Improved ins-marcos-based safety risk assessment method for hydraulic engineering construction

Abstract

This invention discloses an improved INS-MARCOS-based method for assessing the construction safety risks of water conservancy projects, comprising the following steps: First, constructing a construction safety risk assessment index system; second, evaluating expert qualifications (BWM method) and extracting the information entropy of the expert semantic evaluation matrix to verify the scoring validity (AEWM method), generating expert combination weights through combined weighting; subsequently, converting the qualitative evaluations of multiple experts into interval intelligence numbers (INS), and using the interval intelligence weighted average operator (INWA) with expert weights as the index to aggregate multi-source risk information; finally, introducing an improved MARCOS model to construct dual benchmarks of absolute safety and accident limits, calculating the comprehensive utility measure value of each risk indicator and arranging them in descending order of hazard. This invention effectively overcomes the problems of uneven distribution of expert discourse power and information ambiguity in group decision-making, realizing the identification of core risk indicators at water conservancy construction sites, and providing reliable quantitative support for disaster prevention decision-making.

Description

Technical Field

[0001] This invention relates to the field of water conservancy project safety management and multi-criteria decision analysis technology, specifically a water conservancy project construction safety risk assessment method based on an improved INS-MARCOS. Background Technology

[0002] Water conservancy project construction includes multiple stages such as dam main body pouring, deep-buried long-distance water diversion tunnel excavation, steep slope support, and large-scale electromechanical equipment installation. It involves high-altitude cliff operations, cross-operation of large heavy-duty machinery groups, and operations in confined spaces above and below water. In particular, the construction of large-scale water conservancy projects located in deep mountain canyon areas is characterized by extremely complex geological structures, a high susceptibility to sudden fault mudslides, water inrushes, and rock bursts, and severe constraints from flood season and extreme weather conditions. The construction environment is extremely harsh, with long incubation periods for potential hazards and a high degree of unpredictability. It is a highly comprehensive and systematic project.

[0003] Water conservancy projects face harsh construction environments, complex operations, and numerous influencing factors, making them highly susceptible to major safety accidents involving mass casualties. Currently, the industry relies heavily on conventional expert group decision-making and traditional multi-criteria evaluation models for construction safety risk assessments. The weighting of these models often depends on the subjective judgment of the organizers, making it difficult to scientifically quantify the true qualifications and scoring quality of experts. Furthermore, traditional evaluation scales are too general to effectively express and quantify the "hesitation" and uncertainty of experts under complex construction conditions. Therefore, the inherent limitations of expert data dispersion, arbitrary subjective weighting, and the lack of accident limits and measurement benchmarks in water conservancy construction safety assessments can easily lead to inaccurate core risk assessment results. This can result in the failure to accurately identify and promptly eliminate fatal safety hazards during construction, potentially leading to irreversible engineering disasters. Summary of the Invention

[0004] This invention provides a method for assessing the construction safety risks of water conservancy projects based on an improved INS-MARCOS framework, utilizing interval intelligence collection... Representing hesitation information, combined with improvement and Accurately empower expert judgments and leverage improvements The model constructs a dual-utility benchmark to achieve robust assessment of risks in water conservancy construction. To achieve the above objectives, this invention adopts the following technical solution: The improved INS-MARCOS-based method for assessing the safety risks of hydraulic engineering construction includes the following steps: Step S1: Construct a risk assessment index system for water conservancy project construction safety; Step S2: Obtain the expert subjective weights and expert objective weights, and merge the subjective weights and objective weights to obtain the comprehensive weight; Step S3: Collect raw assessment data of each risk assessment indicator from multiple experts, and use the preset semantic set mapping model to convert the experts' qualitative comments into interval intelligence numbers to construct the initial interval intelligence decision matrix. Step S4: Define the ideal solution and the anti-ideal solution, and incorporate them into the intelligent decision-making matrix in the initial interval to form the intelligent decision-making matrix in the extended interval; Step S5: Using the interval-based weighted average operator, with the comprehensive weight as the index, integrate the evaluation information of multiple experts into a comprehensive evaluation value for each risk assessment indicator; Step S6: Use the interval intelligent scoring function to defuzzify the comprehensive evaluation value of each risk assessment indicator to obtain the final comprehensive score of each risk assessment indicator, and calculate the utility ratio of the final comprehensive score relative to the ideal solution score and the anti-ideal solution score respectively. Step S7: Use a nonlinear utility function to perform a fusion calculation on the utility degree ratios to obtain the comprehensive utility score of each index relative to the ideal solution and the irrational solution; Step S8: Sort the risk levels of each risk assessment indicator in descending order based on the comprehensive utility score to obtain the risk assessment result.

[0005] Furthermore, in step S1, the evaluation indicators in the water conservancy project construction safety risk assessment index system include: The primary indicators include the impact of corporate organization, safety supervision, on-site operation-related factors, and construction personnel-related factors; The secondary indicators are as follows: Enterprise organizational impact includes organizational structure and responsibilities, safety production investment, and safety management procedures; safety supervision includes risk monitoring and early warning, supervision and management violations, and work plan arrangements; on-site operation related factors include technical measures, materials and machinery, geological conditions, work environment, and weather conditions; and construction personnel related factors include operational violations, skill errors, intuition and decision-making errors, and personnel quality.

