Tunnel lining construction quality management and control method and system

Abstract

This invention relates to the field of construction quality management, and provides a method and system for quality control of tunnel lining construction. The method includes acquiring first information and second information. The first information includes geological survey parameters of the target tunnel section, BIM model of lining design, concrete material parameters, and construction equipment parameters. The second information includes on-site condition information of the target tunnel section after demolding. Based on the first information, discrete element-finite element coupled simulation and multi-parameter spatial traversal analysis are performed to obtain a sensitivity map of process parameters and a defect risk early warning database for lining construction. The sensitivity map of process parameters, the defect risk early warning database, and the second information are fused to obtain a multi-physics field fusion feature tensor. Based on the multi-physics field fusion feature tensor, source tracing processing is performed to obtain a defect cause diagnosis report. Based on the defect cause diagnosis report, the construction parameters of the next target section are adjusted. This invention achieves proactive early warning of tunnel lining construction quality.

Description

Technical Field

[0001] This invention relates to the field of construction quality management, and more specifically, to a method and system for controlling the quality of tunnel lining construction. Background Technology

[0002] In the field of tunnel engineering construction, the quality of lining construction is directly related to the structural safety and long-term durability of the project. Traditional quality control methods mainly rely on manual experience and discrete point sampling inspection, such as using total stations for cross-sectional measurement or ground-penetrating radar for local non-destructive testing. In recent years, with the application of 3D laser scanning and BIM technology, although it is possible to quickly digitally collect the appearance of the lining and compare it with the design model, which has improved the inspection efficiency and coverage to a certain extent, existing technical solutions are mostly limited to the level of geometric dimension conformity verification. They lack in-depth integrated analysis and accurate source tracing capabilities for internal defects and appearance defects caused by the coupling effect of material characteristics, process parameters and environmental factors during construction. Moreover, the inspection results are disconnected from the process control links, making it difficult to form a closed-loop management of inspection-diagnosis-control. This leads to delayed discovery of quality problems, unclear attribution of causes, and process optimization relying on trial and error. It is impossible to achieve data-driven forward-looking adaptive adjustment of construction parameters and proactive prevention and control of quality risks. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for quality control in tunnel lining construction, so as to improve the above-mentioned problems.

[0004] To achieve the above objectives, the embodiments of this application provide the following technical solutions:

[0005] On the one hand, embodiments of this application provide a method for quality control in tunnel lining construction, the method comprising:

[0006] Obtain first information and second information. The first information includes geological survey parameters of the target section of the tunnel, BIM model of lining design, concrete material parameters and construction equipment parameters. The second information includes on-site condition information of the target section of the tunnel after demolding.

[0007] Based on the first information, discrete element-finite element coupled simulation and multi-parameter spatial traversal analysis are performed to obtain the sensitivity spectrum of process parameters and the defect risk early warning database for lining construction.

[0008] The defect risk warning database and the second information are fused to obtain a multiphysics fusion feature tensor.

[0009] Based on the multiphysics field fusion feature tensor and the process parameter sensitivity map, a defect cause diagnosis report is obtained by performing source tracing processing.

[0010] The construction parameters for the next target section are adjusted based on the defect cause diagnosis report.

[0011] Secondly, embodiments of this application provide a tunnel lining construction quality control system, the system comprising:

[0012] The acquisition module is used to acquire first information and second information. The first information includes geological survey parameters of the target section of the tunnel, BIM model of lining design, concrete material parameters and construction equipment parameters. The second information includes on-site condition information of the target section of the tunnel after demolding.

[0013] The first processing module is used to perform discrete element-finite element coupled simulation and multi-parameter spatial traversal analysis based on the first information to obtain the sensitivity spectrum of process parameters and defect risk warning database for lining construction.

[0014] The second processing module is used to fuse the defect risk warning database and the second information to obtain a multi-physics field fusion feature tensor.

[0015] The third processing module is used to perform source tracing processing based on the multi-physics field fusion feature tensor and the process parameter sensitivity map to obtain a defect cause diagnosis report.

[0016] The fourth processing module is used to adjust the construction parameters of the next target section based on the defect cause diagnosis report.

[0017] Thirdly, embodiments of this application provide a tunnel lining construction quality control device, the device including a memory and a processor. The memory is used to store a computer program; the processor is used to execute the computer program to implement the steps of the above-described tunnel lining construction quality control method.

[0018] Fourthly, embodiments of this application provide a readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described tunnel lining construction quality control method.

[0019] The beneficial effects of this invention are as follows:

[0020] This invention constructs a comprehensive data foundation integrating primary information such as geological surveys, design models, material properties, and equipment parameters with secondary information representing the on-site conditions. Based on discrete element-finite element coupled simulation and multi-parameter spatial ergonomic analysis, it pre-generates sensitivity maps of process parameters and a defect risk early warning database. Furthermore, it deeply integrates prior knowledge with real-time acquired multimodal data to form a multi-physics fusion feature tensor that integrates geometric, mechanical, spectral, and temperature properties. Finally, through intelligent tracing and multi-objective optimization, it achieves accurate diagnosis of defect causes and dynamic adjustment of construction parameters. This effectively overcomes the problems of disconnect between detection and process, one-sided cause analysis, and lagging regulation in traditional methods, realizing closed-loop control of tunnel lining construction quality from passive detection to proactive early warning, and from experience-based judgment to data-driven decision-making.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the tunnel lining construction quality control method described in this embodiment of the invention.

