clouds may miss areas prone to water seepage, while infrared thermography cannot provide precise dimensional information about cracks. Although laser scanners can acquire 3D point cloud data with millimeter-level precision, they only reflect surface morphology and cannot directly identify internal defects such as seepage and voids. For instance, when there is water accumulation behind the tunnel lining, laser point clouds cannot penetrate the surface concrete, causing seepage areas to be misjudged as normal structures. Infrared thermography can locate seepage areas through temperature differences (usually manifested as low-temperature anomalies), but due to the complexity of heat radiation transfer, it is difficult to quantify the geometric parameters of cracks (such as width and depth). For example, a tiny crack with a width of 0.3 mm may not be effectively identified due to thermal conduction attenuation.

[0078] Among them, the collaborative acquisition device refers to the acquisition equipment consisting of a laser scanner and an infrared thermal imager. The two are fixed on the same bracket and can start acquisition synchronously under the control of the server, respectively responsible for generating three-dimensional point cloud data and acquiring thermal imaging parameters.

[0079] Specifically, the server controls a laser scanner to start scanning along the tunnel arch, generating point cloud data that reflects the three-dimensional geometry of the tunnel structure at that location. At the same time, it controls an infrared thermal imager to simultaneously acquire the temperature field distribution on the surface of the structure at that location, obtaining thermal imaging parameters. After the acquisition is completed, the point cloud data and thermal imaging parameters are correlated in real time to the acquisition map of the corresponding acquisition location and transmitted synchronously to the edge computing device. Subsequently, the data is processed and the crack type and parameters are identified through a crack recognition model (a model integrating PointNet++).

[0080] S2, input the point cloud data into the crack identification model for identification, obtain the crack type and crack parameters of the acquired image, and filter the crack parameters and thermal imaging parameters based on the environmental parameters of the acquisition location to obtain fused reliable parameters.

[0081] Understandably, we will input the collected point cloud data into the crack identification model. The training model can be a pre-trained PointNet++ model or an existing neural network model, which will not be elaborated here. The model will automatically identify the crack type and corresponding crack parameters, such as length. At the same time, environmental parameters will be collected, and the environmental interference degree will be calculated to determine the laser weight and infrared weight. The crack parameters and thermal imaging parameters will be filtered, and these weights will be used to generate fused reliable parameters.

[0082] Prior to this, it is not difficult to understand that we would denoise the point cloud. For example, we can use statistical filtering algorithms to remove outliers and correct thermal imaging parameters. For example, we can use non-uniformity correction (NUC) to eliminate differences in infrared detector response, thereby improving the temperature resolution to 0.1℃. Both can be used to simultaneously acquire data from the same area to achieve data alignment. Alternatively, we can use ORB feature matching and Homography transformation to achieve sub-pixel alignment between the point cloud and the thermal image. These are existing technologies and will not be elaborated here.

[0083] Before identification, we pre-train the model. For example, we collect 1000 sets of tunnel crack point cloud data, label the crack type (linear, mesh) and geometric parameters (width, depth), and then expand the dataset to 5000 sets through data augmentation (random rotation, scaling, adding noise) to form a training dataset. We can also optimize the PointNet++ model. For example, the network structure adopts a hierarchical feature learning architecture, with sampling rates set to [0.5, 0.25] and ball query radius set to [0.2m, 0.4m]. In terms of loss function, we combine classification loss (cross-entropy) and regression loss (mean squared error), with a weight ratio of 1:0.5. In terms of training strategy, we use the Adam optimizer with an initial learning rate of 0.001, a batch size of 32, and 200 iterations.

[0084] It should be noted that the tunnel environment is complex, and factors such as dust and water mist can affect the quality of laser and infrared data acquisition, leading to uncertainty in the parameters. This is because environmental interference can cause a single device to output multiple data points, some of which are relatively accurate, while others have significant deviations.

[0085] Therefore, when we obtain the crack parameters and thermal imaging parameters, we will filter the data of both through environmental parameters and finally obtain the fused reliable parameters to update the data of the corresponding crack.

[0086] In some embodiments, step S2 (filtering crack parameters and thermal imaging parameters based on environmental parameters of the acquisition location to obtain fused reliable parameters) includes S21-S24:

[0087] S21, control the environmental sensor to acquire environmental parameters at the acquisition location, the environmental parameters including dust concentration and water mist density.

[0088] Among them, environmental sensors refer to devices that can detect the environmental conditions inside the tunnel in real time, specifically used to collect dust concentration and water mist density.

