A Risk Assessment Method for Highway Tunnel Construction Based on Improved Combined Weighting and Cloud Model
By improving the method of combined weighting and cloud modeling, and combining subjective and objective weights, the problems of weight bias and ambiguity in the risk assessment of highway tunnel construction were solved, and a more accurate and reliable risk level assessment was achieved.
Patent Information
- Application Number
- CN202611324490.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-28
- Publication Date
- 2026-11-17
AI Technical Summary
Existing risk assessment methods for highway tunnel construction suffer from significant deviations in determining indicator weights and an inability to simultaneously reflect randomness and fuzziness, resulting in low accuracy and reliability of construction risk assessments.
By adopting an improved combination weighting and cloud model approach, a risk assessment index system is constructed. Combining subjective and objective weights, a linear combination evaluation model and a cloud model are used to determine the final combination weights. Based on expert scoring data, the membership degree of the risk level is calculated, and the construction risk level is finally determined.
It improves the accuracy and reliability of risk assessment for highway tunnel construction, overcomes the difficulties in consistency verification and the problem of result lag in traditional methods, and enhances the scientific nature and robustness of the assessment results.
Smart Images

Figure CN123117802A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tunnel construction risk assessment technology, and in particular to a method for assessing highway tunnel construction risk based on improved combined weighting and cloud models. Background Technology
[0002] With the increasing number of long tunnels in complex mountainous areas, tunnel construction faces complex geological conditions such as high ground stress, large deformation, sudden water inrush, and fault fracture zones. Construction risks exhibit multi-source, coupled, random, and dynamic characteristics. Current technologies struggle to effectively quantify the coupling relationship between random fluctuations in geological parameters and the fuzzy semantics of expert judgments, limiting the accuracy of early warnings in high-risk areas such as fault fracture zones. Furthermore, traditional models lack dynamic update mechanisms and cannot respond in real-time to parameter changes during construction, causing evaluation results to lag behind actual working conditions. This static and singular assessment system not only weakens the targeted nature of risk management but also increases the probability of safety accidents, necessitating the introduction of a new evaluation framework that can integrate subjective and objective information and characterize uncertainty.
[0003] However, the existing technologies have the following problems: most existing risk assessment methods adopt a single subjective or objective weighting method, which results in large deviations in the determination of indicator weights and an inability to simultaneously reflect randomness and fuzziness, thus leading to low accuracy and reliability of highway tunnel construction risk assessment. Summary of the Invention
[0004] In view of this, the purpose of this invention is to provide a method for assessing the construction risk of highway tunnels based on improved combined weighting and cloud models, so as to alleviate the technical problem of low accuracy and reliability in the assessment of construction risk of highway tunnels.
[0005] In a first aspect, embodiments of this application provide a method for assessing the construction risk of highway tunnels based on improved combined weighting and cloud models, the method comprising: Obtain basic data on the construction of the highway tunnel to be evaluated, and conduct cross-analysis on the basic data to construct a risk assessment index system; A judgment matrix is constructed based on the risk assessment index system. The judgment matrix is then transformed to obtain the optimal transmission matrix. The optimal consistency matrix is determined based on the optimal transmission matrix. Finally, the subjective weight of each index in the risk assessment index system is obtained by geometric averaging the optimal consistency matrix. An evaluation index data matrix is constructed based on the observed values of each index, and the objective weight of each index is determined based on the evaluation index data matrix. Using the subjective weights and objective weights as two weighting subjects, a linear combination evaluation model is constructed. The linear combination evaluation model is solved to determine the improved combination coefficients. Based on the improved combination coefficients, the final combination weight of each indicator is determined. The process involves obtaining expert scoring data for each indicator, determining cloud parameters for each indicator based on the scoring data to obtain an indicator cloud, determining a standard cloud for each risk level based on a pre-defined risk level range, determining the relative membership degree of each indicator relative to each risk level based on the indicator cloud and the standard cloud, and weighting and summarizing the relative membership degrees using the final combined weights to obtain the comprehensive membership degree of each risk level. The risk level with the highest comprehensive membership degree is determined as the final risk assessment level, thus completing the risk assessment for highway tunnel construction.
[0006] In one implementation of the first aspect, the subjective weights are obtained based on the geometric mean of the optimal consistency matrix, as follows: , in, Subjective weighting, Denotes the first step of the optimal consistency matrix. The geometric mean is obtained by taking the geometric mean of the row elements. The number of risk assessment indicators used for comparison at the same level.