[0024] Figure 2 This is a schematic diagram of the tunnel lining construction quality control equipment described in an embodiment of the present invention.

[0025] The diagram is labeled as follows: 800, Tunnel lining construction quality control equipment; 801, Processor; 802, Memory; 803, Multimedia component; 804, I / O interface; 805, Communication component. Detailed Implementation

[0026] 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] Example 1:

[0029] This embodiment provides a method for quality control in tunnel lining construction. It can be understood that this embodiment can be used to lay out a scenario, such as the need for quality control throughout the entire construction process in the secondary lining construction of highway tunnels.

[0030] See Figure 1 The figure shows that the method includes steps S1-S5.

[0031] Step S1: Obtain first information and second information. The first information includes geological survey parameters of the target section of the tunnel, BIM model of lining design, concrete material parameters and construction equipment parameters. The second information includes on-site condition information of the target section of the tunnel after demolding.

[0032] It should be noted that the first piece of information is mainly obtained by integrating existing engineering design data and construction preparation data. Specifically, this includes extracting surrounding rock parameters from geological survey reports, reading lining design geometry and reinforcement information from BIM models, determining material performance parameters from concrete mix design documents, and specifying equipment models and performance based on construction organization design. The second piece of information is captured in real time through a sensor network and mobile acquisition terminals deployed on site. Environmental sensors are used to collect data on temperature, humidity, and dust concentration inside the tunnel after demolding. The appearance and spatial orientation of the lining are simultaneously acquired through integrated laser scanning and multispectral imaging, thus forming a comprehensive information set covering both static design attributes and dynamic environmental conditions.

[0033] Step S2: Based on the first information, perform discrete element-finite element coupled simulation and multi-parameter spatial ergodic analysis to obtain the sensitivity spectrum of process parameters and defect risk warning database for lining construction.

[0034] This step uses numerical simulation to preview the possible quality outcomes under various working conditions before actual construction, shifting quality control from post-event remediation to pre-event prediction, significantly improving the initiative and foresight of management.

[0035] Step S2 further includes steps S21-S24, which specifically include:

[0036] Step S21: Model based on the first information to obtain the initial model for coupled simulation;

[0037] In this step, firstly, based on the surrounding rock grade, elastic modulus, Poisson's ratio, and initial in-situ stress field data from the geological survey report, these are defined as the boundary conditions and material properties of the finite element model using a parametric mapping method. Next, the lining design BIM model is meshed to generate a structural mesh for finite element analysis. Further, based on the aggregate gradation curve in the concrete mix design, this step uses a random distribution algorithm to generate discrete element particle models representing coarse aggregate, mortar, and interfaces within the lining space. Finally, by defining the contact and constraint relationships between the discrete element model and the finite element boundary, a coupled simulation initial model integrating the surrounding rock and lining is formed.

[0038] Step S22: Perform dynamic coupled simulation of the construction process based on the initial coupled simulation model to obtain the concrete density field and stress field dataset;

[0039] This step aims to numerically recreate the key construction processes of lining pouring and solidification, capturing changes in its internal state. The implementation employs a sequential coupling strategy, including: First, within the discrete element method (DEM) framework, the concrete pouring speed and the frequency and amplitude of the vibrator are set according to the construction equipment parameters. By solving the particle motion equations based on Hertzian contact theory, the rearrangement and compaction process of aggregates and mortar under the action of vibration energy is simulated, and the density distribution at different times is output in real time. Then, this density field is transferred as a material property to the finite element model, while simultaneously coupling it with the concrete hydration heat model to solve the thermo-mechanical coupling equations and calculate the stress field caused by temperature changes. The two models exchange data at each computation time step—updating the real-time density calculated by the DEM to the material properties of the finite element model, and using the stress field calculated by the finite element model as the boundary constraint for particle motion in the DEM, thereby dynamically simulating the entire process from fluid to solid state. Compared to static analysis that only focuses on the final form, the dynamic coupling characteristics of this method can record the evolutionary history of how voids form and when stress concentrates, providing time-dimensional data for understanding the causes of defects, which is something that traditional single simulation or static detection cannot achieve.

[0040] Step S23: Perform multi-parameter spatial traversal and quality index extraction based on the concrete density field and stress field dataset to obtain a parameter combination-quality index mapping relationship database;

[0041] The purpose of this step is to establish a quantitative relationship network between process inputs and quality outputs. Using experimental design methods, key process parameters are used as variables, and Latin hypercube sampling is performed within their reasonable value ranges to generate hundreds of parameter combinations with good spatial distribution representativeness. Dynamic coupling simulation of the construction process is performed on each parameter combination, and quantifiable quality indicators are extracted from the simulation results. Finally, all parameter combinations and their corresponding quality indicator results are structured and stored, forming a large mapping database. This step replaces the expensive and time-consuming trial and error in reality with systematic numerical experiments, actively exploring the quality response laws within the entire parameter space, providing a solid data foundation for subsequent analysis. It should be noted that key process parameters include, but are not limited to, pouring speed, vibration frequency, and ambient temperature. Quantifiable quality indicators include, but are not limited to, the average density and coefficient of variation of the entire lining section (reflecting uniformity), the maximum tensile stress value (associated with cracking risk), and the thickness deviation in specific areas.

[0042] Step S24: Perform global sensitivity analysis and generate a graph based on the parameter combination-quality index mapping relationship database to obtain a process parameter sensitivity graph.