[0089] It should be noted that dust concentration is a key factor affecting the clarity of point cloud data from laser scanners, while water mist density is a key factor affecting the temperature detection accuracy of infrared thermal imagers.

[0090] Therefore, we will control the environmental sensors to acquire environmental parameters at the collection location, including dust concentration and water mist density.

[0091] S22, the environmental interference degree is calculated based on the dust concentration and water mist density, and the laser weight and infrared weight are determined based on the environmental interference degree.

[0092] It should be noted that traditional methods lack quantitative calculation of environmental interference and rely on experience to set fixed weights (such as laser weight fixed at 0.6 and infrared weight fixed at 0.4), which cannot be dynamically adjusted according to the real-time environment and are difficult to adapt to the changing environment inside the tunnel (such as water mist at the tunnel entrance and dust in the middle section).

[0093] By calculating the degree of environmental interference, we convert the impact of dust and water mist into quantifiable values, and then dynamically determine the laser weight and infrared weight based on these values, so that the weights can accurately match the real-time environmental interference situation.

[0094] In some embodiments, step S22 (calculating the environmental interference degree based on dust concentration and water mist density, and determining the laser weight and infrared weight based on the environmental interference degree) includes:

[0095] The environmental disturbance level can be obtained using the following formula.

[0096]

[0097] in, For environmental interference, Dust concentration, The density of the water mist;

[0098] Laser weights and infrared weights are determined based on environmental interference levels, and dust concentration is also considered. The concentration of suspended particulate matter (mg / m³) inside the tunnel is measured in real time by a dust sensor, along with the water mist density. It is the volume density (unit: g / m³) of the spray system or natural water mist, calculated by a humidity sensor in combination with the droplet size distribution, and 2×(dust concentration + water mist density + 1) is used for normalization to ensure that the formula output range is [0, 0.5]. The larger the value, the more serious the environmental interference.

[0099] The laser weight and infrared weight are obtained using the following formulas.

[0100]

[0101] in, For laser weights, 0.7 represents the infrared weight, and 0.7 represents the basic weight of the laser in the absence of interference, reflecting its high-precision geometric measurement capability in a clean environment (such as point cloud accuracy at the level of 0.1 mm). This is an inverse attenuation term for environmental interference. The higher the interference, the lower the laser weight. 0.3 is the minimum weight guarantee to ensure that even in a high-interference environment, the laser data still retains 30% contribution (avoiding complete reliance on thermal imaging data). The lower the environmental interference (the less dust), the higher the laser weight, and the more reliable it is.

[0102] S23. Based on laser weighting, crack parameters are screened to obtain reliable crack parameters. Based on infrared weighting, thermal imaging parameters are screened to obtain reliable infrared parameters.

[0103] It should be noted that both the collected crack parameters and thermal imaging parameters are input into the edge computing device (the on-site data processing device). The reliability of the data is automatically calculated using a reliability algorithm. Regarding the reliability calculation of crack parameters, after the laser scanner collects the raw point cloud data, it transmits the data to the edge computing device. The device automatically extracts three key features (point cloud density, neighborhood consistency, and scanning distance angle) for each point cloud data using a reliability calculation algorithm. Then, it calculates the reliability of a single data point (e.g., 75%, 25%) according to preset weights (e.g., density 30%, consistency 35%, distance angle 10%, and the remaining 25% may be related to the device's own accuracy calibration parameters). Similar to laser point clouds, the reliability of infrared thermal imaging data is also automatically calculated based on data quality features. The extracted key features can be temperature consistency, comparison accuracy, etc. These are existing technologies and will not be elaborated here. In other words, the reliability of each parameter can be calculated directly using existing technologies.

[0104] Understandably, the server uses laser weights to filter the credibility of crack parameters to obtain reliable crack parameters, and uses external weights to filter the credibility of thermal imaging parameters to obtain reliable infrared parameters.

[0105] For example, laser weight =0.708, infrared weight =0.65, comparing the reliability of the original laser data: Data A (0.4mm): Reliability 75% ≥ 70.8%, meets the criteria, selected; Data B (0.6mm): Reliability 25% < 70.8%, does not meet the criteria, discarded. The final "reliable laser data" is: 0.4mm (crack width).

[0106] It needs to be explained why laser weighting is used. and infrared weight The reason for filtering, rather than directly selecting the highest confidence level, is that we want to select and retain better data. For example, selecting the highest confidence level under the same environment (such as selecting 60 points on a smoggy day) might result in data that is not actually sufficient being considered valid data. However, by using an environmental threshold for filtering (such as setting the threshold to 70 points on a smoggy day and 50 points on a sunny day), the 60-point data will be filtered out, thus avoiding the retention of invalid data in harsh environments and ensuring that only truly usable data is retained. If no data is found after filtering, the system will continue to detect data until data is available.