[0007] In one implementation of the first aspect, the process of constructing an evaluation index data matrix based on the observed values of each index, and determining the objective weight of each index according to the evaluation index data matrix, includes: An evaluation index data matrix is constructed based on the observed values of each index. Represented as: , in, The number of objects to be evaluated. The number of evaluation index values; Based on the evaluation index data matrix, the correlation coefficient and conflict rate are determined respectively, and expressed as follows: , , in, Indicators and The conflict between them Indicators and The correlation coefficient between them Indicates the first The average value of each evaluation indicator Indicates the first The average value of each evaluation indicator Indicates the first The first evaluation subject in the first The observed values of each evaluation indicator. Indicates the first The first evaluation subject in the first The observed values of each evaluation indicator; The information entropy is determined based on the evaluation index data matrix and is expressed as follows: , in, Indicates the first Information entropy of each evaluation indicator Indicates the first Under the evaluation index, the first The weights of each sample; The objective weights are determined based on the correlation coefficient, the conflict factor, and the information entropy, and are expressed as follows: , in, For the first Objective weights of evaluation indicators.
[0008] In one implementation of the first aspect, the process of constructing a linear combination evaluation model by using the subjective weights and the objective weights as two weighting subjects includes: The set of weight vectors is determined based on the subjective weights and the objective weights, and is represented as follows: , in, Indicates that there is The number of species empowered The number of weights; A linear combination evaluation model is constructed based on the aforementioned set of weight vectors, expressed as follows: , in, For possible evaluation results, These are the coefficients of the linear combination.
[0009] In one implementation of the first aspect, the process of solving the linear combination evaluation model to determine the improved combination coefficients, and determining the final combination weight of each indicator based on the improved combination coefficients, includes: The linear combination evaluation model is constructed into a deviation minimization model using game theory optimization, and the first derivative of the deviation minimization model is optimized to obtain a system of linear equations. Solving the linear system of equations to determine the improved combination coefficients is expressed as follows: , in, To improve the combination coefficients; Substituting the improved combination coefficients into the linear combination model, we obtain the final combination weights, expressed as: , in, This represents the final combined weights.
[0010] In one implementation of the first aspect, the process of determining the relative membership of each indicator with respect to each risk level based on the indicator cloud and the standard cloud includes: The equivalent variance is obtained by combining the entropy and hyperentropy of the index cloud, and the standard equivalent variance is obtained by combining the entropy and hyperentropy of the standard cloud. The one-dimensional Gaussian KL divergence and Jeffreys divergence are determined based on the equivalent variance and the standard equivalent variance. Information similarity is determined based on the KL divergence and the Jeffreys divergence. The information similarity is normalized to obtain the relative membership degree of each risk level.
[0011] In one implementation of the first aspect, the optimal transmission matrix is obtained by performing a transmission transformation on the judgment matrix, as follows: , in, The optimal transfer matrix is represented by the following elements: , in, Evaluation indicators relative to evaluation indicators The comprehensive transfer comparison value, and are the elements of the judgment matrix.
[0012] In one implementation of the first aspect, the optimal consistency matrix is determined based on the optimal transfer matrix and obtained as follows: , in, The optimal consistency matrix is represented by the following elements: , in, Indicators relative to evaluation indicators Scale value, For the natural constant An exponential function with base 0.
[0013] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.
[0014] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description
[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0016] Figure 1 This is a flowchart illustrating the method for assessing construction risks of highway tunnels based on improved combined weighting and cloud models, as provided in an embodiment of the present invention. Figure 2 A comparison diagram of the comprehensive cloud and standard cloud of the Muli Tunnel construction risk assessment criterion layer provided for embodiments of the present invention; Figure 3 A comparison chart of the comprehensive cloud and standard cloud for the construction risk assessment of the Muli Tunnel provided in this embodiment of the invention. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Please see Figure 1 As shown, it is a flowchart illustrating the method for assessing the construction risk of highway tunnels based on improved combined weighting and cloud models provided in an embodiment of the present invention.