[0043] The purpose of this step is to extract instructive patterns from massive mapping relationships, intuitively revealing the weight and manner in which each process parameter affects quality. The specific implementation process is as follows: Using the variance-based Sobol global sensitivity analysis method, the mapping relationship database is mathematically processed to calculate the first-order sensitivity index (measuring its independent contribution to changes in quality indicators) and the total-order sensitivity index (measuring its own contribution and the total contribution of its interaction with other parameters to changes in quality indicators) for each key process parameter. Then, using surrogate model techniques such as Kriging spatial interpolation, these sensitivity indices calculated at discrete parameter points are fitted into a continuous response surface within the entire continuous parameter space. Finally, this response surface is visualized as a two-dimensional or three-dimensional heatmap, which is the process parameter sensitivity map. This step presents the complex interactions between parameters in an intuitive graphical way. For example, it reveals that at low ambient temperatures, the impact of pouring speed on compaction is significantly enhanced, allowing construction managers to easily identify high-risk parameter areas that require key control.

[0044] Following step S24, steps S25-S28 are further included, which specifically include:

[0045] Step S25: Perform defect induction condition simulation based on the initial coupled simulation model and concrete material parameters to obtain a sample set, which includes various typical defect morphologies and causes.

[0046] This step, based on the initial coupled simulation model, sets a series of non-ideal or extreme combinations of process parameters. These include setting the vibration frequency far below the standard requirements to simulate insufficient vibration, setting the pouring speed extremely high to simulate the risk of cold joints, and adjusting the formwork stiffness to simulate support deformation. Subsequently, a dynamic coupled simulation of the construction process is run. In the simulation results, image processing algorithms identify and mark defect states such as low-density areas and high-stress concentration areas, and precisely correlate each defect with the specific parameter settings that caused it—the cause. Finally, the set of all defect morphology-process parameter pairings constitutes the required sample set. This step efficiently and cost-effectively generates a large number of defect samples with clear causal labels through numerical experiments, overcoming the challenges of scarce defect samples and difficulty in tracing their causes in reality.

[0047] Step S26: Based on the sample set, concrete density field and stress field dataset, perform defect pattern feature extraction and quantization to obtain defect feature vector;

[0048] This step extracts multi-dimensional features from the corresponding concrete density and stress field data for each defect instance in the sample set. These features include calculating the geometric, statistical, morphological, and mechanical characteristics of the defect area. After standardization, these features are concatenated to obtain the defect feature vector. Specifically, the geometric features of the defect area include perimeter and area; the statistical features include the average, standard deviation, and skewness of the density within the area; the morphological features include roundness and compactness; and the mechanical features include the stress gradient and maximum principal stress value of the area surrounding the defect.

[0049] Step S27: Based on the defect feature vector, perform defect pattern-process parameter association rule mining to obtain a rule base, which is used to reveal the association between process conditions and defect types;

[0050] This step utilizes association rule learning algorithms in data mining to uncover hidden, strongly correlated patterns in the data. The algorithm scans all samples, identifying frequently occurring item sets. It generates rules in the form of "IF {low vibration frequency, fast pouring speed} THEN {top voids appear}", and calculates the support (frequency of occurrence) and confidence (probability of a defect occurring under this condition) for each rule. By setting preset minimum support and confidence thresholds, strongly correlated rules are filtered out to form a rule base. It should be noted that an item refers to a specific range of characteristic values, such as an average compaction density < 0.8, or a process parameter value, such as a vibration frequency < 8000 rpm.

[0051] Step S28: Construct a database based on the rule base and the parameter combination-quality index mapping relationship database to obtain a defect risk warning database.

[0052] This step uses a parameter combination-quality indicator mapping database as its foundation. This database contains detailed quality responses under various parameter combinations. Then, association rules mined from a rule base are integrated into this foundational database as a logical judgment layer. Specifically, for each parameter combination in the database, all possible defect warning rules that it might trigger are matched, and corresponding confidence levels are assigned. Ultimately, a structured, queryable database is constructed. When a set of process parameters is input, this database can not only query the predicted quality indicators but also immediately trigger relevant warning rules, outputting the possible defect types, locations, and probabilities of occurrence. This step upgrades a simple numerical query database into an intelligent tool capable of logical reasoning and risk warning, providing core decision support for proactive quality risk prevention and control during construction.

[0053] Step S3: Fuse the defect risk warning database and the second information to obtain the multiphysics fusion feature tensor;

[0054] Step S3 further includes steps S31-S34, which specifically include:

[0055] Step S31: Based on the second information and the defect risk warning database, perform adaptive scanning path planning and sensor parameter optimization to obtain an optimized scanning instruction set that adapts to the on-site environment and risk distribution;

[0056] This step analyzes the environmental data in the second set of information. If the dust concentration is high, the laser scanner's emission power is automatically increased and the integration time is extended to enhance the echo signal strength. If the humidity is too high, the multispectral camera is instructed to prioritize imaging in specific bands sensitive to moisture. Simultaneously, based on the defect risk warning database, areas predicted to be prone to defects in historical simulations under these geological and design conditions (e.g., the arch is a high-risk area for cavities) are queried. Based on this, a non-uniform scanning path is generated: in high-risk areas, the deployment of scanning stations is automatically increased, and the scanner's travel speed is reduced to ensure higher point cloud density and resolution in the data acquisition; while in low-risk areas, a sparser path is used to improve overall efficiency. Finally, all these adjustments to the equipment and path are integrated and encapsulated into a complete set of optimized scanning instructions that can be executed by the automated platform. This step enables the data acquisition process to be environmentally adaptive and risk-oriented, fundamentally solving the challenge of ensuring data quality and prioritizing the capture of critical issues under harsh and changing tunnel conditions.

