A method for diagnosing and predicting tunnel lining cracks

Abstract

This invention provides a method for diagnosing and predicting tunnel lining cracks, comprising the following steps: S1 establishing a comprehensive evaluation system for the degree of crack damage in a hierarchical structure; S2 establishing a cumulative damage model for tunnel lining cracks based on the comprehensive evaluation system and influencing factor vectors; S3 establishing a parameter iterative estimation model based on the Monte Carlo Markov chain method to obtain parameter estimation results; S4 obtaining a tunnel lining crack disease diagnosis and prediction model based on the damage accumulation model and parameter estimation results, and diagnosing and predicting tunnel lining crack diseases. This invention, on the one hand, can achieve etiological diagnosis of crack diseases by ranking the weights of influencing factors; on the other hand, it can predict the development of damage over time under different variable combinations, calculating the future development and risk of crack diseases.

Description

Technical Field

[0001] This invention belongs to the field of tunnel structural health diagnosis, specifically relating to a method for diagnosing and predicting tunnel lining cracks. Background Technology

[0002] The number and total mileage of tunnels in highways, railways, and subways are increasing year by year. The diversity of tunnel types and the complexity of terrain also lead to many difficulties in tunnel maintenance. Affected by factors such as design, construction, environment, operation, and maintenance, tunnel lining structures inevitably suffer from structural defects. Among these defects, lining cracking is the most common. Cracks in the tunnel lining can cause durability problems in the lining structure, thereby affecting the normal use of the tunnel, causing structural safety hazards, seriously affecting traffic safety, and leading to tunnel closure or even collapse.

[0003] The causes of tunnel cracking are usually not singular; it is a systemic problem resulting from multiple factors, making it a "difficult-to-treat disease" for tunnel lining structures, presenting numerous challenges in diagnosis and prediction. Diagnosis involves evaluating and prioritizing the factors that induce the continued development or deterioration of lining cracks; prediction involves assessing the future extent of damage to the lining cracks based on multiple contributing factors. Most existing diagnostic methods consider only a single factor, failing to comprehensively consider the various key influencing factors affecting tunnel lining defects. This severely limits the diagnostic capabilities and prevents the development of accurate prediction methods for tunnel lining defects tailored to specific field conditions.

[0004] Numerous factors contribute to tunnel lining cracking, exhibiting significant uncertainty and subjectivity. For timely and targeted maintenance and repair, a method is urgently needed to diagnose the relative importance of various influencing factors, effectively estimating the proportion of different factors in the resulting risks. This would allow for the identification of the root cause and pinpointing the problem, rather than relying solely on subjective human judgment. Furthermore, judging structural safety based on a single indicator or factor can lead to incomplete assessments that fail to reflect potential adverse factors, thereby further increasing tunnel safety risks. Therefore, a method for diagnosing and predicting lining cracks that integrates data from multiple sources is required. Summary of the Invention

[0005] The purpose of this invention is to provide a method for diagnosing and predicting tunnel lining cracks, thereby enabling the diagnosis and prediction of tunnel lining cracks.

[0006] The technical solution of the present invention is as follows:

[0007] A method for diagnosing and predicting cracks in tunnel lining includes the following steps:

[0008] S1 establishes a comprehensive evaluation system for the degree of damage to tunnel lining cracks in a hierarchical structure;

[0009] S2 establishes a cumulative model of crack damage in tunnel lining based on a comprehensive evaluation system for crack damage and vector of influencing factors.

[0010] S3 establishes a parameter iterative estimation model based on the Monte Carlo Markov chain method and obtains the parameter estimation results;

[0011] S4 is based on the damage accumulation model and parameter estimation results to obtain a tunnel lining crack diagnosis and prediction model, which is used to diagnose and predict tunnel lining cracks.

[0012] Furthermore, the comprehensive evaluation system includes: an indicator layer, a criterion layer, and a target layer; the indicator layer includes: crack length, crack width, crack depth, and crack number; the criterion layer includes: crack geometry and crack density; and the target layer includes the comprehensive evaluation value of tunnel lining crack damage.

[0013] Furthermore, the method for obtaining the comprehensive evaluation value of crack damage is specifically as follows:

[0014] Based on the analytic hierarchy process (AHP) and variable fuzzy theory, and combined with multi-source data obtained from engineering experience and practical tunnel lining crack detection methods, a hierarchical comprehensive evaluation system for tunnel lining crack damage was established by selecting criterion layers including geometric morphology and crack density, as well as index layers for crack length, crack width, and crack depth. By calculating weights and membership degrees, the comprehensive evaluation value of crack damage in the lining section was calculated.