[0107] S24, the crack confidence parameter and the infrared confidence parameter are weighted and fused to obtain the fused confidence parameter.

[0108] In some embodiments, step S24 (weighted fusion of the crack confidence parameter and the infrared confidence parameter to obtain a fused confidence parameter) includes:

[0109] The fusion confidence parameter is obtained using the following formula.

[0110]

[0111] in, To integrate trusted parameters, For the crack reliability parameters, For infrared reliable parameters, it's easy to understand that reliable data is extracted from the raw laser and infrared data separately, and then fused using weights. It is used to carry trusted fusion information.

[0112] S3. Based on the crack type, perform customized analysis on the cracks in the acquired image to obtain the newly added grouting holes and grouting sequence for each crack.

[0113] In some embodiments, step S3 (classifying and customizing the cracks in the acquired image based on the crack type to obtain new grouting holes for each crack, and the grouting sequence) includes S31-S33:

[0114] S31. Based on the crack type, the cracks in the acquired image are classified into linear cracks and mesh cracks.

[0115] It should be noted that different types of cracks differ significantly in morphology, propagation path, and impact on structural stability.

[0116] S32, perform a box selection hole location analysis on each of the described line cracks to obtain the newly added grouting holes for each line crack, as well as the grouting sequence.

[0117] It should be noted that linear cracks typically extend in a certain direction, and the arrangement of grouting holes must cover critical locations of the crack (such as both ends and the middle) to ensure sufficient grout penetration. Traditional methods rely on experience. Our approach involves systematically determining the grouting hole locations using a rectangular selection method and developing a reasonable grouting sequence based on the crack's location (top or side wall) and direction (lateral or longitudinal), thereby improving grouting efficiency and coverage quality.

[0118] In some embodiments, step S32 (performing a box-selection hole location analysis on each of the linear cracks to obtain the newly added grouting holes for each linear crack, and the grouting sequence) includes S321-S324:

[0119] S321: The acquired image is processed into coordinates. The extreme values ​​of the crack coordinates of each line are calculated based on the set movement value to obtain the selected extreme values.

[0120] It should be noted that if the selected area is too close to the edge of the crack, the grouting holes constructed later may cause the crack to expand and the grouting effect to be poor.

[0121] Therefore, after the coordinate processing of the acquired image, the preset movement value will be retrieved, which can be set manually according to the actual situation.

[0122] The selection extreme values ​​are calculated based on the coordinate extreme values ​​and the set movement values. For example, the four coordinate extreme values ​​of the selection rectangle boundary (the maximum and minimum values ​​of the horizontal and vertical coordinates of the crack) are determined. Then, the selection extreme values ​​are calculated with the set movement values, for example, X1=Xmin-movement value, X2=Xmax+movement value, Y1=Ymin-movement value, Y2=Ymax+movement value), thus obtaining the selection extreme values. Subsequently, the corresponding four coordinate points (X1,Y1), (X2,Y1), (X1,Y2), and (X2,Y2) can be obtained to determine the selection rectangle.

[0123] S322, Based on the selected extreme values, construct a selection rectangle corresponding to each line crack, and construct a new grouting hole for each line crack at the midpoint of each rectangle side in the selection rectangle.

[0124] It should be noted that traditional grouting hole placement often relies on experience, with points manually selected near the crack. This results in uneven hole distribution, potentially leading to areas with no holes or overly dense hole placement. If holes are only placed around the crack itself, a regular coverage area is not formed, making it impossible to ensure uniform diffusion of grout to the crack and surrounding structure. Therefore, limiting the grouting holes to the midpoint of a rectangular boundary allows them to evenly surround the crack, ensuring that the grout diffuses from the perimeter towards the center of the crack, covering the entire crack area, while avoiding duplicate or missed holes.

[0125] It's easy to understand that, based on the selected extreme values, a bounding rectangle corresponding to each line crack is constructed. Then, new grouting holes for each line crack are created at the midpoints of the sides of this bounding rectangle. For example, using the four selected extreme values ​​(X1, X2, Y1, Y2) as vertex coordinates, a bounding rectangle corresponding to the line crack is constructed in the coordinate-based acquisition image. The midpoint coordinates of the four sides of this bounding rectangle are calculated: top midpoint ((X1+X2) / 2, Y2), bottom midpoint ((X1+X2) / 2, Y1), left midpoint (X1, (Y1+Y2) / 2), and right midpoint (X2, (Y1+Y2) / 2). These four midpoint positions are then marked as new grouting holes for each line crack, with each midpoint corresponding to one hole location.