[0057] Step S32: Perform multimodal data synchronous acquisition according to the optimized scanning instruction set to obtain the original multimodal data stream;

[0058] In this step, the mobile acquisition platform receives and executes an optimized scanning command set. Its core lies in a hardware synchronization triggering mechanism, which sends a unified clock signal to all sensors, including the laser scanner and multispectral camera, ensuring that all data acquisition actions start and stop synchronously within microsecond-level errors. During platform movement and scanning, the platform's built-in positioning and attitude determination system records the platform's precise 3D coordinates and attitude in real time for each frame of data acquisition. Ultimately, the raw data acquired by all sensors, along with their corresponding high-precision spatiotemporal stamps, constitute a raw multimodal data stream that is strictly synchronized in time and continuously correlated in space. It should be noted that the mobile acquisition platform is an automated mobile data acquisition device integrating multiple sensors, a positioning and attitude determination system, and a control unit; this step does not impose specific limitations on the mobile acquisition platform.

[0059] Step S33: Perform pixel-level fusion of the three-dimensional point cloud based on the original multimodal data stream to obtain a three-dimensional point cloud model;

[0060] In this step, the original 3D laser scanning point cloud data in the original multimodal data stream undergoes point cloud preprocessing and multi-station registration. Noise points are removed using a statistical filtering algorithm, and coarse registration based on FPFH features combined with fine registration using an iterative nearest-point algorithm unifies the multi-station scanning data to the same coordinate system, resulting in a high-precision 3D geometric model of the lining surface. Based on the laser ultrasonic vibration response signal in the original multimodal data stream, ultrasonic signal processing and mechanical property inversion processing are performed. Spectral features are obtained by performing a Fast Fourier Transform (FFT) on the time-domain waveform, and inversion calculations are performed using a genetic algorithm or least squares optimization based on the forward model of the material constitutive model to obtain the equivalent elastic modulus and other mechanical property values ​​corresponding to each ultrasonic measurement point. A precise spatial mapping process is performed between the ultrasonic measurement point set with mechanical properties and a high-precision 3D geometric model of the lining surface. Using a non-rigid iterative nearest-point algorithm, the spatial mapping relationship between the sparse ultrasonic measurement point cloud and the dense surface geometric point cloud is constructed and optimized, achieving high-precision alignment and resulting in a spatially precisely registered, sparse geometric-mechanical corresponding point set. Based on this precisely registered, sparse geometric-mechanical corresponding point set, attribute interpolation and model generation are performed. Using the Kriging spatial interpolation method, optimal unbiased estimation of the properties of all unknown vertices on the geometric model is performed based on the known mechanical properties of the measurement points, resulting in a complete geometric-mechanical fused 3D point cloud model, where each vertex contains 3D coordinates and interpolated mechanical properties. This step creates a digital model that combines morphological and mechanical properties.

[0061] Step S34: Perform multi-channel feature extraction and tensor construction based on the three-dimensional point cloud model and the original multimodal data stream to obtain a multi-physics fusion feature tensor.

[0062] Step S34 further includes steps S341-S344, which specifically include:

[0063] Step S341: Extract multi-scale texture and spectral features from the multispectral image in the original multimodal data stream to obtain a multi-scale spectral-texture feature map;

[0064] The purpose of this step is to extract deep information from multispectral data that is highly sensitive to the appearance of the lining. Specifically, this involves inputting image sequences from different bands acquired by a multispectral camera into a pre-trained feature pyramid network. The core structure of this network consists of a bottom-up, progressively abstracting backbone network and a top-down path that fuses multi-scale information. Specifically, the backbone network extracts feature maps at different resolutions at different depths (high-level feature maps are rich in semantic information but lose spatial details, while low-level feature maps retain details well but have weaker semantic information). The feature pyramid network fuses high-level semantic information with the fine details of the lower levels through upsampling and lateral connections, thereby outputting feature maps at multiple scales that contain both rich semantics (e.g., suspected seepage areas) and precise localization (e.g., the precise direction of cracks). Finally, the fused feature maps from all scales are concatenated to obtain a multi-scale spectral-texture feature map. This step can simultaneously capture defects of different scales, from millimeter-level cracks to square meter-level color differences, resolving the contradiction that single-scale analysis cannot take into account both global and local features, and providing a quantitative data foundation that far exceeds the visual perception for a comprehensive assessment of the lining's apparent health status.

[0065] Step S342: Project the multi-scale spectral-texture feature map and the three-dimensional point cloud model to obtain the projected three-dimensional point cloud. Each vertex of the projected three-dimensional point cloud is attached with mechanical, spectral and temperature attributes.

[0066] In this step, the internal and external parameters of the multispectral camera are accurately acquired through camera calibration technology. Then, using a mathematical model of perspective projection transformation, each pixel in the multi-scale spectral-texture feature map is back-projected onto its corresponding 3D vertex in the 3D point cloud model based on its image coordinates and camera parameters. Essentially, this process pastes the texture and spectral information from the 2D image onto the corresponding position in the 3D geometric model. Simultaneously, the acquired temperature data is also mapped onto the 3D vertices in a similar manner through its own calibration parameters. Ultimately, the 3D point cloud, which originally only contained geometric coordinates and mechanical properties, now has each vertex additionally appended with a spectral-texture feature vector from the multispectral image and a temperature value from the thermal imager, forming a highly integrated projected 3D point cloud. The unique effect of this method is that it creates a truly multidimensional spatial dataset, allowing simultaneous querying of the geometric shape, mechanical properties, apparent texture, and surface temperature of any point in 3D space.