[0019] To facilitate understanding of this embodiment, a detailed description of the highway tunnel construction risk assessment method based on improved combined weighting and cloud model disclosed in this embodiment of the invention will be provided first, including: Step S1: Obtain basic data on the construction of the highway tunnel to be evaluated, and conduct cross-analysis on the basic data to construct a risk assessment index system; Step S2: Construct a judgment matrix based on the risk assessment index system, perform a transmission transformation on the judgment matrix to obtain the optimal transmission matrix, determine the optimal consistency matrix based on the optimal transmission matrix, and perform a geometric average on the optimal consistency matrix to obtain the subjective weight of each index in the risk assessment index system. Step S3: Construct an evaluation index data matrix based on the observed values of each index, and determine the objective weight of each index according to the evaluation index data matrix; Step S4: Using the subjective weight and the objective weight as two weighting subjects, construct a linear combination evaluation model, solve the linear combination evaluation model to determine the improved combination coefficient, and determine the final combination weight of each indicator based on the improved combination coefficient; Step S5: Obtain expert scoring data for each indicator, determine the cloud parameters of each indicator based on the scoring data to obtain the indicator cloud, and determine the standard cloud for each risk level based on the pre-set risk level range. Determine the relative membership degree of each indicator relative to each risk level according to the indicator cloud and the standard cloud, and use the final combined weight to perform weighted summation of the relative membership degree to obtain the comprehensive membership degree of each risk level. Determine the risk level with the largest comprehensive membership degree as the final risk assessment level, and complete the highway tunnel construction risk assessment.
[0020] Specifically, this invention constructs a risk assessment index system by cross-analyzing basic tunnel construction data. It uses an improved judgment matrix derived from the optimal transmission matrix to obtain the subjective weights of each index and calculates the objective weights based on the index observation data matrix. These subjective and objective weights are used as two weighting subjects. An improved combination coefficient is solved through a linear combination evaluation model to obtain the final combined weights. An index cloud and a risk level standard cloud are generated based on expert scoring data. The membership degree of each index to the risk level is calculated. After weighted aggregation, the level corresponding to the highest comprehensive membership degree is taken as the tunnel construction risk assessment result. The improved transmission transformation consistency matrix overcomes the difficulty of consistency testing in the traditional analytic hierarchy process. The problem of repeatedly adjusting the judgment matrix is addressed, making the subjective weighting process more efficient and the results more robust. The combination of subjective and objective weighting integrates the inherent laws of expert experience and measured data, avoiding the subjective bias or data distortion of a single weighting method, and improving the scientificity and credibility of the weights. The cloud model can simultaneously characterize the fuzziness and randomness of expert language scoring, making the handling of risk level boundaries more in line with the uncertainties of actual engineering. The evaluation results have better robustness and interpretability. The indicator system constructed from cross-analysis of basic data covers key risk factors in tunnel construction, ensuring the comprehensiveness and pertinence of the evaluation, thereby improving the accuracy and reliability of highway tunnel construction risk assessment.
[0021] Please see Figure 2 and Figure 3 As shown, it is a comparison chart of the comprehensive cloud and standard cloud of the risk assessment criteria layer for the construction of Muli Tunnel, and a comparison chart of the comprehensive cloud and standard cloud of the risk assessment for the construction of Muli Tunnel.
[0022] from Figure 2 As can be seen from the data, in the criterion layer, geological condition risk I1 and surrounding environment risk I2 have the highest risk levels, falling between high and medium risk, but closer to high risk, and are therefore classified as high risk. Personnel and equipment risk I3 is generally between moderate and high risk, and is classified as moderate risk. Tunnel self-risk I4 is generally classified as high risk. Figure 3 It can be seen that the comprehensive risk assessment of the construction risk of the Muli Tunnel entrance section from MZK8+527 to MZK10+000 is between high and medium risk, and is close to high risk. Therefore, the result of the construction risk assessment of the Muli Tunnel is "high risk".
[0023] Specifically, the embodiments of the present invention collect basic data on the highway tunnel project to be evaluated, including: engineering geological survey data; construction organization design; advanced geological forecast data; monitoring and measurement data; construction log; surrounding environment survey data; and expert survey questionnaires. Among them, the engineering geological survey data includes: surrounding rock grade; groundwater conditions; adverse geological conditions; rock bursts; large deformations; and sudden water inrushes.
[0024] Specifically, this embodiment of the invention is based on the construction organization design, construction plan, construction sequence, and sub-item division of the tunnel to be evaluated. Following the principles of "complete coverage, hierarchical exclusivity, accountability, and risk identification," the entire construction process is decomposed into a Work Breakdown Structure (WBS) unit. The WBS unit includes primary and secondary work units. The primary work unit corresponds to the main construction stages or sub-projects that can be independently organized and managed, including portal works, advanced support, tunnel excavation, initial support, invert construction, secondary lining, drainage and waterproofing construction, and ancillary works. The secondary work unit is further divided according to construction procedures and risk control nodes. For example, when using the drill-and-blast method for tunnel excavation, it can be divided into work units such as surveying and setting out, drilling, charging, blasting, ventilation and smoke extraction, handling of loose rocks, muck removal, and support. If the project uses mechanical excavation or other construction methods, the secondary work units are adjusted according to the corresponding construction plan.