[0126] S323, the newly added grouting holes corresponding to both ends of the crack in the line are taken as core holes, and the remaining newly added grouting holes are taken as boundary holes.

[0127] It's easy to understand that the two ends of the crack correspond to the core holes, and the rest are boundary holes. (See [reference]). Figure 3 The newly added grouting holes at both ends along the length shown are the core holes.

[0128] S324, The core holes and boundary holes are sorted according to the crack positions of each line crack to obtain the grouting sequence of each line crack. The crack positions include the top wall position and the side wall position.

[0129] It should be noted that during grouting, the grout tends to flow downwards due to gravity. If the grout is not arranged properly, it may cause the upper part of the grout to seep down, resulting in a void at the top and making it impossible to effectively seal the cracks.

[0130] In some embodiments, step S324 (sorting the core holes and boundary holes according to the crack positions of each line crack to obtain the grouting sequence of each line crack) includes S3241-S324:

[0131] S3241, place the boundary hole in front of the core hole to obtain the initial sequence of cracks in each line.

[0132] It should be noted that the core holes correspond to both ends of the crack (the key areas for inhibiting crack propagation), while the boundary holes correspond to the periphery of the crack (helping to form a grout enclosure area). If the core holes are injected first, the grout may diffuse and flow into the surrounding unconstrained areas, resulting in a weakened reinforcement effect. However, injecting the boundary holes first can form a grout barrier around the crack, constraining the diffusion range of the grout from the subsequent core holes, and ensuring that the grout from the core holes can concentrate and fill both ends of the crack. In other words, the left and right boundary holes are grouted and fixed first to prevent the grout from running around during the top and bottom grouting.

[0133] Therefore, the boundary hole is placed in front of the core hole to obtain the initial sequence of cracks in each line.

[0134] S3242, when the crack location is determined to be the top wall location, the corresponding line crack is taken as the top wall crack, and the initial sequence is taken as the grouting sequence of the top wall crack.

[0135] It is easy to understand that if the crack is located at the top wall, that is, at the top of the tunnel, the grout will not slide down due to gravity. Therefore, the initial sequence can be directly used as the grouting sequence for the top wall crack, that is, the boundary hole is grouted first, followed by the core hole.

[0136] S3243, when the crack location is determined to be a sidewall location, the corresponding line crack is taken as a sidewall crack. If the crack direction of the sidewall crack is transverse, the boundary hole located at the bottom in the initial sequence is adjusted to the first position to obtain the grouting sequence of the transverse sidewall crack.

[0137] It should be noted that when grouting transverse cracks (extending horizontally) in the tunnel sidewall, the grout tends to flow downwards due to gravity, causing the upper part of the grout to seep down, resulting in a void in the upper part and making it impossible to effectively seal the crack.

[0138] Therefore, when the crack is a sidewall crack and the length direction of the crack is transverse, the lower hole should be grouted first. Then the lower boundary hole in the initial sequence is adjusted to the first position to obtain the grouting sequence for the transverse sidewall crack.

[0139] S3244, if the crack direction of the sidewall crack is longitudinal, adjust the core hole located at the bottom in the initial sequence to the front of the core hole located at the top, so as to obtain the grouting sequence of the longitudinal sidewall crack.

[0140] Similarly, when the crack direction of the sidewall crack is longitudinal, the core hole located at the bottom in the initial sequence is adjusted to the front of the core hole located at the top to obtain the grouting sequence of the longitudinal sidewall crack.

[0141] S33, perform connection hole analysis on the network cracks to obtain the newly added grouting holes for each network crack, as well as the grouting sequence.

[0142] It should be noted that the network crack structure is complex, containing main cracks and multiple branch cracks. Traditional uniform hole placement methods are difficult to cover all crack branches, easily leading to uneven grout distribution. Our method uses a connecting polygon approach to incorporate main cracks and branch cracks into a unified hole placement plan, ensuring that the grouting holes can cover the entire crack network and improve the repair effect.

[0143] The subsequent steps involve identifying the main and secondary cracks within the network of cracks, determining the main and secondary endpoints; starting from one main endpoint, connecting adjacent secondary endpoints sequentially, and finally connecting to the main endpoint at the other end to form a connecting polygon; additional grouting holes are then installed at the midpoints of each side of the polygon.