[0067] Step S343: Perform enhancement processing on the projected 3D point cloud to obtain the enhanced feature descriptor;

[0068] In this step, the projected 3D point cloud is treated as a graph structure, where each vertex is a node. The K-nearest neighbor algorithm is used to match the K spatially nearest vertices of each vertex and construct connecting edges to represent local spatial relationships. Then, a graph neural network is used for message passing and feature aggregation. Through message passing, the graph neural network allows the features of each node to interact and aggregate with the features of its neighboring nodes. After multiple layers of graph convolution, the final new feature descriptor generated for each vertex not only includes its own original attributes (mechanical, spectral, temperature) but also incorporates the attribute information of other vertices in its local neighborhood. For example, a point with a normal temperature will have an enhanced feature descriptor containing contextual information about the high-temperature region if the surrounding points are generally hot. This step introduces local spatial correlations compared to viewing each data point in isolation, enabling features to express abnormal regions rather than just abnormal points, greatly improving the robustness and accuracy of subsequent defect segmentation and classification.

[0069] Step S344: Based on the enhanced feature descriptor of each vertex, construct the structured multidimensional feature tensor to obtain the multiphysics fusion feature tensor.

[0070] In this step, the entire spatial range of the 3D point cloud is regarded as a regular voxel grid. The enhanced feature descriptor of each vertex is used as the feature vector of that spatial location. Then, the spatial coordinates of all vertices are used as the spatial dimension of the tensor, and the feature vectors of all vertices are used as the channel dimension. Through tensor concatenation operations, these feature vectors are organized into a higher-dimensional tensor, and finally the multiphysics fusion feature tensor is obtained.

[0071] Step S4: Perform source tracing processing based on the multiphysics field fusion feature tensor and the process parameter sensitivity map to obtain a defect cause diagnosis report;

[0072] Step S4 further includes steps S41-S44, which specifically include:

[0073] Step S41: Based on the multi-physics field fusion feature tensor, perform multi-source heterogeneous feature alignment and fusion to obtain the lining surface defect feature map;

[0074] The purpose of this step is to transform the generated structured tensor data, i.e., the multiphysics fusion feature tensor, into a graph structure that better reflects the spatial correlation and feature complexity of the defect region. The process involves converting the 3D point cloud represented by the multiphysics fusion feature tensor into a graph structure, where each data point becomes a graph node, and the node attributes are the multidimensional feature vector of that point (geometric, mechanical, spectral, temperature, etc.). Then, based on spatial proximity, connecting edges are established between nodes to construct a graph representing the topological relationships of the lining surface. Subsequently, a graph convolutional network is applied to process this graph. After multiple layers of graph convolution, the features of each node are updated to enhanced features that incorporate information from its local neighborhood. Finally, this graph, composed of nodes (carrying enhanced features) and edges (representing spatial relationships), becomes the defect feature map of the lining surface. This step elevates the defect detection problem from independent point analysis to the level of correlated regional analysis, naturally expressing the spatial distribution and morphological relationships of defects through a graph structure, greatly enhancing the ability to represent complex defect patterns.

[0075] Step S42: Based on the defect feature map of the lining surface and the sensitivity map of the process parameters, perform defect pattern matching to obtain a list of defect pattern matching results;

[0076] In this step, the surface defect feature map of the lining is compared with the pre-stored typical defect pattern maps generated by simulation in the process parameter sensitivity map. A specific implementation method is to use a graph similarity algorithm for comparison, calculate the similarity score between the field map and each typical pattern map, and finally generate a list of defect pattern matching results sorted by similarity score. Each item in the list indicates several predefined patterns that are most similar to the current field defect and their matching confidence.

[0077] Step S43: Perform multi-cause probabilistic reasoning based on the defect pattern matching result list to obtain the probability distribution set of defect causes;

[0078] This step first constructs a Bayesian network representing the causal relationship between process parameters and defect types based on the association rule base in the defect risk warning database. This network is a probabilistic graphical model, where nodes represent different process parameters (such as vibration frequency and pouring speed) and defect types, and directed edges between nodes represent causal relationships. The list of defect pattern matching results is input into the Bayesian network as evidence. Then, through the conditional probability relationships defined in the network, probability propagation is performed to calculate the posterior probability of each possible misconfiguration of process parameters (i.e., the cause) occurring under the condition of observing the current defect evidence. Finally, a set of probability distributions for defect causes is output, in the form of cause 1 (insufficient vibration): probability 65%, cause 2 (formwork deformation): probability 30%, etc.

[0079] Step S44: Generate a defect cause diagnosis report based on the probability distribution set of the defect causes.

[0080] In this step, a natural language generation model is pre-set. This model integrates key information from the probability distribution set of defect causes (such as main causes, probability values, and associated process parameters) with the spatial location information of the defect on the BIM model. It automatically populates a structured report template, generating a defect cause diagnosis report that includes main diagnostic conclusions, cause analysis, confidence assessment, defect location diagram, and targeted treatment suggestions.

[0081] Step S5: Adjust the construction parameters for the next target section based on the defect cause diagnosis report.

[0082] It is understandable that the next target segment is the subsequent segment adjacent to the target segment.

