Slurry shield mud film maintenance test device and service life prediction method

Abstract

The invention provides a slurry shield mud film maintenance test device and a service life prediction method, and relates to the technical field of slurry film formation, and the method comprises the steps: obtaining an original data set; performing image preprocessing according to the surface image sequence to obtain preprocessed image data; performing fracture shape evolution feature extraction according to the preprocessed image data to obtain a fracture shape evolution feature set; performing prediction model construction processing according to the fracture form evolution feature set, the mud matching parameter, the stratum physical parameter and the film forming pressure value, and constructing a mud film life prediction model; and life calculation is carried out according to the mud film life prediction model to obtain a mud film residual life prediction value of the mud film from the current state to the failure moment. In combination with automatic monitoring of a mud film observation mechanism and intelligent analysis based on image sequence and physical field parameter fusion, a mud film life prediction model capable of quantitatively characterizing a crack evolution rule and predicting failure time is constructed.

Description

Technical Field

[0001] This invention relates to the field of mud film formation technology, and more specifically, to a mud-water shield tunneling mud film maintenance test device and life prediction method. Background Technology

[0002] In shield tunnel construction, before operations such as cutterhead replacement using pressurized chamber opening technology, a dense mud film needs to be formed at the tunnel face, supplemented by high-pressure gas within the chamber. This ensures effective support for the tunnel face, improves the safety of cutterhead replacement operations, and effectively controls tunnel face stability. The core reliability of this process lies in the long-term airtight stability of the mud film under high pressure, high temperature, and water-rich geological conditions. However, under such complex physical field coupling, the mud film is prone to micro-cracks that can propagate, leading to airtightness failure and threatening tunnel face stability and operational safety. Therefore, accurately evaluating the long-term crack resistance and gas retention performance of the mud film under different mud formulations and geological conditions through laboratory tests before construction is crucial for optimizing the formulation and ensuring construction safety and economy. Currently, the commonly used method in this field is to use a single permeation column device to form the mud film, followed by pressurization or observation tests within the same device. This method can only simulate a single point of operation, and the test environment (pressure, temperature) is limited. The humidity and other factors are difficult to control precisely and maintain constantly, making it impossible to effectively reproduce the real scenario of long-term coupling effects of multiple factors in the formation. In addition, the observation of the mud film cracking process mostly relies on manual visual observation or simple video recording, lacking quantitative and automated monitoring methods for the initiation and expansion of microcracks. Furthermore, it is impossible to intelligently correlate the observed crack morphology evolution with mud material parameters and formation conditions for lifetime prediction, resulting in low experimental efficiency, high costs, and discrete conclusions that are difficult to systematically guide engineering practice. There is an urgent need for an integrated solution that can simulate the real formation maintenance environment, realize efficient parallel testing under multiple working conditions, and automatically perform quantitative analysis and intelligent lifetime prediction of the mud film cracking process.

[0003] Based on the shortcomings of the existing technology, there is an urgent need for a mud-water shield tunneling mud film maintenance test device and a life prediction method. Summary of the Invention

[0004] The purpose of this invention is to provide a slurry shield tunneling mud film maintenance test device and a life prediction method to improve the above-mentioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0005] In a first aspect, this application also provides a lifespan prediction method, including:

[0006] Obtain the raw data set, which includes the surface image sequence of the mud film during the curing in the simulated formation environment, mud slurry ratio parameters, formation physical parameters, and film formation pressure values;

[0007] Image preprocessing is performed based on the surface image sequence. By partitioning and enhancing the contrast of a single frame image to suppress specular highlights and performing adaptive segmentation based on image grayscale statistics, preprocessed image data with non-uniform illumination interference is obtained.

[0008] Based on the preprocessed image data, crack morphological evolution features are extracted. By detecting the crack edge contour and performing morphological connection and background noise filtering on the broken pixel segments, a crack morphological evolution feature set representing the development of cracks from point-like initiation to line-like penetration is obtained.

[0009] Based on the fracture morphology evolution feature set, the mud slurry ratio parameters, the formation physical parameters and the film-forming pressure value, a prediction model is constructed. By fusing the spatiotemporal evolution features of fractures with physical field parameters and using a Bayesian optimization strategy to fit the nonlinear mapping relationship, a mud film lifetime prediction model is constructed.

[0010] Based on the mud film life prediction model and the acquired real-time maintenance data, life calculation is performed, and the predicted value of the remaining life of the mud film from the current state to the time of failure is obtained through model deduction.

[0011] Secondly, this application provides a test device for mud film curing in slurry shield tunneling machines, comprising:

[0012] A test chamber assembly, comprising a test chamber body and a test chamber cover, wherein the test chamber cover is sealed to the test chamber body, and a permeation column placement ring is provided inside the test chamber body;

[0013] A detachable permeation column assembly is disposed within the test chamber assembly. The bottom of the detachable permeation column assembly is fixedly connected to the permeation column placement ring. The detachable permeation column assembly includes an upper permeation column, a detachable permeation column, and a lower permeation column. The detachable permeation column is detachably and sealingly connected between the upper permeation column and the lower permeation column.