[0144] In some embodiments, step S33 (analyzing the connection points of the network cracks to obtain the newly added grouting holes for each network crack, and the grouting sequence) includes S331-S334:

[0145] S331, Obtain the two main endpoints of the main crack and the secondary endpoints of the secondary crack in the network crack.

[0146] In this context, the main crack refers to the longest or widest crack in the network of cracks, typically the backbone of the crack network, i.e., the main crack. The remaining cracks are called secondary cracks. The main endpoints refer to the two endpoints of the main crack along its length, while the secondary endpoints are the endpoints extending from the main cracks and away from them. The main and secondary cracks in the network of cracks can be directly identified using existing technologies, such as identifying the number of pixels in each crack. Tiny cracks with a very small number of pixels are not considered, which is an existing technology and will not be elaborated upon here.

[0147] S332, starting from a main endpoint, connect it to an adjacent but unconnected secondary endpoint on one side of the main gap, take the secondary endpoint as the current main endpoint, repeat the above steps until the remaining secondary endpoints on the same side are connected, then establish a connection with another main endpoint to generate a connection polygon.

[0148] Understandably, by constructing a connecting polygon, the scattered crack endpoints can be connected into a closed region, providing a framework for the subsequent uniform distribution of grouting holes, thereby ensuring that the grouting holes cover all key nodes of the main and secondary cracks, while maintaining the uniformity of the hole location distribution.

[0149] Specifically, select a primary endpoint as the starting point; then, find an adjacent, unconnected secondary endpoint on one side of the primary gap and connect the starting point to that secondary endpoint; next, use that secondary endpoint as the new starting point and continue connecting the next adjacent, unconnected secondary endpoint; repeat this process until all secondary endpoints on one side of the primary gap are connected, then connect the last secondary endpoint to another primary endpoint to form a closed connection polygon, see [link to relevant documentation]. Figure 4 Connect the two endpoints sequentially to form a connected polygon.

[0150] S333, construct new grouting holes for each mesh crack at the center point of each side length in the connected polygon.

[0151] Specifically, arranging grouting holes at the midpoint of each edge of the connecting polygon ensures that the holes are evenly distributed around the entire crack network. This allows the grout to diffuse evenly from multiple directions, covering the entire crack area.

[0152] S334, the newly added grouting holes are enlarged based on the grouting threshold to obtain grouting zones. The newly added grouting holes are sorted in descending order based on the number of crack pixels in each grouting zone to obtain the grouting sequence of each network crack.

[0153] Among them, the grouting threshold refers to the parameter used to determine the influence range of the grouting hole, which can be set manually based on the actual situation.

[0154] It should be noted that each grouting hole has a certain influence range. By setting a grouting threshold, the grouting zone of each hole can be determined. Counting the number of crack pixels within each grouting zone can reflect the severity of the cracks in that area. Sort the grouting holes in descending order of the number of crack pixels to prioritize the treatment of densely cracked areas, thereby improving repair efficiency and effectiveness.

[0155] Specifically, the number of crack pixels contained in each grouting zone is counted. Then, the grouting holes are sorted in descending order of the number of crack pixels. Finally, this sorting is used as the grouting sequence for the network cracks.

[0156] Based on the above embodiments, A1-A2 are also included:

[0157] A1. When it is determined that there is an obstruction at the location of the newly added grouting hole, the corresponding newly added grouting hole is taken as the hole to be moved. The outline point of the obstruction is moved based on the hole position radius of the hole to be moved to obtain the offset outline. The obstruction includes the main reinforcement and bolt holes.

[0158] It should be noted that in actual tunnel structures, the pre-set location of the grouting hole may encounter obstacles such as reinforcing bars or bolt holes. Forcing drilling at the original location could damage the structural reinforcement or prevent effective drilling, affecting construction safety and grouting results. Traditional methods often rely on manual on-site adjustment of the hole position, which is inefficient and lacks standardized procedures. Our system automatically detects obstacles and adjusts their location to ensure the grouting hole avoids structural components while remaining within a reasonable construction range, improving construction feasibility and automation.

[0159] Among them, "obstacle" refers to structural components or holes existing at the preset position of the grouting hole, including main reinforcement (the main steel bars in the tunnel lining) and bolt holes (reserved holes for installing equipment); "hole to be moved" refers to newly added grouting holes whose positions need to be adjusted; "hole radius" refers to the actual radius of the grouting hole; "contour point moving processing" refers to the process of moving the boundary point of the obstacle a certain distance (usually the hole radius) along the normal direction; and "offset contour" refers to the new contour of the obstacle obtained after the moving processing.