[0006] Furthermore, in step S2, the subjective weights are adopted. The calculation method is as follows: Determine the optimal and worst-case indicators, and select the optimal expert from the expert panel based on their qualifications. and the worst expert ; Construct the optimal comparison vector : Optimal Expert Compared to other experts We score the preferences and get: ,in ; Construct the worst comparison vector Other experts Compared to the worst expert We score the preferences and get: ,in ; Solving the linear optimization model yields the optimal subjective weights. Establish the following BWM linear optimization model to minimize the deviation. Minimize:

[0007]

[0008] in: The number of participating experts, Top-ranked experts Compared to the first The relative importance preference scale value of each expert. For the first The expert compared to the least qualified expert The relative importance preference scale value, The first one to be solved The subjective seniority weight of each expert. The most qualified expert to be solved Subjective seniority weighting The least qualified expert to be solved Subjective seniority weighting; The largest absolute error variable used to measure the degree of consistency of the meta-decision-maker preference scaling; Consistency test to calculate consistency ratio ,in This is a consistency indicator under the corresponding scale; when At that time, the judgment is logically consistent.

[0009] Furthermore, in step S2, the objective weights are calculated using the inverse entropy weight method, and the specific process is as follows: extract Experts The original semantic comments given by each indicator are transformed into interval intelligence numbers using a preset mapping rule, and then substituted into the interval intelligence scoring function. Perform deblurring;

[0010] in: and Let these represent the lower and upper bounds of the proper membership function of the numbers in the interval, respectively. and Let these represent the lower and upper bounds of the hesitant membership function of the intelligence number in the interval, respectively. and Let represent the lower and upper bounds of the pseudo-membership function of the intelligence numbers in the interval, respectively; Building a system based on risk indicators Evaluation experts included Real evaluation matrix ,in, Indicates the first The expert commented on the first Precise quantitative scores for each risk indicator; No. The expert's rating was the highest. and The first in its scoring sequence Normalized values ​​of each indicator The calculation formula is:

[0011] Calculate probability proportion :

[0012] Calculate information entropy :

[0013] Calculate objective weights :

[0014] Using inverse entropy To measure the value of information, the higher the value, the higher the weight.

[0015] In step S2, the combined weights Calculated using the linear weighting method:

[0016]

[0017] in: ; The coefficients for assigning subjective weights to experts range from 0 to 1. <1; The expert objective weighting coefficient is assigned, and its value range is [value range missing]. 0< <1 ; The first one calculated by the BWM method The subjective weight of each expert The first value obtained by the anti-entropy weight method The objective weight of each expert.

[0018] Furthermore, in step S3, the three independent components of the intelligence number in the interval are: a true membership interval representing the degree of certainty of the risk existence. Uncertain membership intervals representing the degree of hesitation caused by cognitive limitations And pseudo-membership intervals that characterize the degree of risk. ; in: and These represent the lower and upper bounds of the proper membership function of the intelligence numbers in the interval, respectively, and are used to characterize the credibility of the evaluation information; and These represent the lower and upper bounds of the hesitant membership function of the intelligence number in the interval, respectively, and are used to characterize the degree of uncertainty or ambiguity of the evaluation information; and Let represent the lower and upper bounds of the pseudo-membership function of the intelligence numbers in the interval, respectively, used to characterize the degree of unreliability of the evaluation information; and satisfy: ,and .

[0019] Furthermore, in step S4, the ideal solution is the number of the interval of the minimum value in the semantic set corresponding to the ideal value of the risk indicator, and the anti-ideal solution is the number of the interval of the interval of the maximum value in the semantic set corresponding to the anti-ideal value of the risk indicator.

[0020] Furthermore, in step S5, the calculation formula for the intelligent weighting operator in the interval is:

[0021] in: m The number of risk assessment indicators, This indicates the number of the risk assessment indicator. ; n The number of experts participating in the evaluation. Indicates the evaluation expert's number ,in n This represents the total number of experts who participated in the evaluation. Indicates the first The comprehensive combined weighting of the evaluation experts, and They represent the first The expert commented on the first The lower and upper bounds of the true membership function for evaluating each indicator. and They represent the first The expert commented on the first The lower and upper bounds of the hesitant membership function for evaluating individual indicators. and They represent the first The expert commented on the first The lower and upper bounds of the pseudo-membership function for evaluating each indicator.

[0022] Furthermore, in step S6, the absolute comprehensive score function obtained after defuzzification of the risk assessment index... ):

[0023] in: Indicates the first The absolute comprehensive score is obtained after defuzzifying each risk assessment indicator using a scoring function. and They represent the first The lower and upper limits of the comprehensive true membership degree after aggregating various risk assessment indicators. and They represent the first The lower and upper limits of the overall hesitation membership degree after aggregating various risk assessment indicators. and They represent the first The lower and upper limits of the composite pseudo-membership degree after aggregating various risk assessment indicators.