[0083] Step S5 further includes steps S51-S54, which specifically include:

[0084] Step S51: Construct a multi-objective optimization problem based on the defect cause diagnosis report to obtain a multi-objective optimization function model;

[0085] In this step, the defect cause diagnosis report is analyzed to identify the construction process parameters to be optimized, and then a mathematical model containing multiple objective functions is constructed, specifically:

[0086] ;

[0087] In the above formula, Represents the decision variables, and represents the vector of construction process parameters to be optimized; Indicates the quality deviation target, where, and These represent target quality and predicted quality, respectively. This represents the normalized cost target, where and These represent the construction cost and the maximum allowable cost related to construction parameter X, respectively. This represents the risk penalty function; and These represent the weighting coefficients for the quality deviation target and the cost target, respectively. It should be noted that the specific calculation process for predicting quality is as follows:

[0088] ;

[0089] In the above formula, Indicates the quality of the prediction. Indicates the reference parameters The initial quality prediction value is as follows; Indicates the total number of parameters; Indicates the first The first-order sensitivity coefficients of each parameter are directly derived from the sensitivity spectrum; and They represent the first The and the first The baseline values ​​for each process parameter are taken from the design recommendations used in the current construction. and They represent the first The and the first One construction process parameter; The elements of the Hessian matrix represent the second-order interaction sensitivity of the parameters, which can also be obtained from the spectrum. It should be noted that the specific form of the penalty function is:

[0090] ;

[0091] In the above formula, This represents the penalty weight, which is a large positive number to ensure that the optimizer will try its best to avoid high-risk areas. Indicates the total number of defect types; This represents the predicted risk of occurrence of the k-th type of defect (such as voids or cracks). The calculation method is similar to that of the prediction quality, so it will not be elaborated here. Indicates the first The permissible risk threshold for class defects; This means that a penalty will only be incurred when the predicted risk exceeds a threshold, and the amount of the penalty is proportional to the square of the excess.

[0092] Step S52: Determine the set of candidate process parameter adjustment schemes for the next target segment based on the multi-objective optimization function model and the process parameter sensitivity map;

[0093] Sensitivity maps of process parameters are used as a fast surrogate model. Since directly calling time-consuming coupled simulations for optimization search is computationally intensive, the surrogate model utilizes the sensitivity information in the map to approximately but very quickly predict the impact of parameter changes on quality and risk objectives. Then, a multi-objective genetic algorithm is employed for optimization search. This algorithm initializes a random population of process parameter schemes, calculates the fitness of each scheme on each objective based on the multi-objective optimization function model, and iteratively evolves the population through selection, crossover, and mutation operations, ultimately converging to a Pareto optimal solution set that achieves the best trade-off between quality, cost, and risk objectives—that is, the set of candidate process parameter adjustment schemes for the next objective segment.

[0094] Step S53: Adjust the set of candidate process parameters to conduct a risk assessment and obtain an assessment report;

[0095] This step employs Monte Carlo simulation to consider the normal fluctuations in key parameters that inevitably occur during actual construction for each candidate process parameter scheme (such as slight variations in concrete slump, minor differences in vibrator performance, and fluctuations in ambient temperature). Random numbers are generated based on the historical statistical patterns of these fluctuations, and thousands of random perturbations conforming to a probability distribution are applied to the key parameters of the candidate schemes. Then, using a lightweight surrogate model—the process parameter sensitivity map—the quality performance and risk level of the schemes under these perturbations are quickly simulated. Finally, the probability of quality compliance, cost overrun, and defect risk exceeding limits for each scheme under various perturbations are statistically analyzed, generating a quantitative risk assessment report. This step extends optimization from an ideal deterministic environment to the uncertain real world, quantifying the robustness of each scheme. This ensures that the final recommended scheme is not only theoretically optimal but also practically robust and reliable, significantly reducing the risk of optimization failure due to fluctuations in on-site conditions.

[0096] Step S54: Determine the construction parameter adjustment instructions based on the assessment report.

[0097] In this step, the expected quality improvement effects, cost indicators, and probability of quality achievement of each candidate scheme in the evaluation report are comprehensively considered. The TOPSIS method is used for quantitative comparison, and the optimal scheme is selected from the candidates. Its parameter values ​​are then converted into clear, specific, and operable construction parameter adjustment instructions, such as adjusting the vibration frequency from the current 7500 rpm to 8200 rpm, which will be applied to the lining construction of the next target section. It is understood that using the TOPSIS method for quantitative comparison is a well-known technical solution, and therefore will not be elaborated upon here.

[0098] Example 2:

[0099] This embodiment provides a tunnel lining construction quality control system, which includes an acquisition module, a first processing module, a second processing module, a third processing module, and a fourth processing module, specifically including:

[0100] The acquisition module is used to acquire first information and second information. The first information includes geological survey parameters of the target section of the tunnel, BIM model of lining design, concrete material parameters and construction equipment parameters. The second information includes on-site condition information of the target section of the tunnel after demolding.

[0101] The first processing module is used to perform discrete element-finite element coupled simulation and multi-parameter spatial traversal analysis based on the first information to obtain the sensitivity spectrum of process parameters and defect risk warning database for lining construction.

[0102] The second processing module is used to fuse the defect risk warning database and the second information to obtain a multi-physics field fusion feature tensor.

[0103] The third processing module is used to perform source tracing processing based on the multi-physics field fusion feature tensor and the process parameter sensitivity map to obtain a defect cause diagnosis report.