[0015] Furthermore, the comprehensive evaluation value of the crack damage includes four damage levels: healthy, sub-healthy, diseased, and critically diseased.

[0016] Furthermore, S2 specifically refers to: based on the traditional proportional hazards model and survival analysis method, considering the influence of various potential influencing factors, establishing a mathematical relationship between the comprehensive evaluation value of crack damage and the vector of influencing factors, namely, the damage accumulation model.

[0017] Furthermore, the damage accumulation model is as follows:

[0018]

[0019] In the formula, D is the comprehensive evaluation value of crack damage, α and β are the scale parameter and shape parameter in the Weibull distribution, and X = (x0, x1, x2, ..., x k ) T Let A be a vector of covariates with k+1 variables, and let B be the coefficient vectors of the covariates with k+1 coefficients each.

[0020] Furthermore, S3 specifically refers to:

[0021] Based on the detection data from engineering cases, potential influencing factors affecting tunnel lining cracks were initially screened. Then, the values ​​of various potential influencing factors were statistically analyzed, and univariate factor analysis was performed on all sections of the tunnel to examine the significant relationship between the severity of cracks and each potential influencing factor. On this basis, data normalization preprocessing was performed. Using the comprehensive evaluation value of crack damage as the dependent variable and multiple influencing factors, including time, as independent variables, the parameter distribution of each variable was solved in reverse.

[0022] Technical effects of the present invention:

[0023] The tunnel lining crack diagnosis and prediction method provided by this invention can, on the one hand, diagnose the cause of cracks by ranking the weights of the factors affecting cracks; on the other hand, it can predict the development of damage over time under different combinations of variables and calculate the future development and risk of cracks. Attached Figure Description

[0024] The accompanying drawings illustrate various embodiments generally by way of example rather than limitation, and are used, together with the specification and claims, to explain embodiments of the invention. Where appropriate, the same reference numerals are used in all drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the apparatus or method.

[0025] Figure 1 A schematic diagram of the overall process of the method of the present invention is shown;

[0026] Figure 2 A Bayesian network diagram of the comprehensive evaluation value of lining crack damage according to the present invention is shown. Detailed Implementation

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] This invention provides a method for diagnosing and predicting cracks in tunnel lining, specifically including the following steps:

[0029] (1) First, establish a pyramid-shaped hierarchical structure for the evaluation system of lining crack damage indicators: The first layer is the target layer, which has only one target object, namely the comprehensive evaluation result of lining crack damage, denoted by A; The second layer is set as the criterion layer, which consists of features that directly reflect the development of lining cracks, including two aspects: geometric shape and density, denoted by B1 and B2 respectively, and the two are combined into the criterion layer factor set; The third layer is the indicator layer, which consists of factors related to the characteristics of the criterion layer: Geometric shape includes the length, width and depth of cracks, denoted by C1, C2 and C3; Density only has one indicator, the number of cracks, denoted by C4.

[0030] In this invention, the indicators selected in the comprehensive evaluation system for tunnel lining crack damage should be distinguished from various influencing factors. That is, the comprehensive evaluation value of tunnel lining crack damage is the "result," while the multiple, multi-source factors that cause or induce tunnel lining cracking and continued deterioration are the "causes." The indicators selected for the former should be the most obvious and naturally quantifiable characteristics of the lining cracks; while the latter, as multi-source potential influencing factors related to lining cracks, are expressed using the influencing factor vector in the damage accumulation model, and are analyzed and diagnosed as causes leading to tunnel cracking.

[0031] Table 1 Comprehensive Evaluation System for Tunnel Crack Damage

[0032]

[0033] To ensure that the judgment criteria for the indicator layer can be linked to the test samples and the evaluation values ​​of tunnel lining crack damage status, and to prepare for the application of the fuzzy-analytic hierarchy process, it is necessary to first determine the judgment criteria for each indicator. The judgment criteria refer to the state of the indicator layer when its value falls within a certain range, and this state is further divided into several levels based on the severity of the situation.

[0034] Based on domestic and international standards for judging tunnel lining crack levels, a "multi-parameter" tunnel crack classification method is proposed. Crack length *l*, crack width *b*, and crack depth are used as evaluation indicators. The crack depth is represented by *K*, which is the ratio of the average crack depth to the thickness of the secondary lining structure. The classification standards are shown in Table 2.