[0024] Furthermore, in step S7, the formula for calculating the comprehensive utility value is:

[0025] In the formula, , and These are the individual utility function values ​​of the solution relative to the ideal solution and the anti-ideal solution, respectively. Indicates the first The ratio of the utility of each risk assessment indicator to the ideal solution. Indicates the first The ratio of the utility of a risk assessment indicator to the anti-ideal solution.

[0026] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention innovatively employs the Brown-Wood Method (BWM) to assign weights based on experts' objective qualifications, while simultaneously introducing the Anti-Entropy Weighted Method (AEWM) to extract the information entropy of the expert scoring matrix and calculate expert opinion bias. This mechanism can adaptively identify and significantly reduce invalid scores in the evaluation process, ensuring that the final decision-making power leans towards authoritative experts, thus guaranteeing the objectivity of risk assessment from the outset.

[0027] 2. Addressing the limitations of expert knowledge in complex water conservancy projects, this invention abandons the crude real-number averaging method, which easily leads to information homogenization. Relying on the mathematical properties of the In-Interval Intelligence Set (INS), the "true membership degree (certainty of occurrence)," "hesitation degree (uncertainty)," and "false membership degree (certainty of non-occurrence)" from multiple expert opinions are independently and completely integrated in the INWA operator using the expert's comprehensive weight as the exponent. This preserves the uncertainty characteristics of the decision-making process to the greatest extent and achieves a high degree of integration of group opinions.

[0028] 3. This invention improves upon traditional evaluation models by establishing dual benchmarks of absolute safety (ID) and accident limit (AID) based on MARCOS logic. Through a nonlinear comprehensive utility function, the model can achieve an "exponential amplification" measurement effect on critical risk factors. This method overcomes the shortcomings of traditional evaluation methods, such as low indicator discrimination, susceptibility to inversion, and lack of distinction between primary and secondary factors. It achieves accurate ranking of core risk indicators at water conservancy construction sites, providing reliable quantitative decision support for the precise allocation of disaster prevention resources on site. Attached Figure Description

[0029] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0030] Figure 1 A flowchart illustrating a method for assessing the construction safety risks of water conservancy projects based on an improved INS-MARCOS, provided for an embodiment of the present invention; Figure 2 This is a structural diagram of the water conservancy project construction safety risk assessment index system provided in an embodiment of the present invention; Detailed Implementation To make the technical means, creative features, achieved objectives, and effects of this invention readily understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0031] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Please refer to... Figure 1 As shown, this invention proposes an improved INS-MARCOS-based method for assessing the construction safety risks of water conservancy projects. The specific steps are as follows: S1. Construct a risk assessment index system for construction safety of water conservancy projects; For the characteristics of water conservancy project construction, see Figure 2 This embodiment is based on the traditional Based on the existing framework of the (Construction Human Factors Analysis and Classification System), an evaluation index system covering four dimensions was constructed: Impact of corporate organization: Examine the construction unit's organizational structure, investment in safety production, and the completeness of its safety management procedures; Safety supervision: Examine the implementation of on-site risk monitoring and early warning mechanisms and the effectiveness of hazard investigation; Factors related to on-site operations: The key considerations are the combined effects of complex geological conditions (such as faults, mudslides, and water inrushes), the condition of construction machinery, and the natural climate environment. Factors related to construction personnel: Examine the construction personnel's violations of regulations, habitual skill errors, and risk perception capabilities.

[0032] The four primary indicators are further subdivided into multiple secondary indicators, forming a complete tree-structured evaluation model that provides a comprehensive logical basis for the quantitative assessment of risk.

[0033] Primary indicators include organizational impact. Safety supervision Factors related to on-site operations Factors related to construction personnel ; The secondary indicator is specifically: organizational impact of enterprises. Including organizational structure and responsibilities Investment in safe production Safety Management Procedures Safety supervision Including risk monitoring and early warning Supervision and management violations Work plan arrangement On-site operation related factors Including technical measures Materials and Machinery Geological conditions Working environment Changes in weather conditions Factors related to construction personnel Including operational violations Skill errors Intuition and Decision Errors Personnel quality .

[0034] S2. Obtain the subjective and objective weights of experts, and integrate the subjective and objective weights to obtain the comprehensive weight; To address the shortcomings of single weighting methods that cannot simultaneously consider both experiential knowledge and data fluctuations, this invention employs an improved weighting mechanism that combines subjective and objective factors. The expert weights are combined and assigned.

[0035] based on Subjective empowerment of law: Based on the experts' qualifications, the most qualified experts were selected. The worst-qualified expert A preference vector is constructed using the 1-9 scaling method: the preference vector of the optimal expert relative to other experts. ; Preference vector of other experts relative to the worst expert .