[0014] An environmental control mechanism is provided, comprising a test environment control unit, a pressure holding component, a humidity holding component, and a temperature holding component. The test environment control unit is fixedly installed on the outside of the test chamber and is point-connected to the pressure holding component, the humidity holding component, and the temperature holding component, respectively.

[0015] A mud film observation mechanism, comprising an image acquisition component and an air pressure monitoring component, wherein the image acquisition component is disposed within the test chamber assembly, and the air pressure monitoring component is connected to the interior of the test chamber assembly; Both the test chamber assembly and the detachable permeation column assembly are made of transparent material.

[0016] Furthermore, the pressure maintaining assembly includes an air compressor, a pressurization pipe, an electronic barometer, an electrically controlled vent valve, and a pressure regulating valve. The air compressor is connected to the test chamber assembly via the pressurization pipe. The electronic barometer is fixedly mounted on the test chamber body. The pressure regulating valve is mounted on the pressurization pipe. The electrically controlled vent valve is fixedly mounted on the test chamber cover. The humidity maintaining assembly includes an electronic hygrometer, an electrically controlled humidifier, and a water tank. The electronic hygrometer is mounted on the test chamber body, and the electrically controlled humidifier and the water tank are located inside the test chamber body. The temperature maintaining assembly includes an electronic thermometer, a temperature control assembly, an electrically controlled air heater, and a ventilation fan. The electronic thermometer and the temperature control assembly are fixedly mounted on the test chamber body, and the electrically controlled air heater and the ventilation fan are located inside the test chamber body.

[0017] Furthermore, the pressure monitoring component includes a barometer box and at least one barometer, the barometer being connected to the detachable permeation column assembly.

[0018] Furthermore, the mud film observation mechanism also includes a mud film vent outlet and a mud film vent outlet pipe. The mud film vent outlet is located at the bottom of the detachable permeable column placement ring, and the mud film vent outlet pipe connects the mud film vent outlet to the barometer.

[0019] Furthermore, the barometer is a graduated cylindrical liquid barometer.

[0020] The beneficial effects of this invention are as follows:

[0021] This invention combines an independently replaceable and detachable permeation column assembly with a precise environmental control mechanism to achieve long-term, parallel, and environmentally constant curing tests on multiple mud film samples within a single test chamber. Simultaneously, by integrating automatic monitoring from a mud film observation mechanism with intelligent analysis based on image sequences and physical field parameters, a mud film life prediction model is constructed that can quantitatively characterize crack evolution and predict failure time. This systematically improves the efficiency and depth of mud film performance research, providing a reliable basis for optimizing mud slurry ratios and construction parameters, and ensuring the safety and economy of shield tunneling under pressure. 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 1This is a schematic diagram of the mud-water shield tunnel mud film curing test device described in the embodiment of the present invention; Figure 2 This is a schematic diagram of the detachable permeation column assembly; Figure 3 for Figure 1 AA section view in the middle; Figure 4 for Figure 3 BB section view in the middle; Figure 5 This is a flowchart of the lifetime prediction method described in an embodiment of the present invention.

[0024] The diagram shows: 1. Test chamber assembly; 11. Test chamber body; 12. Test chamber cover; 13. Permeation column placement ring; 2. Removable permeation column assembly; 21. Upper permeation column; 22. Removable permeation column; 23. Lower permeation column; 3. Environmental control mechanism; 31. Overall test environment control; 32. Pressure maintenance assembly; 321. Air compressor; 322. Pressurization pipe; 323. Electronic barometer; 324. Electrically controlled vent valve; 325. Pressure regulating valve; 3 3. Humidity maintenance component; 331. Electronic hygrometer; 332. Electrically controlled humidifier; 333. Water tank; 34. Temperature maintenance component; 341. Electronic thermometer; 342. Temperature control assembly; 343. Electrically controlled air heater; 344. Ventilation fan; 4. Mud film observation mechanism; 41. Image acquisition component; 42. Air pressure monitoring component; 421. Barometer box; 422. Barometer; 423. Mud film vent outlet; 424. Mud film vent outlet pipe. Detailed Implementation

[0025] 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.

[0026] 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.

[0027] In actual shield tunnel engineering, the strata ahead of the tunnel face are a complex system full of uncertainties. Under pressure, the drilling mud seeps into the strata pores to form a mud film, which serves as a key barrier to isolate the high-pressure gas inside the chamber from the external soil and water pressure. Its long-term stability directly determines the safe window period for pressurized entry operations. However, the underground environment is not static; it is a dynamic space with strong coupling effects from multiple physical fields, such as stress field, seepage field, chemical field, and temperature field. On the one hand, the heterogeneity of the strata itself (such as abrupt changes in permeability coefficient and the existence of interlayers) will cause the drilling mud to produce an uneven filter cake structure during the film-forming stage, forming inherent weak points. On the other hand, during the opening operation, which lasts for hours or even days, the continuous action of high-pressure gas inside the chamber, the seepage erosion of surrounding groundwater, the micro-disturbances caused by the release of the original stress of the strata, and the property evolution of the drilling mud material itself during hydration and consolidation processes are not isolated factors, but rather mutually excite and superimpose each other. For example, localized high-permeability pathways can exacerbate water migration and accelerate the loss of fine particles from the mud film, thus inducing the initiation of microcracks under continuous pressurized gas. The appearance of microcracks then becomes a new dominant seepage channel, further intensifying erosion and forming a positive feedback loop of "seepage-damage," ultimately leading to penetrating rupture of the mud film far before reaching its expected service life, resulting in a sudden loss of airtightness. Such failures are often sudden and insidious. Traditional, single, static laboratory tests cannot reproduce this multi-field, long-term, coupled evolutionary process, and it is even more difficult to quantitatively predict the "safe lifespan" of a specific mud formulation in a specific formation to maintain effective airtightness before construction. This makes on-site decision-making largely dependent on experience, containing significant risks. Therefore, there is an urgent need for an experimental and evaluation method that can highly simulate the above-mentioned complex coupling effects and can monitor and intelligently predict the mud film performance degradation process in real time.