[0035] Table 2 Classification based on tunnel crack length, width, and depth

[0036]

[0037] Furthermore, the comprehensive damage value of lining cracks is a value calculated and analyzed in this invention by combining the length, width, depth, and density of lining cracks. Obviously, there is no readily available corresponding judgment standard in current specifications. In order to define and describe the specific state of cracks after quantifying the comprehensive evaluation of cracks in engineering cases, a comprehensive evaluation level classification method for tunnel lining crack damage is derived by analogy with the highway tunnel health status level classification methods in different types of tunnel maintenance standards at home and abroad.

[0038] Table 3. Explanation of Comprehensive Evaluation Level of Tunnel Lining Crack Damage

[0039]

[0040] In the process of comprehensively evaluating the damage of lining cracks, the primary task is to determine the membership degree. Determining the membership degree essentially involves establishing a mapping from the sample detection results of the index layer indicators in the domain of discourse to the standard intervals of each level. This mapping reflects the degree to which the sample detection values ​​of the index layer indicators belong to a certain fuzzy concept, that is, the situation of each damage state level.

[0041] The detection value of the index layer index of a certain section sample in the comprehensive evaluation system for tunnel lining crack damage is known to be x. ij (i represents the criteria at the criterion level, i = 1, 2, 3, ..., m; j represents the indicator at the indicator level belonging to criterion level i, j = 1, 2, 3, ..., n). The comprehensive evaluation value judgment standard range for indicator j at the indicator level, level h (h = 1, 2, 3, ..., k), is [a, b]. jh .

[0042] Because the distance between construction joints in the tunnel lining used in this invention is 10m, the samples used for subsequent evaluation system analysis and model calculation are defined as follows when preprocessing the detection data: 20 meters is considered a segment, each segment contains two construction sections, and the crack damage on the lining of a 20-meter segment is used as a segment sample to statistically calculate the comprehensive crack damage value of the sample. The number of cracks on each segment sample is the index layer index of the crack density of that sample, and the index layer indexes for crack length, width, and depth are all selected from the maximum values ​​of all cracks on that segment sample.

[0043] Based on the interval [a, b] jh The variation range [c,d] of indicator j and level h in the indicator layer is determined by equations (1) and (2). jh and M jh value.

[0044]

[0045] For criterion i in the criterion layer, the relative difference functions (3), (4) and the relative membership formula (5) are applied to calculate the detected value x of the index in the index layer ij For the judgment standard interval [a, b] jh The membership vector ij R”

[0046] When x < M, its relative difference function is:

[0047]

[0048] When x > M, its relative difference function is:

[0049]

[0050] Establish the relative membership vector of the index in the index layer to the evaluation level j R”(j = 1, 2, 3, 4), then there is:

[0051] i R” = ( i1 R” i2 R” i3 R” i4 R”) (6)

[0052] Finally, the first-level fuzzy comprehensive evaluation is completed through the weighted average type fuzzy comprehensive evaluation model, and the calculation in the fuzzy comprehensive evaluation process is through the multiplication of matrices, as shown in formula (7):

[0053] ω'·R” = R' (7)

[0054] In the formula: ω'——The weight vector of the index in the index layer;

[0055] R”——The single-factor evaluation matrix;

[0056] R'——The membership vector of the corresponding criterion in the criterion layer to the comment set V.

[0057] For the geometric features in the criterion layer, the weight vector of the index in the index layer is calculated based on the “1~9” digital scale method according to the previous research results and engineering experience, which will not be elaborated here, and the calculation result is given as:

[0058] ω' = (ω1” ω2” ω3”) = (0.385 0.385 0.230) (8)

[0059] In addition, for the crack density in the criterion layer, since there is only one index, the crack number, under its index layer, the membership degree is directly obtained using Table 4 of the membership degree:

[0060] Table 4 Membership degree of crack number

[0061]

[0062] (4) Take the first-level fuzzy comprehensive evaluation results of all evaluation models. i R', constitutes the single-factor evaluation set for the secondary evaluation { i Let R'} form the evaluation matrix R for the second-level fuzzy comprehensive evaluation. In the second-level fuzzy comprehensive evaluation process, the second-level fuzzy comprehensive evaluation result—the membership vector Z of the criterion level to the evaluation grade—is calculated by matrix multiplication of the evaluation matrix R and the weight vectors of the criteria at the criterion level. Then:

[0063] Z=ω·R=ω·(1R'2R')=(z1 z2 z3 z4) (9)

[0064] In the formula: ω — weight vector of the criteria layer;

[0065] z1 z2 z3 z4 — These correspond to the membership degrees of the four evaluation levels, respectively.