[0036] To obtain the weights, a linearly constrained optimization model is constructed to minimize the maximum absolute deviation:

[0037]

[0038] in: The number of participating experts, Top-ranked experts Compared to the first The relative importance preference scale value of each expert. For the first The expert compared to the least qualified expert The relative importance preference scale value, The first one to be solved The subjective seniority weight of each expert. The most qualified expert to be solved Subjective seniority weighting The least qualified expert to be solved Subjective seniority weighting; The largest absolute error variable used to measure the degree of consistency of the meta-decision-maker preference scaling; Solve the model to obtain the expert subjective weight vector. After the model is solved, the consistency ratio needs to be calculated. Only when If the preference logic is valid, then the preference scale needs to be readjusted until the consistency requirement is met.

[0039] Based on the anti-entropy weight method ( Expert objective weighting adjustment: The construction site of the water conservancy project has a total of There are 10 risk assessment indicators, and a total of 100 experts participated in the evaluation. First extract the bits. Experts The original semantic comments given by each indicator are transformed into interval intelligence numbers using a preset mapping rule, and then substituted into the interval intelligence scoring function. Perform deblurring:

[0040] in :and Let these represent the lower and upper bounds of the proper membership function of the numbers in the interval, respectively. and Let these represent the lower and upper bounds of the hesitant membership function of the intelligence number in the interval, respectively. and Let represent the lower and upper bounds of the pseudo-membership function of the numbers in the interval, respectively.

[0041] Therefore, a risk indicator-based approach is constructed. Evaluation experts included Real evaluation matrix .in, Indicates the first The expert commented on the first The precise quantitative score of each risk indicator.

[0042] No. The expert's rating was the highest. and The first in its scoring sequence Normalized values ​​of each indicator The calculation formula is:

[0043] Calculate probability proportion :

[0044] Calculate information entropy :

[0045] Calculate objective weights :

[0046] Using inverse entropy To measure the value of information, the higher the value, the higher the weight.

[0047] Combined weights Calculated using the linear weighting method:

[0048] in: , and These represent the expert subjective weight and expert objective weight allocation coefficients, respectively, with a value range of [value range missing]. , The first one calculated by the BWM method The subjective weight of each expert The first value obtained by the anti-entropy weight method The objective weights of each expert are used to adjust the proportion of subjective qualifications and objective scoring quality in the final weighting.

[0049] In this embodiment, through Solving the optimization model using this method to obtain expert subjective weights can effectively reduce inconsistencies caused by personal preferences among water conservancy experts. Subsequently, we introduce... The core significance of this method lies in the fact that its objective data is directly derived from the expert's interval-based semantic evaluation matrix. Under complex working conditions in water conservancy projects, if an expert's evaluation sequence tends to be homogeneous due to cognitive limitations or avoidance of responsibility, the information entropy of their evaluation sequence will significantly increase. This study extracts... The objective data of each column vector in the matrix are used to calculate the posterior objective weights of the experts using the inverse entropy weighting method. This mechanism can automatically identify invalid scores with low discriminative power and high hesitation, effectively filtering out redundant and poor-quality information in group decision-making.

[0050] S3. Construct the initial interval intelligent decision matrix; To address the issue of expert hesitation during assessments due to insufficient geological data in deeply buried underground engineering projects, this invention abandons traditional deterministic scoring and introduces a pre-defined 7-level semantic set (e.g., from "extremely low risk" to "extremely high risk") and maps it to an interval-based intelligent number. .

[0051] The membership interval that represents the certainty of the existence of risk; The wider the uncertainty range introduced for innovation, the more ambiguous the experts' understanding of the indicator and the stronger their hesitation. The membership interval represents the degree of confidence that the risk does not exist.

[0052] In the formula middle: A number represents the intelligence within an interval and is the basic unit for evaluating information. (Truth-membership) represents the degree to which the assessor is certain that the risk factor exists. Indeterminacy-membership represents the degree of hesitation caused by incomplete information or limitations in expert knowledge. F (Falsity-membership) represents the degree to which the assessor is certain that the risk factor does not exist. L and U represent the lower bound and upper bound of the corresponding component intervals, respectively. and Let represent the lower and upper bounds of the proper membership function of the intelligence numbers in the interval, respectively, used to characterize the reliability of the evaluation information. and Let represent the lower and upper bounds of the hesitant membership function of the intelligence number in the interval, respectively, used to characterize the degree of uncertainty or fuzziness of the evaluation information. and Let represent the lower and upper bounds of the pseudo-membership function of the intelligence numbers in the interval, respectively, used to characterize the degree of unreliability of the evaluation information. And satisfy: ,and .

[0053] S4. Construct the intelligent decision-making matrix in the extended interval; Based on the initial interval intelligence decision matrix, this embodiment anchors the absolute benchmark: directly extracting the corresponding extremely low-risk interval intelligence numbers from the semantic database as the ideal solution. Extract the values ​​corresponding to extremely high risks as the anti-ideal solution. .Will and Incorporate the initial interval intelligent decision matrix to form the extended interval intelligent decision matrix. This provides a solid benchmark for subsequent evaluation.