[0028] Example 1: like Figure 5 As shown, this embodiment provides a lifetime prediction method, including steps S100 to S500.

[0029] Step S100: Obtain the raw data set, which includes the surface image sequence of the mud film during the curing in the simulated formation environment, mud slurry ratio parameters, formation physical parameters and film formation pressure values;

[0030] Understandably, the acquired raw data set specifically includes: real-time images of the mud film surface acquired by an image acquisition component deployed within the test chamber assembly, and a sequence of mud film surface images obtained by frame decomposition of the video using a cross-platform computer vision library (OpenCV). This sequence records the evolution of the mud film's apparent morphology under environmental influences; mud slurry ratio parameters determined through laboratory testing, mainly including specific gravity reflecting solid content, Marvel funnel viscosity characterizing flow and slag-carrying capacity, swelling ratio determining viscosity, and sand content affecting the structural framework; geological physical parameters obtained through geological exploration, mainly the permeability coefficient determining the mud slurry infiltration rate and the porosity affecting the mud film adhesion foundation and drainage path; and film-forming pressure values ​​simulating actual soil chamber support pressure recorded by the air pressure monitoring component 42. These data provide interrelated initial information inputs for subsequent analysis from four dimensions: visual performance, material formulation, geological environment, and mechanical conditions.

[0031] Step S200: Perform image preprocessing based on the surface image sequence. By partitioning and enhancing the contrast of a single frame image to suppress specular highlights and performing adaptive segmentation based on image grayscale statistics, preprocessed image data with non-uniform illumination interference is obtained.

[0032] It should be noted that during long-term maintenance experiments, due to the reflection and refraction of a constant light source on the transparent plexiglass enclosure, the acquired mud film images generally exhibit a non-uniform illumination gradient with excessive brightness in the center and dimness at the edges. Simultaneously, the wet mud film surface produces localized specular highlights. These interferences severely obscure the appearance of microcracks, easily masking the initially formed hairline microcracks. Therefore, this step involves partitioned contrast enhancement of single-frame images to adaptively suppress overexposure in highlight areas while brightening details in shadow areas, thereby balancing the illumination effect across the entire image. Subsequently, adaptive segmentation is performed based on the image's own grayscale statistical characteristics to initially separate darker pixel areas that may belong to cracks from the mud film background, ultimately resulting in a pre-processed image free from non-uniform illumination interference.

[0033] Step S300: Extract crack morphological evolution features based on preprocessed image data. By detecting crack edge contours and performing morphological connection and background noise filtering on broken pixel segments, a crack morphological evolution feature set representing the development of cracks from point-like initiation to line-like penetration is obtained.

[0034] Understandably, the typical evolution pattern of mud film cracks originates from scattered microscopic point-like defects, gradually extending and connecting into visible linear cracks, and may eventually develop into continuous fissures. This step first locates these discontinuous features by detecting the crack edge contours. Since early cracks may appear discontinuous due to particle obscuring or uneven lighting, morphological connection operations are needed to bridge these broken pixel segments to restore the true continuous morphology of the cracks. Simultaneously, the inherent particle inhomogeneity of the mud itself generates a large number of discrete noise points in the image, similar to micro-cracks, which must be filtered out by analyzing the geometric features of connected regions. Finally, a series of dynamic evolution data characterizing the crack length, area, and connectivity over time are extracted.

[0035] Step S400: Based on the fracture morphology evolution feature set, mud ratio parameters, formation physical parameters and film formation pressure value, a prediction model is constructed. By integrating the spatiotemporal evolution features of fractures with physical field parameters and using a Bayesian optimization strategy to fit the nonlinear mapping relationship, a mud film lifetime prediction model is constructed.

[0036] It should be noted that the lifespan of the mud film is not solely determined by visible fractures, but is influenced by a combination of factors including mud material properties, formation geological conditions, and the pressure exerted. These factors exhibit strong nonlinear coupling. This step involves multi-source fusion of fracture spatiotemporal evolution features with mud mix parameters, formation physical parameters, and film-forming pressure values ​​to construct a comprehensive feature space. To learn these multi-dimensional, nonlinear correlations from limited experimental data and avoid overfitting, a Bayesian optimization strategy is preferred to automatically search for the optimal hyperparameter configuration. This process aims to fit the intrinsic relationship between fracture propagation rate and multi-physics conditions, thereby constructing a mud film lifespan prediction model with generalization capabilities.

[0037] Step S500: Based on the mud film life prediction model and the acquired real-time maintenance data, calculate the lifespan and obtain the predicted value of the remaining lifespan of the mud film from the current state to the time of failure through model deduction.