[0025] Specifically, based on the sources of risk, the objects affected by the risk, and the responsibility for risk control, the embodiments of the present invention decompose the risks of highway tunnel construction into four categories of RBS risks: geological condition risks, surrounding environment risks, personnel and equipment risks, and tunnel-specific risks. Among them, geological condition risks originate from surrounding rock, groundwater, and unfavorable geological formations; surrounding environment risks originate from adjacent buildings, roads, pipelines, and the ecological environment; personnel and equipment risks originate from personnel behavior, organizational management, machinery and equipment, and working conditions; and tunnel-specific risks originate from structural form, burial depth, support condition, and construction quality.
[0026] Specifically, in classifying risk items, the embodiments of the present invention use engineering survey data, construction organization design, monitoring records, existing accident cases and expert investigation results as evidence, and add or delete specific risk items according to the principle of "distinguishing sources, non-duplication of indicators, monitorable risks and corresponding measures".
[0027] Specifically, in this embodiment of the invention, the overall risk of highway tunnel construction is taken as the target layer, and the geological condition risk, surrounding environment risk, personnel and equipment risk, and tunnel itself risk are taken as the criteria layer. The WBS work unit and RBS risk category are cross-referenced to identify candidate risk events corresponding to the combination of "specific process - specific risk source", and observable and evaluable risk events are set as the indicator layer.
[0028] Specifically, in this embodiment of the invention, candidate indicators are selected according to the following rules: they have engineering data, monitoring data, or expert judgment as a basis; they have a clear causal or correlation relationship with the tunnel construction safety results; the meanings of different indicators are not redundant; they can be evaluated through quantitative data or rating scales; and they can correspond to specific control measures. After review by the project technical leader, safety management personnel, and relevant professional experts, a three-level risk assessment indicator system is formed.
[0029] Specifically, this invention focuses on the core objective of risk assessment in highway tunnel construction, breaking down the complex assessment problem into a hierarchical structure of target layer, criterion layer, and indicator layer, clarifying the hierarchical relationship and internal logic of indicators at each level, and building a clear analytical framework for subsequent weight calculation.
[0030] Specifically, embodiments of the present invention use a three-level scaling method (-1, 0, 1) to construct a judgment matrix, containing... Judgment matrix of each indicator Represented as: , Among them, the elements in the matrix Indicates the first The risk assessment indicator is relative to the first The importance of each risk indicator.
[0031] In this embodiment of the invention, a scale value of -1 represents a factor. With factors Compared to unimportant; a scale value of 0 indicates that the factor is not important. With factors Equally important; a scale value of 1 indicates that the factor is... With factors It is more important than that.
[0032] Specifically, the embodiments of the present invention are based on a judgment matrix. Calculate the optimal transfer matrix By summing and averaging the row and column elements of the judgment matrix, the initial judgment information is integrated and optimized to obtain the optimal transmission matrix. Represented as: , in, As an indicator relative to indicators The comprehensive transfer comparison value, according to Perform calculations. The number of risk assessment indicators used for comparison at the same level.
[0033] Specifically, embodiments of the present invention construct an optimal consistency matrix based on an optimal transmission matrix. ,according to Calculate the optimal consistency matrix ,in Indicators relative to indicators The scale value; For the natural constant An exponential function with base α maps the linearly transformed judgment information into a consistent positive matrix, providing a stable mathematical foundation for weight calculation and yielding the optimal consistency matrix. Represented as: .
[0034] Specifically, the embodiments of the present invention address the optimal consistency matrix. The The geometric mean is obtained by taking the geometric mean of the row elements. Then, the geometric mean of all rows is normalized to obtain the final subjective weights, completing the weight solution process of the improved analytic hierarchy process. The subjective weights are expressed as: .