[0066] Since only two factors are set in the criterion layer in this invention, the product scaling method is used to calculate the weight vector of the criterion layer:

[0067]

[0068] After obtaining the membership vector Z, the membership vector single-value processing method is used. The result is the comprehensive evaluation value of crack damage, denoted by D (Damage), and the calculation formula is as follows:

[0069]

[0070] Combining the tunnel lining crack damage evaluation level description table and the crack damage comprehensive evaluation value D, the crack damage comprehensive value status description table is shown in Table 5:

[0071] Table 5. Description of Crack Damage Overall Value

[0072] Damage level Health (B) Sub-health (1A) Disease (2A) Critically ill (3A) Overall value D 25～37.5 37.5～62.5 62.5～87.5 87.5～100

[0073] (2) In the classic Cox proportional hazards model, the analysis of the research object always revolves around the life t of the engineering structure. In order to explore the development of the damage degree of lining cracks over time, this invention incorporates multiple potential influencing factors into the model in the form of a covariate vector. Combined with the comprehensive evaluation value of crack damage obtained above, the Weibull distribution is used as the probability distribution of the basic risk function to establish the relationship between the comprehensive evaluation value of crack damage and multiple potential influencing factors.

[0074] The cumulative distribution function F(D) of the comprehensive crack damage value is obtained by replacing the original survival time t with the comprehensive crack damage value D, as shown in equation (12):

[0075]

[0076] Where: D—Comprehensive evaluation value of crack damage;

[0077] α, β — scale and shape parameters in the Weibull distribution;

[0078] X = (x0, x1, x2, ..., x k ) T —A covariate vector with k+1 variables;

[0079] A and B are the coefficient vectors of the covariates, each with k+1 coefficients.

[0080] The research object of this invention is tunnel lining cracking. The data-driven approach uses discrete, repeated measurement data, and expresses the scale parameter α and shape parameter β using generalized estimation equations of covariate vectors and their coefficient vectors. Due to the characteristics of the Weibull distribution, any appropriate functional form can be chosen for the generalized equations of the shape parameter α and scale parameter β; here, an exponential function is used as an example.

[0081] α=g1(X,A)=exp(a0+a1x1+a2x2+...+a k x k (13)

[0082] β=g2(X,B)=exp(b0+b1x1+b2x2+...+b k x k (14)

[0083] In equation (12), the structural life or survival time t in the traditional Cox proportional hazards model is replaced by the comprehensive evaluation value D of lining crack damage. This transformation is the core of this invention. Because the proportional hazards model uses F(t) to describe the accumulated risk of failure or instability when the object's survival time exceeds time t, while the damage accumulation model uses F(D) to describe the risk of lining crack damage accumulating beyond a certain limit. This is the essential reason why this model is called the damage accumulation model.

[0084] By rearranging terms in equation (12), we can obtain the damage prediction function. Assume that at a certain moment, n tunnel lining crack samples are detected and collected, and the magnitude of the comprehensive damage evaluation value of these samples follows a Weibull distribution. Assume P... i=i% represents the probability that i% of the partial damage composite values ​​in the segment sample are less than or equal to a certain value D. According to the Weibull damage accumulation model, we can derive from equation (12):

[0085]

[0086] Moving the numerical value D as the dependent variable to the left side of the equation, the above equation can be transformed into:

[0087]

[0088] For a two-parameter Weibull distribution, the approximate cumulative failure probability value can be obtained using the median rank formula, where MR represents the percentile of the i-th sample at a confidence level of P = 50%. The empirical formula for calculating the median rank is:

[0089]

[0090] Therefore, the expression for D in the case of discrete samples is:

[0091]

[0092] This invention incorporates a time variable into the influencing factor vector X, because the service life of the tunnel itself is a potential factor affecting the lining structure. By changing the time variable, equation (12) can be used to predict the probability of the future comprehensive damage value exceeding a certain limit, and equation (18) can be used to predict the comprehensive damage evaluation value of a section sample at a future time. It should be noted that the time variable here is not the same concept as the time replaced by the comprehensive damage value mentioned earlier. The former is the service life of the structure, while the latter is the lifespan in the structural reliability analysis.