[0038] After model training, this method requires real-time lifetime assessment of new mud film samples under actual maintenance conditions. This step first extracts and vectorizes newly acquired real-time maintenance data, including the latest surface images, mud parameters, formation parameters, and pressure values, in accordance with historical data, forming a real-time feature vector recognizable by the model. Subsequently, this vector is input into the pre-constructed mud film lifetime prediction model for forward extrapolation. The model outputs a sequence of crack development trend indicators based on current conditions. Finally, by comparing and extrapolating this predicted trend with predefined failure criteria, such as calculating the time required for crack indicators to reach the critical penetration threshold, the remaining lifetime prediction value of the mud film from its current state to the point of functional failure is obtained, achieving a leap from condition monitoring to lifetime early warning.

[0039] Further, step S200 includes steps S210 to S230.

[0040] Step S210: Perform specular reflection suppression processing on wet surface according to the surface image sequence. By dividing a single frame image into multiple sub-regions and performing independent adaptive histogram equalization, the gray value of the highlight area is suppressed while limiting the contrast stretching degree in each sub-region, and a single frame image with enhanced contrast is obtained.

[0041] Step S220: Extract the crack foreground based on the contrast-enhanced single-frame image. Calculate the inter-class variance of the image to determine the optimal threshold for segmenting the grayscale image into foreground and background, and obtain a preliminary binary image of the crack foreground.

[0042] Step S230: Based on the preliminary binary image of the crack foreground, perform background texture noise filtering. By analyzing the geometric features of the connected regions, distinguish between discrete noise formed by mud particles and continuous linear crack structures to obtain preprocessed image data that eliminates non-uniform illumination interference.

[0043] Specifically, step S210 addresses the specular highlights and non-uniform lighting issues arising from the wetted surface of the mud film by employing a Limiting Contrast Adaptive Histogram Equalization (CLAHE) algorithm. This algorithm divides a single-frame image into multiple sub-regions and performs independent adaptive histogram equalization. While strictly limiting the contrast stretching within each sub-region, it effectively suppresses the grayscale values ​​of the highlight areas. This process strongly suppresses specular reflection while accurately preserving the granular skeletal texture that characterizes the mud film's structure, preventing subsequent steps from misinterpreting the natural roughness of the mud as noise. This results in a single-frame image with significantly improved contrast and preserved details. Building upon this, step S220 focuses on separating the crack target. Based on the contrast-enhanced single-frame image, crack foreground extraction is performed. The Otsu's method (Maximum Between-Class Variance) is used to calculate the between-class variance of the image to determine the optimal threshold for segmenting the grayscale image into foreground and background. This method, based on the principle of maximizing between-class variance, automatically adapts to different image grayscale distributions, enabling preliminary and accurate extraction of crack regions from complex backgrounds, resulting in a preliminary binary image of the crack foreground. However, the binary image not only contains real cracks, but also contains a large number of discrete noise points formed by mud particles. Therefore, step S230 performs background texture noise filtering based on the preliminary crack foreground binary image. By analyzing the geometric features such as the length, area and aspect ratio of each connected region, it is possible to effectively distinguish between real cracks with continuous, slender linear structures and approximately circular, small-area particle noise points, and finally obtain preprocessed image data that eliminates non-uniform illumination interference.

[0044] Further, step S300 includes steps S310 to S330.

[0045] Step S310: Perform crack edge contour detection processing based on preprocessed image data. By calculating the image gradient, identify areas with abrupt changes in pixel grayscale values ​​and locate potential crack edges to obtain an edge intensity map containing edge position and direction information.

[0046] Step S320: Perform fracture crack segment connection processing based on the edge intensity map. Apply morphological dilation operation to bridge the edge pixel discontinuities caused by uneven lighting or mud film particle occlusion, connect discrete edge points into a continuous linear structure, and obtain a preliminary connected crack skeleton map.

[0047] Step S330: Based on the preliminary connected crack skeleton map, background particle noise is filtered out and feature quantization is performed. By analyzing the length, area and aspect ratio of the connected region, the elongated connected domain formed by real cracks and the approximately circular small area noise formed by mud particles are distinguished, and a crack morphology evolution feature set representing the development of cracks from point-like initiation to line-like penetration is obtained.

[0048] Specifically, step S310 employs the Canny edge detection algorithm, which identifies abrupt changes in pixel grayscale values ​​by calculating image gradients, thereby accurately locating potential crack edges and obtaining an edge intensity map that clearly outlines the contours and contains positional and directional information. This algorithm, due to its excellent signal-to-noise ratio and positioning accuracy, effectively conforms to the evolutionary pattern of cracks developing from points to lines. Step S320 aims to address the edge discontinuity problem caused by uneven illumination or mud film particle obstruction. Based on the edge intensity map, it performs fractured crack segment connection processing, applying morphological dilation operations to bridge discrete edge pixels, reconnecting the originally broken small crack segments into a continuous linear structure, thus obtaining a preliminary connected crack skeleton map. Step S330 quantifies the skeleton map. Based on the preliminary connected crack skeleton map, background particle noise is filtered out and feature quantization is performed. By analyzing the geometric features such as the length, area and aspect ratio of each connected region, the elongated connected domain formed by real cracks and the approximately circular small-area pseudo-noise points generated by the unevenness of mud particles are effectively distinguished. Finally, the length, area and connectivity indicators that can characterize the development of cracks from point-like initiation to line-like penetration are extracted, namely the crack morphology evolution feature set.