[0035] Specifically, this invention employs an improved analytic hierarchy process (AHP) to solve for the subjective weights of risk indicators: a judgment matrix is constructed using a three-level scaling method; the optimal transmission matrix is obtained through row and column summation and averaging; then, the optimal consistency matrix is generated through exponential transformation; finally, the subjective weights are obtained by geometric averaging and normalization of each row. The three-level scaling method reduces the difficulty of expert judgment and subjective arbitrariness, avoiding repeated adjustments caused by unsuccessful consistency checks. The optimal transmission matrix integrates and optimizes the initial judgment information through row and column summation and averaging, suppressing the excessive influence of individual judgment biases and making the importance transmission relationship between indicators more robust. The optimal consistency matrix is constructed by exponential mapping, satisfying the consistency requirements of the judgment matrix and overcoming the problems of cumbersome consistency checks, blind adjustments, and unstable results in the traditional AHP. Geometric averaging and normalization make the weight allocation smoother and more reasonable, avoiding extreme weights and improving the reliability and discriminative power of the weight results, thereby improving the accuracy and reliability of highway tunnel construction risk assessment.
[0036] Specifically, embodiments of the present invention construct a dataset matrix; for example, it is assumed that there exists... Individuals to be evaluated and If there are 10 evaluation index values, then the constructed evaluation index data matrix is represented as follows: , in, It indicates the first The first evaluation object corresponds to the first j Each evaluation indicator value.
[0037] In this embodiment of the invention, due to the differences in the dimensions and orders of magnitude of each indicator, the original data needs to be standardized. The range transformation method is used to achieve dimensionless processing of positive and negative indicators.
[0038] Specifically, for positive indicators, the numerical values in the embodiments of the present invention are... The larger the value, the better the evaluation effect. In this case, a positive processing method is used, and the processing formula is: , For inverse indicators, a larger value actually indicates a worse evaluation effect. In this case, a reverse processing method is used, and the processing formula is as follows: , in, , These are the maximum and minimum values of the indicator, respectively; , This is the standardized dimensionless value.
[0039] Specifically, in this embodiment of the invention, standard deviation is used to characterize the dispersion of data within each indicator. The larger the standard deviation, the more significant the fluctuation of the indicator among samples, the richer the information it contains, and the higher its weight should be assigned. The formula for calculating the standard deviation is: , , in, For the first The average of the indicators, For the first j The standard deviation of each indicator.
[0040] Specifically, the embodiments of this invention calculate the conflict of indices, and the correlation coefficient is often used in the CRITIC method. To reflect the conflict between data, a high correlation coefficient between two indicators indicates a strong correlation between them; conversely, a small difference between two indicators suggests a high degree of information overlap, thus requiring a reduction in their weighting. The formula for calculating indicator conflict is as follows: , , in, Indicators and The conflict between them; Indicators and The correlation coefficient between them is calculated using the absolute value.
[0041] Specifically, this embodiment of the invention introduces the entropy method to compensate for the shortcomings of the CRITIC method in characterizing the dispersion of indicators. The entropy value information of each indicator is calculated according to the entropy method. Entropy, as a measure of the uncertainty of probability distribution, can effectively reflect the distribution characteristics of indicator data: the smaller the entropy value, the higher the degree of variation of the indicator and the greater the information content.
[0042] Specifically, the embodiments of the present invention calculate the first The first item under the indicator Weights of each sample , represented as: , Then, based on the weights, the entropy values of each indicator are converted into information utility using the entropy method, and these differences are normalized, thereby achieving objective weighting and obtaining the first... The information entropy of an indicator is expressed as: , in, Indicates the first Information entropy of each indicator.
[0043] Specifically, the embodiments of the present invention calculate the objective weights of each indicator, expressed as follows: , in, For the first The objective weight of each indicator is calculated, and the denominator is the sum of the importance coefficients of all indicators, which serves a normalization function.
[0044] Specifically, this invention constructs an evaluation index data matrix and performs dimensionless processing on both positive and negative indices through range transformation to eliminate dimensional differences. It measures the dispersion of each index data based on standard deviation, calculates the conflict between indices using correlation coefficients, and introduces entropy to calculate the information entropy of each index to compensate for the shortcomings of the CRITIC method in characterizing the dispersion of indices. The objective weight of each index is obtained by combining standard deviation, conflict, and information entropy. Standard deviation reflects the degree of index variation, conflict reflects the degree of information overlap between indices, and entropy reflects the uncertainty of data distribution. The three factors combined comprehensively characterize the importance of indices from three dimensions: dispersion, independence, and information content. This overcomes the problem of a one-sided perspective in a single objective weighting method, thereby improving the accuracy and reliability of highway tunnel construction risk assessment.