[0104] The fourth processing module is used to adjust the construction parameters of the next target section based on the defect cause diagnosis report.

[0105] In one specific embodiment of this disclosure, the first processing module further includes a first processing unit, a second processing unit, a third processing unit, and a fourth processing unit, specifically including:

[0106] The first processing unit is used to model based on the first information to obtain the initial model of the coupled simulation.

[0107] The second processing unit is used to perform dynamic coupled simulation of the construction process based on the initial coupled simulation model to obtain a dataset of concrete density field and stress field.

[0108] The third processing unit is used to perform multi-parameter spatial traversal and quality index extraction based on the concrete density field and stress field dataset to obtain a database of parameter combination-quality index mapping relationship.

[0109] The fourth processing unit is used to perform global sensitivity analysis and generate a spectrum based on the parameter combination-quality index mapping relationship database to obtain a process parameter sensitivity spectrum.

[0110] In one specific embodiment of this disclosure, the fourth processing unit is followed by a fifth processing unit, a sixth processing unit, a seventh processing unit, and an eighth processing unit, specifically including:

[0111] The fifth processing unit is used to perform defect induction condition simulation based on the initial coupled simulation model and concrete material parameters to obtain a sample set, which includes various typical defect morphologies and causes.

[0112] The sixth processing unit is used to extract and quantify defect pattern features based on the sample set, concrete density field and stress field dataset, to obtain defect feature vectors.

[0113] The seventh processing unit is used to mine defect pattern-process parameter association rules based on the defect feature vector to obtain a rule base, which is used to reveal the association between process conditions and defect types.

[0114] The eighth processing unit is used to construct a database based on the rule base and the parameter combination-quality index mapping relationship database to obtain a defect risk warning database.

[0115] In one specific embodiment of this disclosure, the second processing module further includes a ninth processing unit, a tenth processing unit, an eleventh processing unit, and a twelfth processing unit, specifically including:

[0116] The ninth processing unit is used to perform adaptive scanning path planning and sensor parameter optimization based on the second information and the defect risk warning database to obtain an optimized scanning instruction set adapted to the field environment and risk distribution;

[0117] The tenth processing unit is used to perform multimodal data synchronous acquisition according to the optimized scanning instruction set to obtain the original multimodal data stream;

[0118] The eleventh processing unit is used to perform pixel-level fusion of the three-dimensional point cloud based on the original multimodal data stream to obtain a three-dimensional point cloud model.

[0119] The twelfth processing unit is used to perform multi-channel feature extraction and tensor construction based on the three-dimensional point cloud model and the original multimodal data stream to obtain a multi-physics fusion feature tensor.

[0120] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0121] Example 3:

[0122] Corresponding to the above method embodiments, this embodiment also provides a tunnel lining construction quality control device. The tunnel lining construction quality control device described below and the tunnel lining construction quality control method described above can be referred to in correspondence.

[0123] Figure 2 This is a block diagram illustrating a tunnel lining construction quality control device 800 according to an exemplary embodiment. Figure 2 As shown, the tunnel lining construction quality control equipment 800 may include: a processor 801 and a memory 802. The tunnel lining construction quality control equipment 800 may also include one or more of the following: a multimedia component 803, an I / O interface 804, and a communication component 805.

[0124] The processor 801 controls the overall operation of the tunnel lining construction quality control equipment 800 to complete all or part of the steps in the aforementioned tunnel lining construction quality control method. The memory 802 stores various types of data to support the operation of the tunnel lining construction quality control equipment 800. This data may include, for example, instructions for any application or method operating on the tunnel lining construction quality control equipment 800, as well as application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the tunnel lining construction quality control equipment 800 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, and an NFC module.

[0125] In an exemplary embodiment, the tunnel lining construction quality control device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the tunnel lining construction quality control method described above.

[0126] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the tunnel lining construction quality control method described above. For example, the computer-readable storage medium may be the memory 802 including the program instructions described above, which may be executed by the processor 801 of the tunnel lining construction quality control device 800 to complete the tunnel lining construction quality control method described above.

[0127] Example 4:

[0128] Corresponding to the above method embodiments, this embodiment also provides a readable storage medium. The readable storage medium described below can be referred to in conjunction with the tunnel lining construction quality control method described above.

[0129] A readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the tunnel lining construction quality control method described in the above method embodiments.

[0130] Specifically, the readable storage medium can be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.