[0049] Further, step S400 includes steps S410 to S430.

[0050] Step S410: Perform multi-source heterogeneous feature fusion processing based on fracture morphology evolution feature set, mud ratio parameters, formation physical parameters and film formation pressure value. By associating and integrating the spatiotemporal sequence of fracture length and area with mud specific gravity, viscosity, sand content and formation porosity and permeability coefficient, a comprehensive feature vector for model training is obtained.

[0051] Step S420: Based on the comprehensive feature vector, perform model training and hyperparameter optimization. By inputting data into the gradient boosting framework and using a Gaussian process as a surrogate model to guide the search for the optimal hyperparameter combination of the model, the optimized prediction model basis is obtained.

[0052] Step S430: Based on the optimized prediction model, the prediction function is constructed. By expressing the mapping relationship between fracture propagation rate and multi-physics field conditions as a weighted function of mud parameters, formation parameters and film-forming pressure, the mud film lifetime prediction model is constructed.

[0053] Specifically, step S410 performs multi-source heterogeneous feature fusion processing. Its core involves integrating the spatiotemporal sequences of length and area representing the dynamic expansion of fractures extracted from the image sequence with the physical properties of the mud (including specific gravity, Marshall funnel viscosity, swelling ratio, and sand content) and the physical properties of the formation (porosity, permeability coefficient) and film-forming pressure values. This results in a comprehensive feature vector that integrates visual evolution information and multi-physics conditions, providing a comprehensive data foundation for model learning. Step S420 performs model training and hyperparameter optimization on this comprehensive feature vector. To adapt to the common small-sample data scenarios in experiments and capture complex nonlinear relationships between features, this step uses the gradient boosting algorithm as the basic framework for model training. Simultaneously, to address the critical impact of model configuration on prediction accuracy and avoid overfitting, a Bayesian hyperparameter optimization mechanism is introduced. Specifically, a Gaussian process surrogate model automatically and efficiently searches for the optimal hyperparameter combination of the gradient boosting model in the parameter space, thereby obtaining a foundation for an optimized prediction model with high generalization ability. Step S430 then proceeds to construct the final prediction function based on this optimization, aiming to obtain an interpretable and computable lifetime prediction model. This model expresses the complex mapping relationship between fracture propagation rate and multiphysics conditions as an explicit weighted function of mud parameters, formation parameters, and film-forming pressure. The model can be represented as follows:

[0054] ;

[0055] In the formula, The target mud film is the predicted remaining life of the mud film from its current state to the time of failure. For the first Each eigenvector has components , , and These represent specific gravity, Marvife funnel viscosity, swelling ratio, and sand content, respectively. The index of the feature vector; The total number of eigenvectors; For the first The sensitivity weights of each feature vector are obtained through machine learning training; Formation porosity; Film-forming pressure; The formation permeability coefficient; These are the correction values, which can be determined during the experiment.

[0056] Finally, a mud film lifetime prediction model was constructed. To further ensure the model's robustness under different formation permeability and porosity conditions, this embodiment incorporates a 5-fold cross-validation mechanism for cyclic stress testing. This mechanism divides the observed mud film image sequence and the environmental data collected by sensors into five subsets, which are used in turn as validation sets to verify the prediction accuracy. This effectively prevents the model from overfitting due to excessive focus on texture noise under specific laboratory lighting conditions, ensuring the stability of the prediction results when simulating different geological conditions at the tunnel face.

[0057] Further, step S500 includes steps S510 to S530.

[0058] Step S510: Perform real-time feature extraction and vectorization processing based on real-time maintenance data. By performing preprocessing and crack morphology feature extraction on the real-time acquired mud film surface image, and combining real-time mud ratio, formation parameters and film formation pressure, a real-time feature vector is obtained.

[0059] Step S520: Perform forward model extrapolation based on the real-time feature vector, and input it into the mud film lifetime prediction model to calculate the prediction index sequence characterizing the development trend of mud film cracks under the current multi-physics field conditions.

[0060] Step S530: Map the remaining effective gas retention time according to the predicted index sequence. By comparing and deducing the crack development trend output by the model with the preset failure criteria, the predicted value of the remaining life of the mud film from the current state to the failure time is obtained.