[0093] (3) This invention focuses on the diagnosis of inducing factors of crack defects, driven by information collected on relevant potential factors. The first part consists of factors reflecting the degree of lining crack damage: crack length, crack width, crack depth, and crack number. Each 20m segment was considered, and the investigated tunnel group case included 180 segments with lining crack defects, which were then labeled. Following the calculation steps, the vector D = {d} of the comprehensive crack damage evaluation value for the 180 samples was obtained. i}, i = (1, 2, ..., 180).

[0094] The other part analyzes the potential factors that cause the cracks to worsen, including: the grade of the surrounding rock, the location of the crack, whether it is a left or right tunnel, whether it is near a construction joint, the water leakage situation, the direction of the crack, and the detection time.

[0095] First, univariate factor analysis is used to preliminarily screen the significance of factors on the disease: the greater the difference in the average comprehensive damage value of the lining under different categories of the same factor, the more significant the factor's impact on the disease; otherwise, the factor's impact on the disease is considered insignificant.

[0096] This standard can be used to quantify significance: if the difference between different factors or values ​​exceeds 5%, then the effect of the factor is significant; otherwise, it is not significant and the factor can be excluded.

[0097] In the case applied in this invention, the three factors of surrounding rock grade, whether it is near the construction joint, and whether it is located in the left or right tunnel have a weak impact on the comprehensive value of crack damage. Therefore, the factors to be included in the analysis of the covariate vector are further selected as: crack direction, crack location, water leakage situation, and time.

[0098] Since the variables have different types and ranges, they need to be normalized. Data normalization preprocessing makes the covariates represented by different factors comparable. Let x1 represent the crack direction, x2 represent the crack location, x3 represent the crack leakage, and t represent time. The S-curve nonlinear function is used to normalize variables x1, x2, and x3; uniform distribution normalization is used to normalize variable t. After normalization, the range of the four variables is unified to [0,1], and the results are shown in Table 6.

[0099] Table 6. Vector Normalization Results of Influencing Factors

[0100]

[0101] It's important to note that x4 here is the normalized time variable, and x4 has no dimensions. The unnormalized time variable is still expressed in t, with the unit being years. The relationship between t and x4 after normalization for a uniform distribution is:

[0102]

[0103] After data preprocessing, it can be incorporated as a covariate into the damage accumulation model:

[0104] α=g1(X,A)=exp(a0+a1x1+a2x2+a3x3+a4x4) (20)

[0105] β=g2(X,B)=exp(b0+b1x1+b2x2+b3x3+b4x4) (21)

[0106] The covariate coefficients to be estimated are a0…a4 and b0…b4. a1 and b1 are the covariate coefficients corresponding to the influencing factor crack direction parameter x1, a2 and b2 for the crack location parameter x2, a3 and b3 for the water leakage parameter x3, and a4 and b4 for the time variable x4. Furthermore, a0 and b0 are constant terms. The introduction of constant terms serves two main purposes: first, there are coupling relationships between different factors. This paper does not focus on analyzing such potential relationships, as these coupling relationships are not within the scope of single-variable factor analysis. Therefore, constant terms are added to the covariates to consider these influences and avoid significant model errors; second, in tunnel maintenance, inspections cannot detect all influencing factors of lining cracks. Adding constant terms can reflect the comprehensive influence of these unknown factors.

[0107] Based on the modeling logic of the damage accumulation model, node types are classified and Bayesian networks are constructed as follows: Figure 2 As shown. Since prior information about the covariate coefficients is unavailable, it is assumed that all parameters follow an uninformative prior distribution, and that a0…a4 and b0…b4 both follow a normal distribution with a mean of 0 and a variance of 10. This invention uses the Markov MCMC method to generate samples of the posterior distribution and employs the No-U-Turn sampling algorithm within the Pyro framework to calculate the posterior distribution of the parameters, achieving high computational efficiency and accurate results.

[0108] The parameter estimation results focus on the mean and Monte Carlo error. Generally, when using the MCMC method, a Monte Carlo error below 5% is considered sufficiently small, indicating iterative convergence. Bayesian estimation calculates a probability distribution, which cannot be directly applied to function operations. Therefore, in subsequent diagnostic and predictive applications using model methods, the mean of the parameter estimates is used directly.

[0109] (3) The first function of this invention is to diagnose the importance of influencing factors of tunnel lining cracks by comparing the covariate coefficients a. i The sign and absolute value of the value can be used to diagnose and rank the factors that induce the continued development of lining cracks or the deepening of damage in a certain section, so as to truly find the cause objectively and quantitatively, rather than relying on subjective experience and human conjecture, as shown in Table 7.

[0110] Table 7. Diagnostic methods for the importance of influencing factors.