[0054] S5. Integration of expert group decision-making information; Using the interval weighted average The operator, using the expert comprehensive weights obtained in step S2 as an index, integrates the evaluation information of multiple experts into a comprehensive evaluation value for each risk assessment indicator. ; The formula for calculating the interval weighted operator is:

[0055] in: m The number of risk assessment indicators, This indicates the number of the risk assessment indicator. ; n The number of experts participating in the evaluation. Indicates the evaluation expert's number ,in n This represents the total number of experts who participated in the evaluation. Indicates the first The comprehensive combined weighting of the evaluation experts, and They represent the first The expert commented on the first The lower and upper bounds of the true membership function for evaluating each indicator. and They represent the first The expert commented on the first The lower and upper bounds of the hesitant membership function for evaluating individual indicators. and They represent the first The expert commented on the first The lower and upper bounds of the pseudo-membership function for evaluating each indicator.

[0056] The physical significance of the interval-based intelligent weighted operator lies in: The calculation logic for the components is inherently flawed; if any one key indicator carries an extremely high risk, the overall risk value will rapidly approach a high level. The component can preserve the product of the expert's degree of hesitation in each dimension without loss.

[0057] For example, in water conservancy projects such as the construction of diversion tunnels, because geological surveys cannot completely cover all rock strata details, experts often have a mindset of "it may be medium risk, but it may also be higher" when giving assessments. This invention utilizes... Independent components capture this hesitation, in During integration, this uncertain information is losslessly incorporated into the final calculation, ensuring the accurate restoration of the safety margin.

[0058] S6. Calculate the degree of utility; The comprehensive evaluation values ​​of each risk assessment indicator are defuzzified using the interval-based intelligent scoring function to obtain the final comprehensive score of each risk assessment indicator. The utility ratio of the final comprehensive score relative to the ideal solution score and the anti-ideal solution score is calculated respectively. Comprehensive scoring function The calculation formula is:

[0059] in: Indicates the first The absolute comprehensive score is obtained after defuzzifying each risk assessment indicator using a scoring function. and They represent the first The lower and upper limits of the comprehensive true membership degree after aggregating various risk assessment indicators. and They represent the first The lower and upper limits of the overall hesitation membership degree after aggregating various risk assessment indicators. and They represent the first The lower and upper limits of the composite pseudo-membership degree after aggregating several risk assessment indicators; This formula is applicable to high uncertainties. and high pseudo-membership Deductions are made to ensure that the final score objectively reflects the severity of the risk.

[0060] S7. Calculate the overall utility score; The utility ratios are calculated by integrating the nonlinear utility function to obtain the comprehensive utility score of each index relative to the ideal solution and the irrational solution. Calculate the deviation / approximation factor of the evaluated scheme relative to the absolute safety benchmark and the absolute hazard benchmark:

[0061] Subsequently, the sub-utility function of a single dimension is calculated using compromise utility logic. .

[0062] Finally, the overall utility score, which integrates both sense of location and sense of distance, is derived:

[0063] In the formula, , and These are the individual utility function values ​​of the solution relative to the ideal solution and the anti-ideal solution, respectively. Indicates the first i The ratio of the utility of each risk assessment indicator to the ideal solution. Indicates the first i The ratio of the utility of a risk assessment indicator to the anti-ideal solution.

[0064] exist In the operators, scoring functions, and utility formulas: It is the first The comprehensive intelligence assessment value of the interval after weighted integration of the indicators. It is the first The final combined weight of the experts. The score function value is used to convert multidimensional interval numbers into comparable real-valued scalars. The ideal solution and the anti-ideal solution are respectively used as the upper and lower limits for evaluation. It is an indicator The utility ratio (ratio value) relative to the ideal solution and the antiideal solution, respectively. The final comprehensive utility function value is the ultimate quantitative basis for determining the risk level.

[0065] Unlike simple ranking methods (such as traditional methods) ), Model calculation and It's a comparison between indicators and standards. This means... The results have absolute reference value: when When the risk level is below the set threshold, even if the indicator poses the greatest risk among multiple indicators, this provides a scientific basis for the red line management of water conservancy construction risks.

[0066] S8. Risk Assessment and Output: Based on the comprehensive utility score, the risk level of each risk assessment indicator is ranked in descending order. Overall utility score Analysis is performed. This score ranges from 0 to 1; the higher the value, the closer the construction system is to the ideal safety state. The overall effectiveness score... The score is analyzed. It ranges from 0 to 1; a higher value indicates the construction system is closer to the ideal safety state. If the calculated score for a certain working condition is... If the value drops significantly (e.g., falls below the medium-risk threshold of 0.6), project managers can assess the situation based on various indicators. The intermediate scores before operator integration are used to trace back to the "weak links" that lead to increased risks (such as the discovery of a sharp increase in the objective weight of geological condition indicators or a widening of the expert hesitation range leading to deductions), thereby enabling precise engineering remedial measures to be taken and achieving a logical closed loop from risk assessment to precise prevention and control.