[0160] First, the system adds the location of the grouting hole and the bolt hole in the image. Since the shield tunnel segments are all cast uniformly, the position of the reinforcing bars is fixed and can be directly updated to the image of each shield tunnel segment. When the system detects that the distance between the center of the grouting hole and the obstacle is less than the radius of the hole, it determines that the hole is to be moved. Next, the system extracts the contour points of the obstacle and moves all contour points outward based on the radius of the hole. This can be done by connecting all contour points with the center point of the contour and then moving the hole outward by the radius of the hole along the direction of the connection, so that the hole position determined at the moved position will not intersect with the obstacle.

[0161] A2, take the outline of the connecting polygon or the selected rectangle as the wrapping outline, take the intersection of the offset outline and the wrapping outline as the offset position, and move the center point of the hole to be moved to the offset position.

[0162] Understandably, you can select the appropriate wrapping contour based on the type of crack (using a connecting polygon for mesh cracks and a bounding rectangle for line cracks). The offset contour will intersect with the wrapping contour. Using the intersection as the offset position will not affect the obstruction while ensuring that the crack is covered. Finally, move the center point of the hole to be moved to the offset position to complete the automatic adjustment of the hole position. If there are multiple holes, you can choose any one to move.

[0163] S4. Based on the acquisition location, update the fusion reliability parameters, newly added grouting holes, and grouting sequence to the tunnel twin model to generate a tunnel display model.

[0164] It is easy to understand that constructing a tunnel twin model corresponding to the tunnel will update the previously collected cracks to the tunnel twin model, update the parameters of the crack in the fused confidence parameters to the corresponding crack location, and mark the newly added grouting holes around the crack and the grouting sequence of the grouting holes at the crack location.

[0165] Based on the above embodiments, it also includes:

[0166] The shield tunnel segments corresponding to each crack in the tunnel display model are obtained as damaged segments. Based on the maintenance starting point in the tunnel display model, the damaged segments within the set range are counted sequentially to obtain the maintenance set sequence.

[0167] Understandably, the process involves retrieving the tunnel display model, identifying the shield tunnel segment to which each crack belongs, and marking these cracked segments as damaged segments. Next, the maintenance starting point in the tunnel display model (such as the tunnel entrance) is determined. Then, according to a set range (such as within 50 meters), damaged segments within 50 meters are counted and sorted sequentially from the maintenance starting point and categorized into maintenance sets (e.g., damaged segments starting from 0-50 meters form maintenance set 1, and damaged segments from 50-100 meters form maintenance set 2). Finally, all maintenance sets are arranged according to the regional division order to form maintenance set sequence 1-2.

[0168] Based on the number of crack pixels in the damaged segments, the damaged segments in each segment maintenance set in the maintenance set sequence are sorted in descending order to obtain the segment maintenance sequence.

[0169] It is understandable that the damaged segments in each segment maintenance set in the maintenance set sequence are sorted in descending order based on the number of crack pixels in the damaged segments to obtain the segment maintenance sequence.

[0170] That is, there is a sequential order of processing within a set, and each set also has a sequential order.

[0171] Furthermore, during maintenance, for line cracks and mesh cracks with similar pixel count deviations, mesh cracks will be treated first. Therefore, this solution is a continuous process. The grouting sequence of the segments will be determined, and then the treatment sequence of each crack in the segment will be determined. Cracks with a large number of pixels will be treated first, and when the number is similar, mesh cracks will be treated first. Each grouting hole at the crack also has its own grouting sequence, giving construction personnel a clear construction order. This allows them to prioritize the treatment of areas with greater potential risks and proceed sequentially according to the maintenance path, ensuring that high-risk segments are treated first.

[0172] See Figure 5 This is a schematic diagram of a tunnel structure health processing system based on multimodal data fusion provided in an embodiment of the present invention. The tunnel structure health processing system based on multimodal data fusion includes:

[0173] The control module is used to control the collaborative acquisition device to acquire data on the tunnel structure, and obtain point cloud data and thermal imaging parameters of the acquisition location.

[0174] The input module is used to input the point cloud data into the crack recognition model for recognition, obtain the crack type and crack parameters of the acquired image, and filter the crack parameters and thermal imaging parameters based on the environmental parameters of the acquisition location to obtain fused reliable parameters.

[0175] The analysis module is used to classify and customize the cracks in the acquired image based on the crack type, and obtain the new grouting holes and grouting sequence for each crack.

[0176] The update module is used to update the fused reliable parameters, newly added grouting holes and grouting sequence to the tunnel twin model according to the acquisition location, and generate a tunnel display model.