[0061] Step S510 is responsible for converting real-time maintenance data into model-recognizable input. By performing preprocessing and crack morphology feature extraction processes identical to those used in training data on the real-time acquired mud film surface images, the quantitative characteristics of the cracks at the current moment are obtained. This is then combined with the real-time synchronously acquired mud mix ratio, formation parameters, and film-forming pressure to form a real-time feature vector matching the model input dimension. This vector represents the current state of the mud film. Step S520 is the model calculation and extrapolation stage. Based on this real-time feature vector, forward extrapolation processing is performed. By inputting it into the already constructed mud film lifetime prediction model, the model, based on its internally learned parameter-crack evolution complex mapping relationship, calculates the possible development trajectory of indicators such as crack length and area over a future period, thus obtaining a prediction index sequence. This sequence reveals the potential failure trend of the mud film under the current multiphysics field conditions. Step S530 transforms the trend prediction into specific and actionable engineering indicators. Based on the predicted indicator sequence, the remaining effective air-tightness time is mapped. By comparing and extrapolating the crack development trend output by the model (such as crack penetration) with pre-defined failure criteria (such as the critical threshold for airtightness loss) based on engineering experience or standards, the estimated time required for the current state to develop to the critical point is calculated, ultimately yielding the predicted value of the remaining mud film life. This prediction logic directly transforms the evolution patterns of mud film parameters, formation parameters, and crack morphology obtained in the laboratory into numerical decision-making suggestions, providing solid scientific support for dynamically adjusting mud slurry ratios and ensuring the stability of the tunnel face during pressurized cutterhead replacement operations at the shield tunneling site.

[0062] Example 2:

[0063] like Figure 1 and Figure 2As shown, this embodiment provides a mud-water shield tunneling mud film curing test device, which includes a test chamber assembly 1, a detachable permeable column assembly 2, an environmental control mechanism 3, and a mud film observation mechanism 4. The test chamber assembly 1 includes a test chamber body 11 and a test chamber cover 12, with the cover 12 sealed to the body 11. A permeable column placement ring 13 is provided inside the test chamber body 11. The detachable permeable column assembly 2 is disposed inside the test chamber assembly 1, with its bottom fixedly connected to the permeable column placement ring 13. The detachable permeable column assembly 2 includes an upper permeable column 21, a detachable permeable column 22, and a lower permeable column 23, with the detachable permeable column 22 detachably and sealed between the upper permeable column 21 and the lower permeable column 23. The environmental control mechanism 3 includes... The test chamber assembly 1 includes a test environment control unit 31, a pressure holding component 32, a humidity holding component 33, and a temperature holding component 34. The test environment control unit 31 is fixedly installed on the outside of the test chamber 11. The test environment control unit 31 is connected to the pressure holding component 32, the humidity holding component 33, and the temperature holding component 34 at points. The mud film observation mechanism 4 includes an image acquisition component 41 and a pressure monitoring component 42. The image acquisition component 41 is installed inside the test chamber assembly 1, and the pressure monitoring component 42 is connected to the inside of the test chamber assembly 1. Both the test chamber assembly 1 and the detachable permeation column assembly 2 are made of transparent material. The core inventive idea of ​​this embodiment lies in constructing an integrated experimental platform that separates preparation and curing, accurately simulates the environment, and automatically integrates observation. By employing independently detachable permeation column components, the two stages of mud film preparation (film formation) and long-term performance curing (gas retention) are decoupled, enabling standardized prefabrication and parallel testing of mud film samples, significantly improving research efficiency. By integrating and precisely controlling the environmental control mechanism 3 for pressure, humidity, and temperature, the real environment of long-term coupling of multiple physical fields in the formation is systematically simulated, making the experimental conditions closer to engineering practice. Simultaneously, by utilizing transparent materials and combining image acquisition and air pressure monitoring, a non-invasive, multi-dimensional quantitative tracking method is achieved, combining visual observation of crack initiation and propagation on the mud film surface with physical monitoring of airtightness failure. Ultimately, these designs collectively upgrade the traditional isolated and inefficient single-point test into a highly efficient and precise research platform capable of systematically exploring the influence of mud slurry ratio, formation parameters, and curing environment on the long-term performance of mud films. This provides a solid data foundation and technical support for revealing the performance degradation mechanism of mud films, establishing life prediction models, and optimizing construction parameters.

[0064] Preferably, such as Figure 4As shown, the pressure maintaining assembly 32 includes an air compressor 321, a pressurization pipe 322, an electronic barometer 323, an electrically controlled vent valve 324, and a pressure regulating valve 325. The air compressor 321 is connected to the test chamber assembly 1 through the pressurization pipe 322. The electronic barometer 323 is fixedly mounted on the test chamber body 11. The pressure regulating valve 325 is mounted on the pressurization pipe 322. The electrically controlled vent valve 324 is fixedly mounted on the test chamber cover 12. The humidity maintaining assembly 33 includes an electronic hygrometer 331 and an electrically controlled humidifier 332. The test chamber 11 is equipped with a water tank 333 and an electronic hygrometer 331. The electronically controlled humidifier 332 and the water tank 333 are located inside the test chamber 11. The temperature holding assembly 34 includes an electronic thermometer 341, a temperature control assembly 342, an electronically controlled air heater 343, and a ventilation fan 344. The electronic thermometer 341 and the temperature control assembly 342 are fixedly installed on the test chamber 11, while the electronically controlled air heater 343 and the ventilation fan 344 are located inside the test chamber 11. The pressure maintaining component 32 achieves long-term dynamic and stable control of the simulated chamber pressure through a closed loop of air compressor 321 supply, pressure regulating valve 325 coarse adjustment, electronic barometer 323 monitoring, and electrically controlled vent valve 324 fine adjustment, so as to accurately simulate the support pressure of the soil chamber; the humidity maintaining component 33 actively increases humidity through a closed loop of electronic hygrometer 331 monitoring and electrically controlled humidifier 332 execution, simulating the wetting and seepage softening effect of groundwater on the mud film; the temperature maintaining component 34, through the design of electronic thermometer 341 monitoring, temperature control assembly 342 heating, electrically controlled air heater 343 heating, and ventilation fan 344 driving circulation, ensures uniform temperature inside the chamber while providing a heat source, so as to simulate the influence of ground temperature or construction heat. Under the coordination of the overall test environment control 31, these three systems can independently and synchronously control the three key environmental parameters of pressure, humidity and temperature at the set values ​​for a long time. This allows for a high degree of replication of the complex, constant or changing real service conditions that mud films are subjected to underground in the laboratory, providing an accurate and reliable environmental basis for studying the long-term performance degradation mechanism under such multi-field coupling effects.