[0067] Implementation Case: To verify an improvement proposed in this invention To assess the scientific validity and effectiveness of the risk assessment method for water conservancy project construction safety, this embodiment selects a large-scale water conservancy project under construction in southwestern China as an application case. Located in a deep mountain canyon area with complex geological structures, frequent mudslides, water inrushes, and rock bursts, and situated during the rainy season, the risk factors at the construction site exhibit a high degree of uncertainty and ambiguity.

[0068] This embodiment aims to accurately identify the most critical "core risk sources" in the current construction phase by ranking the 15 secondary safety risk indicators constructed above. To this end, the project team assembled an evaluation panel consisting of four experts with different levels of experience: experts (Chief Project Engineer, 20 years of experience) Expert (Senior Engineer, 10 years of experience), Expert (Supervising Engineer, 5 years of experience) and experts (On-site safety officer, 1 year of experience).

[0069] The 15 secondary indicators are: organizational impact of enterprises Including organizational structure and responsibilities Investment in safe production Safety Management Procedures Safety supervision Including risk monitoring and early warning Supervision and management violations Work plan arrangement On-site operation related factors Including technical measures Materials and Machinery Geological conditions Working environment Changes in weather conditions Factors related to construction personnel Including operational violations Skill errors Intuition and Decision Errors Personnel quality .

[0070] Based on the on-site investigation, the four experts adopted a pre-set 7-level semantic set. The degree of danger of 15 risk indicators was independently scored, as shown in Table 1; the evaluation set and interval intelligence data were used. The mapping relationship is shown in Table 2.

[0071] Table 1. Interval-based semantic scoring table

[0072] Table 2. Semantic Evaluation of Construction Risks in Water Conservancy Projects and Interval Mapping Table of Intelligence Numbers

[0073] Table 3 Interval Intelligence Representation and Matrix of Evaluation Information

[0074] To ensure the scientific nature of group decision-making, this embodiment adopts... The combined discourse power of experts in the subjective-objective coupling model calculation.

[0075] The method empowers core technical personnel with a dominant voice by assessing their objective resumes, including professional background, years of experience, and historical decision-making accuracy.

[0076] In this embodiment, the decision-maker needs to evaluate the qualifications and decision-making authority of the four experts in order to assign them corresponding subjective weights.

[0077] Selecting the best expert :expert (Project Chief Engineer, with the most senior experience).

[0078] Selecting the worst expert :expert (On-site safety officer, the least experienced).

[0079] The meta-decision-maker uses the 1-9 scale to identify the optimal expert. Importance preference vector relative to other experts And other experts relative to the worst expert Importance Preference Vector :

[0080]

[0081] Linear optimization model formula:

[0082]

[0083] The specific values ​​in Tables 1 and 2 and expert number Substituting each of the above constraints, we obtain the fully expanded mathematical model: Objective function:

[0084] Optimal preference absolute value constraint (to Replace with 1, 2, 4, 8): for

[0085] for

[0086] for

[0087] for

[0088] The absolute value constraint of worst preference (will) Replace with 8, 4, 2, 1): for

[0089] for

[0090] for

[0091] for

[0092] Normalization and nonnegativity constraints:

[0093]

[0094] Substituting the above equation into the normalization condition (the sum of the weights is 1):

[0095]

[0096]

[0097] Solving for the optimal expert The weights are:

[0098] Then, substituting this back into the previous proportional relationship, we can obtain the weights of the other experts:

[0099]

[0100]

[0101] According to the BWM method's inspection criteria, the consistency ratio must be calculated. To demonstrate the scientific validity of the decision. Formula: ; In this example, the maximum preference scaling Consulting the BWM standard conformance index database reveals that the corresponding scale is 8. Value is

[0102] The obtained consistency error

[0103] therefore,

[0104] because This fully demonstrates that the decision-makers' choice of this set of preference scales (1, 2, 4, 8) has extremely high logical consistency, and the resulting expert subjective weight data is valid.

[0105] AEWM's expert objective empowerment: To accurately capture hesitation, the interval-based defuzzification of semantic evaluation does not employ a coarse linear real number mapping (such as 1-6 points). Instead, it rigorously substitutes semantic variables into the interval-based intelligence scoring function. Perform precise conversion:

[0106] The exact scalar score corresponding to level 7 semantics is calculated as follows:

[0107]

[0108] Based on this set of precise scores, the extreme values ​​and ranges of the scoring sequences for 15 risk indicators by four experts were extracted: expert Scoring distribution Range = 0.300 expert Scoring distribution Range = 0.550 expert Scoring distribution

[0109] expert Scoring distribution

[0110] Data standardization Probability weight The range method was used to perform internal normalization on the scores of each expert (based on the expert's score). (Using an example to illustrate the calculation). Experts Six out of the 15 indicators were given. 7 2 .