[0045] Specifically, in this embodiment of the invention, a basic set of weight vectors is determined based on subjective weights and objective weights, namely... ,in,( ).
[0046] In this embodiment of the invention, Indicates that there is The number of species empowered The number of weights is given. Since this embodiment only includes subjective weights and objective weights, therefore... The constructed linear combination evaluation model is expressed as: , in, For possible evaluation results, These are the coefficients of the linear combination.
[0047] Specifically, this embodiment of the invention is based on the concept of "collective rationality" in game theory, and constructs a deviation minimization model with the optimization objective of minimizing the deviation between the overall result and each original weight, as follows: .
[0048] In this embodiment of the invention, the essence of the optimization objective is to make the comprehensive weight as close as possible to the information characteristics of each set of original weights, so as to avoid the deviation of a single weight dominating the final result.
[0049] Specifically, this embodiment of the invention utilizes the properties of matrix differentiation to optimize the first derivative of the objective function of the optimization objective, obtaining a system of linear equations, expressed as: .
[0050] Specifically, in this embodiment of the invention, solving the linear equation system yields the optimal combination weight coefficients. In this step, the optimal combination weight coefficients may have negative values or a sum that is not equal to 1, which does not satisfy the game theory principle of "...". Due to the constraints of the original coefficients, an improved rule is introduced to address the limitations of the original coefficients: taking the absolute value of the original coefficients and then normalizing them, we obtain improved combination coefficients that satisfy non-negativity and sum to 1. The improved combination coefficients are expressed as: .
[0051] In this embodiment of the invention, the improved combination coefficients break through the constraints of traditional game theory, ensuring that the coefficients can be effectively applied in any scenario, thereby improving the robustness of the method.
[0052] Specifically, in this embodiment of the invention, the improved combination coefficients are substituted into the linear combination model to obtain the final combination weights, which are expressed as follows: .
[0053] In this embodiment of the invention, the final combined weight integrates the engineering experience of subjective weights with the data characteristics of objective weights, which not only retains the core information of the original weights, but also reduces the bias of single weighting through game equilibrium logic.
[0054] Specifically, this invention constructs a linear combination evaluation model by using subjective and objective weights as two weighting subjects. The optimal combination coefficients are obtained with the goal of minimizing game-theoretic deviation. During the solution process, an improved rule of taking the absolute value and then normalizing is introduced for the original coefficients that may be negative or non-normalized, resulting in improved combination coefficients that satisfy non-negativity and normalization constraints. These improved coefficients are then substituted into the linear model to obtain the final combination weights of each indicator. The goal of minimizing game-theoretic deviation ensures that the comprehensive weights are as close as possible to the original subjective and objective weights, balancing expert experience judgment with the patterns of measured data. This avoids the bias of a single weighting method dominating the final result. The improved combination coefficients, through absolute value normalization, enhance the robustness and engineering applicability of the method under different data scenarios. The final combination weights simultaneously inherit the logical rationality of subjective weighting and the data sensitivity of objective weighting, making the highway tunnel construction risk assessment results both credible to experts and supported by field data. This makes the evaluation conclusions more accurate and reliable, thereby improving the accuracy and reliability of highway tunnel construction risk assessment.
[0055] Specifically, this invention relies on cloud model-related theories and uses a cloud generator to calculate standard cloud parameters for corresponding indicators, including hyperentropy. The value of is generally a constant, and the expectation and entropy of the standard cloud are expressed as follows: , in, and These represent the upper and lower limits of the set evaluation range, respectively.
[0056] Specifically, this embodiment of the invention is based on expert ratings of the actual conditions of the indicators, and uses an inverse cloud model to derive the mean, variance, and cloud parameters of each indicator's rating. The rating variance is expressed as: , Cloud parameters are represented as follows: .
[0057] Specifically, in this embodiment of the invention, after obtaining the comprehensive evaluation cloud and standard cloud for each indicator, cloud similarity is introduced to quantify the evaluation level. The higher the cloud similarity value, the higher the degree of fit between the current risk status and the corresponding evaluation level. The final evaluation level is the evaluation interval level with the largest similarity value.
[0058] Specifically, the embodiments of the present invention improve the relevant calculations of the cloud model. Instead of directly calculating according to the traditional similarity, it introduces one-dimensional Gaussian KL divergence and Jeffreys divergence, also known as symmetric information distance, and determines the final risk assessment level by calculating the relative membership degree.