[0065] Preferably, the pressure monitoring component 42 includes a barometer housing 421 and at least one barometer 422, which is connected to the detachable permeation column assembly 2. This design establishes a physical leak detection channel for each individual mud film sample by directly connecting at least one barometer 422 to the detachable permeation column assembly 2.

[0066] Preferably, such as Figure 3 As shown, the mud film observation mechanism 4 also includes a mud film venting outlet 423 and a mud film venting outlet pipe 424. The mud film venting outlet 423 is located at the bottom of the placement ring of the detachable permeable column 22, and the mud film venting outlet pipe 424 connects the mud film venting outlet 423 with the barometer 422.

[0067] Preferably, the barometer 422 is a graduated cylindrical liquid barometer.

[0068] 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 changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A lifespan prediction method, characterized in that, include: Obtain the raw data set, which includes the surface image sequence of the mud film during the curing in the simulated formation environment, mud slurry ratio parameters, formation physical parameters, and film formation pressure values; Image preprocessing is performed based on the surface image sequence. By partitioning and enhancing the contrast of a single frame image to suppress specular highlights and performing adaptive segmentation based on image grayscale statistics, preprocessed image data with non-uniform illumination interference is obtained. Based on the preprocessed image data, crack morphological evolution features are extracted. By detecting the crack edge contour and performing morphological connection and background noise filtering on the broken pixel segments, a crack morphological evolution feature set representing the development of cracks from point-like initiation to line-like penetration is obtained. Based on the fracture morphology evolution feature set, the mud slurry ratio parameters, the formation physical parameters and the film-forming pressure value, a prediction model is constructed. By fusing the spatiotemporal evolution features of fractures with physical field parameters and using a Bayesian optimization strategy to fit the nonlinear mapping relationship, a mud film lifetime prediction model is constructed. Based on the mud film life prediction model and the acquired real-time maintenance data, life calculation is performed, and the predicted value of the remaining life of the mud film from the current state to the time of failure is obtained through model deduction.

2. The lifetime prediction method according to claim 1, characterized in that, Image preprocessing is performed based on the surface image sequence. This involves partitioning and enhancing the contrast of a single frame image to suppress specular highlights and adaptive segmentation based on image grayscale statistics to obtain preprocessed image data free from non-uniform illumination interference. The data includes: Based on the surface image sequence, wet surface specular reflection suppression processing is performed. By dividing a single frame image into multiple sub-regions and performing independent adaptive histogram equalization, the gray value of the highlight region is suppressed while limiting the contrast stretching degree in each sub-region, resulting in a single frame image with enhanced contrast. Based on the contrast-enhanced single-frame image, crack foreground extraction processing is performed. The inter-class variance of the image is calculated to determine the optimal threshold for segmenting the grayscale image into foreground and background, resulting in a preliminary crack foreground binary image. Background texture noise is filtered out based on the preliminary crack foreground binary image. By analyzing the geometric features of connected regions, discrete noise formed by mud particles and continuous linear crack structures are distinguished, and preprocessed image data with non-uniform illumination interference is obtained.

3. The lifetime prediction method according to claim 1, characterized in that, Based on the preprocessed image data, crack morphological evolution features are extracted. By detecting crack edge contours and performing morphological connection and background noise filtering on broken pixel segments, a crack morphological evolution feature set representing the development of cracks from point-like initiation to linear continuity is obtained, including: Based on the preprocessed image data, crack edge contour detection processing is performed. By calculating the image gradient, regions with abrupt changes in pixel grayscale values ​​are identified and potential crack edges are located, resulting in an edge intensity map containing edge position and orientation information. The fractured crack segments are connected according to the edge intensity map. The discontinuity of edge pixels caused by uneven lighting or mud film particles is bridged by applying morphological dilation operation, and the discrete edge points are connected into a continuous linear structure to obtain a preliminary connected crack skeleton map. Based on the preliminary connected crack skeleton diagram, background particle noise is filtered out and feature quantization is performed. By analyzing the length, area and aspect ratio of the connected region, the elongated connected domain formed by real cracks and the approximately circular small-area noise formed by mud particles are distinguished, and a crack morphology evolution feature set representing the development of cracks from point-like initiation to linear penetration is obtained.