[0111] Normalized value:

[0112] Normalized value:

[0113] Normalized value:

[0114] Normalized sum of columns:

[0115] Calculate the probability weight:

[0116] Calculate the information entropy of each expert With inverse entropy According to the information entropy formula ; Computational experts Information entropy:

[0117]

[0118] Similarly, the information entropy and inverse entropy values ​​of all experts were calculated. ): (The scoring is clear and highly differentiated) (The scores are highly similar) (Very low information content)

[0119]

[0120] Generate AEWM objective weights

[0121] Perform overall normalization on the anti-entropy value. :

[0122]

[0123]

[0124]

[0125] Expert portfolio weighting: To balance subjective engineering experience with objective data quality, an allocation coefficient is introduced.

[0126] Perform linear weighted fusion

[0127] The final results are shown in Table 4: Table 4 Summary of Calculation Results of Subjective and Objective Combination Weights

[0128] Using the interval weighted average The operator, after obtaining the combined weights of four experts... Subsequently, multi-source information aggregation and dual-benchmark measurement were performed on 15 water conservancy project construction safety risk indicators, and core risk sources were accurately identified based on the ranking results.

[0129] Due to the high degree of ambiguity and uncertainty inherent in the risks at water conservancy construction sites, the judgments of a single expert are often hesitant. This study employs an interval-based weighted average operator. Based on the comprehensive weight of experts The index aggregates independent assessments of the same risk indicator from multiple experts into a comprehensive range. .

[0130] Let a certain risk indicator The intelligence assessment value from the four experts is:

[0131] Then its comprehensive evaluation value The calculation formula is:

[0132] Indicators For example: Based on the initial evaluation matrix, the four experts... The semantic comments are respectively .

[0133] The corresponding intervals contain the following intelligence numbers:

[0134]

[0135]

[0136]

[0137] Known to have passed The calculated combined weights of the four experts are as follows:

[0138] Substitute the above data rigorously into the formula for calculating the six components of the Interval Intelligent Weighted Average (INWA) operator: Calculate the true membership interval :

[0139]

[0140]

[0141]

[0142] Calculate the hesitation interval :

[0143]

[0144]

[0145]

[0146] (Note: To avoid the base being 0, which would render the expression meaningless, when...) It usually takes the minimum value

[0147] In this example (Minimum is 0.10) Calculate the pseudo-membership interval :

[0148]

[0149] The final comprehensive assessment value for China's intelligence range: Based on the above INWA multi-source information aggregation, The interval intelligence number after incorporating the weights of 4 experts is:

[0150] Defuzzing calculation of the scoring function: Will Substitute into the scoring function The deblurring process was performed, and the results are shown in Table 5:

[0151]

[0152]

[0153] Table 5. Risk Indicators of INWA Operator: Comprehensive Range Assessment Matrix and Scoring Table

[0154] In the MARCOS model, an ideal solution needs to be constructed. With anti-ideal solution For this application scenario, the following definitions apply: Absolutely safe ideal solution This indicates that the project is in a completely risk-free state, and is set as the lowest risk level in the semantic set. .

[0155] Accident Limit Anti-ideal Solution This represents the extreme state where a catastrophic accident could occur at any time in the project, and is set as the highest risk level in the semantic set. .

[0156] Substituting this into the scoring function, the dual-benchmark score is obtained as follows:

[0157]

[0158] Calculate the utility ratio of each risk indicator. : Calculate the overall score of each risk indicator Compared to and The utility ratio. The ratio represents the relative degree to which the risk source is "absolutely safe" and "absolutely dangerous":

[0159]

[0160] Calculate the comprehensive utility function : Comprehensive utility function The risk level that can nonlinearly amplify risk indicators approaching the accident limit is calculated using the following formula:

[0161] The higher the value, the closer the current state is to the accident limit after considering the opinions of multiple experts, and the higher the degree of danger.

[0162] Indicators and For example, the utility ratio calculation: After completing the INWA multi-source information aggregation and scoring function defuzzification, the absolute comprehensive scores of two typical risk indicators were extracted. :

[0163]

[0164] The utility ratio is calculated using the formula from the MARCOS model: calculate Utility ratio:

[0165] calculate Overall utility value:

[0166] calculate Utility ratio:

[0167] calculate Overall utility value:

[0168] Substituting all 15 risk indicators into the INS-MARCOS model for iterative calculations, the final utility ratio and overall utility measurement results were obtained. Based on... The risk indicators are sorted in descending order, as detailed in Table 6.

[0169] Precisely pinpointing the core risk sources: Calculation results indicate that geological conditions Operational violations Overall utility value The scores of 6.8368 and 5.9232, significantly higher than other indicators, were accurately identified as the critical core risks under the current working conditions. This conclusion is highly consistent with the objective engineering realities of large-scale water conservancy projects, namely, the frequent occurrence of geological disasters in deeply buried tunnels and the difficulty in supervising illegal operations on the front line.