[0059] Specifically, this embodiment of the invention utilizes a Python forward cloud generator to input the digital features of a standard cloud and a comprehensive evaluation cloud, generating corresponding cloud maps. Comparing the comprehensive evaluation cloud map with the standard cloud map provides a clear view of the evaluation level results. The comprehensive evaluation cloud and the standard cloud exhibit inherent similarity based on their cloud models, and this similarity can be quantified by calculating their closeness. The cloud parameters and equivalent variance are related to the... Each indicator's cloud parameter is represented as follows: , No. The cloud parameters for each standard level are represented as follows: .
[0060] Specifically, embodiments of the present invention synthesize entropy and hyperentropy to obtain an equivalent variance that integrates the uncertainty scale. The equivalent variance of each indicator is expressed as: , No. The standard equivalent variance of each standard level is expressed as follows: , in, As expected, Characterizing discreteness, Characterization Fluctuations; This is a comprehensive uncertainty scale.
[0061] Specifically, this invention does not directly calculate cloud similarity using traditional methods. Instead, it introduces one-dimensional Gaussian KL divergence and Jeffreys divergence. KL divergence, also known as relative entropy, is an indicator used to measure the "degree of difference" between two probability distributions. It describes the amount of information loss when using distribution Q to approximate distribution P. One-dimensional Gaussian KL divergence equates a cloud to a one-dimensional Gaussian distribution. The KL divergence formula for two one-dimensional Gaussian distributions is expressed as: .
[0062] Jeffreys divergence is a symmetric measure of probability distribution dissimilarity, proposed to address the asymmetry limitations of KL divergence, and is used to measure index clouds. With standard level cloud The symmetrical difference between them can be expressed by the Jeffreys divergence formula as: , in, When the two distributions are consistent, .
[0063] Then, the information distance is mapped to information similarity using information similarity and membership degree, and expressed as: .
[0064] Specifically, the embodiments of the present invention address each indicator. The information similarity is normalized row by row to obtain its relative membership degree to each level, which is expressed as: , in, Represented as indicators Belongs to the level The degree of relative subordination .
[0065] Specifically, in this embodiment of the invention, the relative membership degree is weighted and summarized using the final combined weight to obtain the comprehensive membership degree of each risk level, and the risk level with the largest comprehensive membership degree is determined as the final risk assessment level.
[0066] Specifically, this invention generates indicator clouds for each indicator and standard clouds for each risk level using a cloud model. Entropy and hyperentropy are combined into an equivalent variance, and the cloud is equated to a one-dimensional Gaussian distribution. One-dimensional Gaussian KL divergence and symmetric Jeffreys divergence are introduced to calculate the information distance between the indicator cloud and the standard cloud. This distance is then mapped to information similarity and normalized row-wise to obtain the relative membership degree of each indicator to each risk level. A weighted summation using a combination of subjective and objective factors yields the comprehensive membership degree of each risk level. The level corresponding to the highest membership degree is used as the final evaluation result. KL divergence and Jeffreys divergence replace traditional cloud similarity, measuring the closeness between clouds from the perspective of overall probability distribution differences, thus avoiding the influence of cloud droplet randomness. The similarity fluctuations of the indicators and the evaluation results are more stable and accurate. Jeffreys divergence is a symmetric improvement of KL divergence, which makes up for the one-way comparison bias caused by the asymmetry of KL divergence. It makes the measurement of the difference between the indicator cloud and the standard cloud more objective and reasonable. The equivalent variance jointly represents entropy and hyperentropy, which comprehensively reflects the discreteness and uncertainty fluctuations of the cloud model. It makes the fuzziness and randomness of risk information more fully reflected. After the relative membership normalization, the degree of fit between each indicator and each level is comparable. Combined with the weighted summation of combined weights, it conveys subjective experience and objective data information, reduces the error of single weighting or single similarity calculation, and thus improves the accuracy and reliability of highway tunnel construction risk assessment.
[0067] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present application should be included within the scope of protection of the present application.