[0111]

[0112] If tunnel maintenance is neglected, existing structural defects will worsen, and new damage or secondary defects will appear in other parts, including lining cracks. The second function of this invention is to calculate the future development of defects, allowing for targeted inspections and maintenance based on predictions to prevent the aforementioned situations from occurring.

[0113] The prediction is mainly divided into two aspects: First, given the variables x1, x2, x3, time x4 and risk control level among the influencing factors, the comprehensive evaluation value of lining crack damage under given working conditions can be calculated over time using formula (16), and a curve about D-x4 can be plotted to control the development of damage in the future.

[0114] Second, given a preferred comprehensive damage evaluation value, the risk of the comprehensive damage value of the tunnel lining structure exceeding the preferred value in the future is predicted, provided that the comprehensive damage value of the tunnel lining structure has not yet exceeded the preferred value at the current time. Assuming the current time is x4, and setting the comprehensive damage value control level to D = 82.5, the conditional risk of the comprehensive damage value exceeding a certain limit in the x4 + Δx4 time interval is calculated according to formula (12), and a function curve of the conditional risk P with respect to Δx4 is plotted. One practical engineering application of this invention is that a crack defect inspection mode can be obtained based on the prediction curve. This is a non-periodic inspection mode, and the inspection cycle will change accordingly with the service life of the tunnel lining structure to control the development of tunnel lining crack defects to always be kept below the preferred risk: x4 of the prediction curve represents the current service life of the lining structure, Δx4 represents the next inspection interval, and the subsequent tunnel structure inspection work is scheduled based on Δx4.

[0115] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for diagnosing and predicting cracks in tunnel lining, characterized in that, Includes the following steps: S1 establishes a comprehensive evaluation system for the degree of damage to tunnel lining cracks in a hierarchical structure; S2 establishes a cumulative model of crack damage in tunnel lining based on a comprehensive evaluation system of crack damage degree and influencing factor vectors; Specifically, S2 is: based on the traditional proportional risk model and survival analysis method, considering the influence of various potential influencing factors, a mathematical relationship is established between the comprehensive evaluation value of crack damage and the vector of influencing factors, namely, the damage accumulation model. The damage accumulation model is as follows: In the formula, D is the comprehensive evaluation value of crack damage, α and β are the scale parameter and shape parameter in the Weibull distribution, and X = (x0, x1, x2, ..., x k ) T Let A be a vector of covariates with k+1 variables, and let B be the coefficient vector of the covariates with k+1 coefficients respectively. S3 establishes a parameter iterative estimation model based on the Monte Carlo Markov chain method and obtains the parameter estimation results; S4 is based on the damage accumulation model and parameter estimation results to obtain a tunnel lining crack diagnosis and prediction model, and to diagnose and predict tunnel lining cracks. The specific method for obtaining the comprehensive evaluation value of crack damage is as follows: Based on the analytic hierarchy process (AHP) and variable fuzzy theory, and combined with multi-source data obtained from engineering experience and practical tunnel lining crack detection methods, a hierarchical comprehensive evaluation system for tunnel lining crack damage was established by selecting criterion layers including geometric morphology and crack density, as well as index layers for crack length, crack width, and crack depth. By calculating weights and membership degrees, the comprehensive evaluation value of crack damage in the lining section was calculated.

2. The method for diagnosing and predicting tunnel lining cracks according to claim 1, characterized in that, The comprehensive evaluation system includes: an indicator layer, a criterion layer, and a target layer; the indicator layer includes: crack length, crack width, crack depth, and crack number; the criterion layer includes: crack geometry and crack density; and the target layer includes the comprehensive evaluation value of tunnel lining crack damage.

3. The method for diagnosing and predicting tunnel lining cracks according to claim 1, characterized in that, The comprehensive evaluation value of crack damage includes four damage levels: healthy, sub-healthy, diseased, and critically diseased.

4. The method for diagnosing and predicting tunnel lining cracks according to claim 1, characterized in that, Specifically, S3 is: Based on the test data of engineering cases, potential influencing factors affecting tunnel lining cracks were initially screened. Then, the values ​​of various potential influencing factors were statistically analyzed, and univariate factor analysis was performed on all sections of the tunnel to examine the significant relationship between the severity of cracks and each potential influencing factor. On this basis, data normalization preprocessing was then performed. Using the comprehensive evaluation value of crack damage as the dependent variable and multiple influencing factors, including time, as independent variables, the parameter distribution of each variable is solved in reverse.

Images