[0177] The present invention also provides a readable storage medium storing a computer program, which, when executed by a processor, is used to implement the methods provided in the various embodiments described above.

[0178] The readable storage medium can be a computer storage medium or a communication medium. A communication medium includes any medium that facilitates the transfer of computer programs from one location to another. A computer storage medium can be any available medium accessible to a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application-Specific Integrated Circuit (ASIC). Alternatively, the ASIC can be located in a user equipment. Of course, the processor and the readable storage medium can also exist as discrete components in a communication device. The readable storage medium can be a read-only memory (ROM), random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0179] The present invention also provides a program product including executable instructions stored in a readable storage medium. At least one processor of the device can read the executable instructions from the readable storage medium, and the at least one processor executes the executable instructions to cause the device to implement the methods provided in the various embodiments described above.

[0180] In the embodiments of the above-described device, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0181] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for tunnel structure health processing based on multimodal data fusion, characterized in that, include: The control and collaborative acquisition device acquires data on the tunnel structure, obtaining point cloud data and thermal imaging parameters of the acquisition location. The point cloud data is input into the crack identification model for identification, and the crack type and crack parameters of the acquired image are obtained. The crack parameters and thermal imaging parameters are filtered based on the environmental parameters of the acquisition location to obtain fused reliable parameters. Based on the crack type, the cracks in the acquired images are classified and customized to obtain newly added grouting holes and grouting sequences for each crack, including: Cracks in the acquired images are classified based on crack type, resulting in linear cracks and mesh cracks. By performing a box-selection hole location analysis on each of the aforementioned linear cracks, the newly added grouting holes and grouting sequence for each linear crack are obtained, including: The acquired image is processed into coordinates, and the extreme values ​​of the crack coordinates of each line are calculated based on the set movement value to obtain the selected extreme values; Based on the selected extreme values, a box selection rectangle corresponding to each line crack is constructed, and a new grouting hole for each line crack is constructed at the midpoint of each rectangle side in the box selection rectangle. The newly added grouting holes corresponding to both ends of the linear crack are taken as core holes, and the remaining newly added grouting holes are taken as boundary holes. The core holes and boundary holes are sorted according to the crack locations of each line crack to obtain the grouting sequence of each line crack. The crack locations include the top wall location and the side wall location, including: By placing the boundary hole in front of the core hole, the initial sequence of cracks in each line is obtained; When the crack location is determined to be the top wall location, the corresponding linear crack is taken as the top wall crack, and the initial sequence is taken as the grouting sequence of the top wall crack; When the crack location is determined to be a sidewall location, the corresponding line crack is taken as a sidewall crack. If the crack direction of the sidewall crack is transverse, the boundary hole located at the bottom in the initial sequence is adjusted to the first position to obtain the grouting sequence of the transverse sidewall crack. If the crack direction of the sidewall crack is longitudinal, the core hole located at the bottom in the initial sequence is adjusted to the front of the core hole located at the top to obtain the grouting sequence of the longitudinal sidewall crack. The connection points of the network cracks were analyzed to obtain the newly added grouting holes for each network crack, as well as the grouting sequence, including: Obtain the two main endpoints of the main cracks and the secondary endpoints of the secondary cracks in the network crack; Starting from a main endpoint, connect it to an adjacent but unconnected secondary endpoint on one side of the main gap. Use this secondary endpoint as the current main endpoint and repeat the above steps until all other secondary endpoints on the same side are connected. Then, establish a connection with another main endpoint to generate a connection polygon. New grouting holes are constructed at the center points of each side of the connected polygon to create each mesh-like crack; The newly added grouting holes are enlarged based on the grouting threshold to obtain grouting zones. The newly added grouting holes are sorted in descending order based on the number of crack pixels in each grouting zone to obtain the grouting sequence of each network crack. Based on the data collection location, the fusion reliability parameters, newly added grouting holes, and grouting sequence are updated to the tunnel twin model to generate a tunnel display model.

2. The method according to claim 1, characterized in that, The crack parameters and thermal imaging parameters are filtered based on environmental parameters from the acquisition location to obtain fused reliable parameters, including: The environmental sensor is controlled to acquire environmental parameters at the acquisition location, including dust concentration and water mist density; The environmental interference degree is calculated based on the dust concentration and water mist density, and the laser weight and infrared weight are determined based on the environmental interference degree. Crack parameters are screened based on laser weights to obtain reliable crack parameters, and thermal imaging parameters are screened based on infrared weights to obtain reliable infrared parameters. The crack confidence parameter and the infrared confidence parameter are weighted and fused to obtain the fused confidence parameter.