Claims
1. A method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models, characterized in that, include: Obtain basic data on the construction of the highway tunnel to be evaluated, and conduct cross-analysis on the basic data to construct a risk assessment index system; A judgment matrix is constructed based on the risk assessment index system. The judgment matrix is then transformed to obtain the optimal transmission matrix. The optimal consistency matrix is determined based on the optimal transmission matrix. Finally, the subjective weight of each index in the risk assessment index system is obtained by geometric averaging the optimal consistency matrix. An evaluation index data matrix is constructed based on the observed values of each index, and the objective weight of each index is determined based on the evaluation index data matrix. Using the subjective weights and objective weights as two weighting subjects, a linear combination evaluation model is constructed. The linear combination evaluation model is solved to determine the improved combination coefficients. Based on the improved combination coefficients, the final combination weight of each indicator is determined. The process involves obtaining expert scoring data for each indicator, determining cloud parameters for each indicator based on the scoring data to obtain an indicator cloud, determining a standard cloud for each risk level based on a pre-defined risk level range, determining the relative membership degree of each indicator relative to each risk level based on the indicator cloud and the standard cloud, and weighting and summarizing the relative membership degrees using the final combined weights to obtain the comprehensive membership degree of each risk level. The risk level with the highest comprehensive membership degree is determined as the final risk assessment level, thus completing the risk assessment for highway tunnel construction.
2. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The subjective weights are obtained by taking the geometric mean of the optimal consistency matrix, and are acquired as follows: , in, Subjective weighting, Denotes the first step of the optimal consistency matrix. The geometric mean is obtained by taking the geometric mean of the row elements. The number of risk assessment indicators compared at the same level.
3. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The process of constructing an evaluation index data matrix based on the observed values of each index, and determining the objective weight of each index based on the evaluation index data matrix, includes: An evaluation index data matrix is constructed based on the observed values of each index. Represented as: , in, The number of objects to be evaluated. The number of evaluation index values; Based on the evaluation index data matrix, the correlation coefficient and conflict rate are determined respectively, and expressed as follows: , , in, Indicators and The conflict between them Indicators and The correlation coefficient between them Indicates the first The average value of each evaluation indicator Indicates the first The average value of each evaluation indicator Indicates the first The first evaluation subject was in the first The observed values of each evaluation indicator. Indicates the first The first evaluation subject was in the first The observed values of each evaluation indicator; The information entropy is determined based on the evaluation index data matrix and is expressed as follows: , in, Indicates the first Information entropy of each evaluation indicator Indicates the first Under the evaluation index, the first The weights of each sample; The objective weights are determined based on the correlation coefficient, the conflict factor, and the information entropy, and are expressed as follows: , in, For the first Objective weights of evaluation indicators.
4. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The process of constructing a linear combination evaluation model by using the subjective weights and the objective weights as two weighting subjects includes: The set of weight vectors is determined based on the subjective weights and the objective weights, and is represented as follows: , in, Indicates that there is The number of species empowered The number of weights; A linear combination evaluation model is constructed based on the aforementioned set of weight vectors, expressed as follows: , in, For possible evaluation results, These are the coefficients of the linear combination.
5. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The process of solving the linear combination evaluation model to determine the improved combination coefficients, and determining the final combination weight of each indicator based on the improved combination coefficients, includes: The linear combination evaluation model is constructed into a deviation minimization model using game theory optimization, and the first derivative of the deviation minimization model is optimized to obtain a system of linear equations. Solving the linear system of equations to determine the improved combination coefficients is expressed as follows: , in, To improve the combination coefficients; Substituting the improved combination coefficients into the linear combination model, we obtain the final combination weights, expressed as: , in, This represents the final combined weights.
6. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The process of determining the relative membership of each indicator with respect to each risk level based on the indicator cloud and the standard cloud includes: The equivalent variance is obtained by combining the entropy and hyperentropy of the index cloud, and the standard equivalent variance is obtained by combining the entropy and hyperentropy of the standard cloud. The one-dimensional Gaussian KL divergence and Jeffreys divergence are determined based on the equivalent variance and the standard equivalent variance. Information similarity is determined based on the KL divergence and the Jeffreys divergence. The information similarity is normalized to obtain the relative membership degree of each risk level.
7. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The optimal transmission matrix is obtained by performing a transmission transformation on the judgment matrix, and is acquired as follows: , in, The optimal transfer matrix is represented by the following elements: , in, Evaluation indicators Compared to evaluation indicators The comprehensive transfer comparison value, and are the elements of the judgment matrix.
8. The method for assessing construction risk of highway tunnels based on improved combined weighting and cloud models according to claim 1, characterized in that, The optimal consistency matrix is determined based on the optimal transfer matrix and is obtained as follows: , in, The optimal consistency matrix is represented by the following elements: , in, Indicators Compared to evaluation indicators Scale value, For the natural constant An exponential function with base 0.