[0131] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0132] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for quality control in tunnel lining construction, characterized in that, include: Obtain first information and second information. The first information includes geological survey parameters of the target section of the tunnel, BIM model of lining design, concrete material parameters and construction equipment parameters. The second information includes on-site condition information of the target section of the tunnel after demolding. Based on the first information, discrete element-finite element coupled simulation and multi-parameter spatial ergonomic analysis are performed to obtain the sensitivity spectrum of process parameters and the defect risk early warning database for lining construction. The defect risk warning database and the second information are fused to obtain a multiphysics fusion feature tensor. Based on the multiphysics field fusion feature tensor and the process parameter sensitivity map, a defect cause diagnosis report is obtained by performing source tracing processing. The construction parameters for the next target section are adjusted based on the defect cause diagnosis report. The process of performing discrete element-finite element coupled simulation and multi-parameter spatial traversal analysis based on the first information includes: Based on the first information, a model is created to obtain the initial model for coupled simulation. Based on the initial coupled simulation model, dynamic coupled simulation of the construction process is performed to obtain data sets of concrete density field and stress field. Based on the concrete density field and stress field dataset, multi-parameter spatial traversal and quality index extraction are performed to obtain a database of parameter combination-quality index mapping relationship; Based on the parameter combination-quality index mapping relationship database, a global sensitivity analysis and graph generation are performed to obtain a process parameter sensitivity graph. The process, including global sensitivity analysis and graph generation based on the parameter combination-quality index mapping database, further includes: Based on the initial coupled simulation model and concrete material parameters, a simulation of defect induction conditions is performed to obtain a sample set, which includes various typical defect morphologies and causes. Based on the sample set, concrete density field and stress field dataset, defect pattern features are extracted and quantified to obtain defect feature vectors. Based on the defect feature vector, defect pattern-process parameter association rules are mined to obtain a rule base, which is used to reveal the association between process conditions and defect types. A database is constructed based on the rule base and the parameter combination-quality index mapping relationship database to obtain a defect risk warning database. The fusion of the defect risk warning database and the second information includes: Based on the second information and the defect risk warning database, adaptive scanning path planning and sensor parameter optimization are performed to obtain an optimized scanning instruction set that adapts to the field environment and risk distribution; Multimodal data is synchronously acquired according to the optimized scanning instruction set to obtain the original multimodal data stream; Pixel-level fusion of the three-dimensional point cloud is performed based on the original multimodal data stream to obtain a three-dimensional point cloud model; Multi-channel feature extraction and tensor construction are performed based on the three-dimensional point cloud model and the original multimodal data stream to obtain a multi-physics fusion feature tensor.

2. The method for quality control of tunnel lining construction according to claim 1, characterized in that, Based on the multiphysics fusion feature tensor and the process parameter sensitivity map, a defect cause diagnosis report is obtained, including: Based on the multi-physics field fusion feature tensor, multi-source heterogeneous feature alignment and fusion are performed to obtain the lining surface defect feature map; Based on the defect feature map of the lining surface and the sensitivity map of the process parameters, defect pattern matching is performed to obtain a list of defect pattern matching results. Based on the defect pattern matching result list, multi-cause probabilistic reasoning is performed to obtain a set of probability distributions of defect causes; A defect cause diagnosis report is generated based on the probability distribution set of the defect causes.

3. The method for quality control of tunnel lining construction according to claim 1, characterized in that, Based on the defect cause diagnosis report, the construction parameters for the next target section are adjusted, including: Based on the defect cause diagnosis report, a multi-objective optimization problem is constructed to obtain a multi-objective optimization function model; The set of candidate process parameter adjustment schemes for the next target segment is determined based on the multi-objective optimization function model and the process parameter sensitivity map. A risk assessment is conducted based on the set of candidate process parameter adjustment schemes, and an assessment report is obtained. The construction parameter adjustment instructions are determined based on the assessment report.

4. A tunnel lining construction quality control system, characterized in that, include: The acquisition module is used to acquire first information and second information. The first information includes geological survey parameters of the target section of the tunnel, BIM model of lining design, concrete material parameters and construction equipment parameters. The second information includes on-site condition information of the target section of the tunnel after demolding. The first processing module is used to perform discrete element-finite element coupled simulation and multi-parameter spatial traversal analysis based on the first information to obtain the sensitivity spectrum of process parameters and defect risk warning database for lining construction. The second processing module is used to fuse the defect risk warning database and the second information to obtain a multi-physics field fusion feature tensor. The third processing module is used to perform source tracing processing based on the multi-physics field fusion feature tensor and the process parameter sensitivity map to obtain a defect cause diagnosis report. The fourth processing module is used to adjust the construction parameters of the next target section based on the defect cause diagnosis report; The first processing module includes: The first processing unit is used to model based on the first information to obtain the initial model of the coupled simulation. The second processing unit is used to perform dynamic coupled simulation of the construction process based on the initial coupled simulation model to obtain a dataset of concrete density field and stress field. The third processing unit is used to perform multi-parameter spatial traversal and quality index extraction based on the concrete density field and stress field dataset to obtain a database of parameter combination-quality index mapping relationship. The fourth processing unit is used to perform global sensitivity analysis and generate a spectrum based on the parameter combination-quality index mapping relationship database to obtain a process parameter sensitivity spectrum. The fourth processing unit further includes: The fifth processing unit is used to perform defect induction condition simulation based on the initial coupled simulation model and concrete material parameters to obtain a sample set, which includes various typical defect morphologies and causes. The sixth processing unit is used to extract and quantify defect pattern features based on the sample set, concrete density field and stress field dataset, to obtain defect feature vectors. The seventh processing unit is used to mine defect pattern-process parameter association rules based on the defect feature vector to obtain a rule base, which is used to reveal the association between process conditions and defect types. The eighth processing unit is used to construct a database based on the rule base and the parameter combination-quality index mapping relationship database to obtain a defect risk warning database; The second processing module includes: The ninth processing unit is used to perform adaptive scanning path planning and sensor parameter optimization based on the second information and the defect risk warning database to obtain an optimized scanning instruction set adapted to the field environment and risk distribution; The tenth processing unit is used to perform multimodal data synchronous acquisition according to the optimized scanning instruction set to obtain the original multimodal data stream; The eleventh processing unit is used to perform pixel-level fusion of the three-dimensional point cloud based on the original multimodal data stream to obtain a three-dimensional point cloud model. The twelfth processing unit is used to perform multi-channel feature extraction and tensor construction based on the three-dimensional point cloud model and the original multimodal data stream to obtain a multi-physics fusion feature tensor.

Images