3. The method according to claim 2, characterized in that, The environmental interference level is calculated based on dust concentration and water mist density. The laser weight and infrared weight are then determined based on this environmental interference level, including: The environmental disturbance level can be obtained using the following formula. in, For environmental interference, Dust concentration, The density of the water mist; The laser weight and infrared weight are determined based on the degree of environmental interference. The laser weight and infrared weight are obtained using the following formulas. in, For laser weights, For infrared weights.

4. The method according to claim 2, characterized in that, The weighted fusion of the crack confidence parameter and the infrared confidence parameter to obtain the fused confidence parameter includes: The fusion confidence parameter is obtained using the following formula. in, To integrate trusted parameters, For the crack reliability parameters, These are the infrared reliable parameters.

5. The method according to claim 1, characterized in that, Also includes: When it is determined that there is an obstruction at the location of the newly added grouting hole, the corresponding newly added grouting hole is taken as the hole to be moved. The outline points of the obstruction are moved based on the hole position radius of the hole to be moved to obtain the offset outline. The obstruction includes the main reinforcement and bolt holes. Use the outline of the connecting polygon or the selected rectangle as the wrapping outline, use the intersection of the offset outline and the wrapping outline as the offset position, and move the center point of the hole to be moved to the offset position.

6. A tunnel structure health processing system based on multimodal data fusion, characterized in that, include: The control module is used to control the collaborative acquisition device to acquire data on the tunnel structure, and obtain point cloud data and thermal imaging parameters of the acquisition location. The input module is used to input the point cloud data into the crack recognition model for recognition, obtain the crack type and crack parameters of the acquired image, and filter the crack parameters and thermal imaging parameters based on the environmental parameters of the acquisition location to obtain fused reliable parameters. The analysis module is used to classify and customize the cracks in the acquired images based on the crack type, and to obtain the newly added grouting holes and grouting sequences for each crack, including: Cracks in the acquired images are classified based on crack type, resulting in linear cracks and mesh cracks. By performing a box-selection hole location analysis on each of the aforementioned linear cracks, the newly added grouting holes and grouting sequence for each linear crack are obtained, including: The acquired image is processed into coordinates, and the extreme values ​​of the crack coordinates of each line are calculated based on the set movement value to obtain the selected extreme values; Based on the selected extreme values, a box selection rectangle corresponding to each line crack is constructed, and a new grouting hole for each line crack is constructed at the midpoint of each rectangle side in the box selection rectangle. The newly added grouting holes corresponding to both ends of the linear crack are taken as core holes, and the remaining newly added grouting holes are taken as boundary holes. The core holes and boundary holes are sorted according to the crack locations of each line crack to obtain the grouting sequence of each line crack. The crack locations include the top wall location and the side wall location, including: By placing the boundary hole in front of the core hole, the initial sequence of cracks in each line is obtained; When the crack location is determined to be the top wall location, the corresponding linear crack is taken as the top wall crack, and the initial sequence is taken as the grouting sequence of the top wall crack; When the crack location is determined to be a sidewall location, the corresponding line crack is taken as a sidewall crack. If the crack direction of the sidewall crack is transverse, the boundary hole located at the bottom in the initial sequence is adjusted to the first position to obtain the grouting sequence of the transverse sidewall crack. If the crack direction of the sidewall crack is longitudinal, the core hole located at the bottom in the initial sequence is adjusted to the front of the core hole located at the top to obtain the grouting sequence of the longitudinal sidewall crack. The connection points of the network cracks were analyzed to obtain the newly added grouting holes for each network crack, as well as the grouting sequence, including: Obtain the two main endpoints of the main cracks and the secondary endpoints of the secondary cracks in the network crack; Starting from a main endpoint, connect it to an adjacent but unconnected secondary endpoint on one side of the main gap. Use this secondary endpoint as the current main endpoint and repeat the above steps until all other secondary endpoints on the same side are connected. Then, establish a connection with another main endpoint to generate a connection polygon. New grouting holes are constructed at the center points of each side of the connected polygon to create each mesh-like crack; The newly added grouting holes are enlarged based on the grouting threshold to obtain grouting zones. The newly added grouting holes are sorted in descending order based on the number of crack pixels in each grouting zone to obtain the grouting sequence of each network crack. The update module is used to update the fused reliable parameters, newly added grouting holes and grouting sequence to the tunnel twin model according to the acquisition location, and generate a tunnel display model.

Images