4. The lifetime prediction method according to claim 1, characterized in that, Based on the fracture morphology evolution feature set, the mud mix ratio parameters, the formation physical parameters, and the film-forming pressure value, a prediction model is constructed. By fusing the spatiotemporal evolution characteristics of fractures with physical field parameters and using a Bayesian optimization strategy to fit the nonlinear mapping relationship, a mud film lifetime prediction model is obtained, including: Based on the fracture morphology evolution feature set, the mud ratio parameters, the formation physical parameters and the film formation pressure value, a multi-source heterogeneous feature fusion process is performed. By associating and integrating the spatiotemporal sequence of fracture length and area with the mud specific gravity, viscosity, sand content and formation porosity and permeability coefficient, a comprehensive feature vector for model training is obtained. Based on the comprehensive feature vector, model training and hyperparameter optimization are performed. By inputting data into the gradient boosting framework and using a Gaussian process as a surrogate model to guide the search for the optimal hyperparameter combination of the model, the optimized prediction model basis is obtained. Based on the optimized prediction model, a prediction function is constructed. By expressing the mapping relationship between fracture propagation rate and multiphysics conditions as a weighted function of mud parameters, formation parameters, and film-forming pressure, a mud film lifetime prediction model is obtained.

5. The lifetime prediction method according to claim 1, characterized in that, Based on the mud film life prediction model and the acquired real-time maintenance data, life calculation is performed. The predicted remaining lifespan of the mud film from its current state to its failure time is obtained through model deduction, including: Based on the real-time maintenance data, real-time feature extraction and vectorization are performed. By performing preprocessing and crack morphology feature extraction on the real-time acquired mud film surface image, and combining the real-time mud ratio, formation parameters and film formation pressure, a real-time feature vector is obtained. Based on the real-time feature vector, the model is forward extrapolated and then input into the mud film lifetime prediction model to calculate a sequence of prediction indicators characterizing the development trend of mud film cracks under the current multiphysics field conditions. Based on the predicted index sequence, the remaining effective gas retention time is mapped. By comparing the crack development trend output by the model with the preset failure criteria, the predicted value of the remaining life of the mud film from the current state to the failure time is obtained.

6. A test device for mud film curing in slurry shield tunneling machines, characterized in that, include: Test chamber assembly (1), the test chamber assembly (1) includes a test chamber body (11) and a test chamber cover (12), the test chamber cover (12) is sealed to the test chamber body (11), and a permeation column placement ring (13) is provided inside the test chamber body (11); A detachable permeation column assembly (2) is disposed inside the test chamber assembly (1). The bottom of the detachable permeation column assembly (2) is fixedly connected to the permeation column placement ring (13). The detachable permeation column assembly (2) includes an upper permeation column (21), a detachable permeation column (22), and a lower permeation column (23). The detachable permeation column (22) is detachably and sealed between the upper permeation column (21) and the lower permeation column (23). An environmental control mechanism (3) is provided, comprising a test environment control unit (31), a pressure holding component (32), a humidity holding component (33), and a temperature holding component (34). The test environment control unit (31) is fixedly installed on the outside of the test chamber body (11), and the test environment control unit (31) is connected to the pressure holding component (32), the humidity holding component (33), and the temperature holding component (34) respectively. Mud film observation mechanism (4), the mud film observation mechanism (4) includes an image acquisition component (41) and an air pressure monitoring component (42), the image acquisition component (41) is set inside the test chamber component (1), and the air pressure monitoring component (42) is connected to the inside of the test chamber component (1); The test chamber assembly (1) and the detachable permeation column assembly (2) are both made of transparent material.

7. The slurry shield tunneling mud film curing test device according to claim 6, characterized in that: The pressure maintaining assembly (32) includes an air compressor (321), a pressurizing pipe (322), an electronic barometer (323), an electrically controlled vent valve (324), and a pressure regulating valve (325). The air compressor (321) is connected to the test chamber assembly (1) through the pressurizing pipe (322). The electronic barometer (323) is fixedly mounted on the test chamber body (11). The pressure regulating valve (325) is mounted on the pressurizing pipe (322). The electrically controlled vent valve (324) is fixedly mounted on the test chamber cover (12). The humidity maintaining assembly (33) includes an electronic hygrometer (331), an electrically controlled humidifier (332), and... The water tank (333) and the electronic hygrometer (331) are installed on the test chamber body (11). The electronically controlled humidifier (332) and the water tank (333) are installed inside the test chamber body (11). The temperature holding assembly (34) includes an electronic thermometer (341), a temperature control assembly (342), an electronically controlled air heater (343), and a ventilation fan (344). The electronic thermometer (341) and the temperature control assembly (342) are fixedly installed on the test chamber body (11). The electronically controlled air heater (343) and the ventilation fan (344) are installed inside the test chamber body (11).

8. The slurry shield tunneling mud film curing test device according to claim 6, characterized in that: The pressure monitoring component (42) includes a barometer box (421) and at least one barometer (422), which is connected to the detachable permeation column component (2).

9. The slurry shield tunneling mud film curing test device according to claim 8, characterized in that: The mud film observation mechanism (4) also includes a mud film venting outlet (423) and a mud film venting outlet pipe (424). The mud film venting outlet (423) is located at the bottom of the detachable permeable column (22) placement ring. The mud film venting outlet pipe (424) connects the mud film venting outlet (423) with the barometer (422).

10. The slurry shield tunneling mud film curing test device according to claim 8, characterized in that: The barometer (422) is a graduated cylindrical liquid barometer.

Images