[0170] In summary, compared with traditional methods for evaluating the safety of water conservancy project construction, the "method for evaluating the safety risks of water conservancy project construction based on improved INS-MARCOS" proposed in this invention exhibits the following significant technical advantages: A subjective and objective expert weighting mechanism was constructed to prevent manipulation, thus avoiding the drawbacks of directly assigning weights to experts in traditional evaluations. The objective qualifications of experts are determined through Black-White Weighted Analysis (BWM), while AEWM (Anti-Entropy Weighted Analysis) is introduced to verify data quality. This mechanism can adaptively identify and significantly reduce perfunctory and undifferentiated evaluations that "all assign medium risk" in group decision-making, ensuring that the final decision-making power is tilted towards experts who are both experienced and conscientious.

[0171] This invention achieves lossy aggregation of multi-source fuzzy information. Addressing the limitations of expert cognition under complex operating conditions in water conservancy projects, this invention does not employ simple real-number averaging, but instead relies on the mathematical properties of interval intelligence sets (INS) to aggregate the "certainty" in expert opinions. "Hesitation level ()" ")" and "degree of negation ()" The INWA operator performs independent and complete 6-dimensional space integration. This preserves the uncertainty characteristics of the decision-making process to the greatest extent and avoids the early loss of key risk information.

[0172] An improved MARCOS model was introduced, establishing absolute safety. With the limits of accidents Dual benchmarks. Through a nonlinear comprehensive utility function. The model can produce an "exponential amplification" measurement effect on risk factors that are hidden but extremely fatal in the eyes of many experts, overcoming the "reverse order" defect common in traditional TOPSIS or VIKOR methods, and achieving accurate labeling of core risks at water conservancy construction sites.

[0173] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of the invention. Any equivalent substitutions or partial modifications made by those skilled in the art under the technical concept and logical principles of the present invention, such as parallel application of this evaluation system to risk source identification in other complex infrastructure fields such as bridge engineering and underground utility tunnel engineering, should still be included within the patent protection scope of the present invention.

[0174] Table 6. Ranking of Risk Indicators Based on Multi-Expert Group Decision-Making

Claims

1. A hydraulic engineering construction safety risk assessment method based on an improved INS-MARCOS, characterized in that, Includes the following steps: Step S1: Construct a risk assessment index system for water conservancy project construction safety; Step S2: Obtain the expert subjective weights and expert objective weights, and merge the subjective weights and objective weights to obtain the comprehensive weight; Step S3: Collect raw assessment data of each risk assessment indicator from multiple experts, and use the preset semantic set mapping model to convert the experts' qualitative comments into interval intelligence numbers to construct the initial interval intelligence decision matrix. Step S4: Define the ideal solution and the anti-ideal solution, and incorporate them into the intelligent decision-making matrix in the initial interval to form the intelligent decision-making matrix in the extended interval; Step S5: Using the interval-based weighted average operator, with the comprehensive weight as the index, integrate the evaluation information of multiple experts into a comprehensive evaluation value for each risk assessment indicator; Step S6: Use the interval intelligent scoring function to defuzzify the comprehensive evaluation value of each risk assessment indicator to obtain the final comprehensive score of each risk assessment indicator, and calculate the utility ratio of the final comprehensive score relative to the ideal solution score and the anti-ideal solution score respectively. Step S7: Use a nonlinear utility function to perform a fusion calculation on the utility degree ratios to obtain the comprehensive utility score of each index relative to the ideal solution and the irrational solution; Step S8: Sort the risk levels of each risk assessment indicator in descending order based on the comprehensive utility score to obtain the risk assessment result.

2. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 1, characterized in that, In step S1, the evaluation indicators in the water conservancy project construction safety risk assessment index system include: The primary indicators include the impact of corporate organization, safety supervision, on-site operation-related factors, and construction personnel-related factors; The secondary indicators are as follows: Enterprise organizational impact includes organizational structure and responsibilities, safety production investment, and safety management procedures; safety supervision includes risk monitoring and early warning, supervision and management violations, and work plan arrangements; on-site operation related factors include technical measures, materials and machinery, geological conditions, work environment, and weather conditions; and construction personnel related factors include operational violations, skill errors, intuition and decision-making errors, and personnel quality.

3. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 1, characterized in that, In step S2, the subjective weight is calculated by a method, and the specific process includes: determining the best and worst indicators, selecting the best and worst experts from the group of experts according to their qualifications and worst indicators ; Constructing optimal comparison vectors : optimal expert : score relative to other experts : score relative to other experts : score relative to other experts ; constructing a worst expert comparison vector : other experts relative to the worst expert are scored to obtain: where ; Solving linear optimization model to obtain optimal subjective weights The BWM linear optimization model is established as follows to minimize the deviation minimize: wherein: is the number of experts participating in the evaluation, is the most senior expert, is the relative importance preference scale value of the first expert relative to the second expert, is the relative importance preference scale value of the second expert relative to the most junior expert, is the relative importance preference scale value of the first expert relative to the most junior expert, is the subjective seniority weight of the first expert to be solved, is the subjective seniority weight of the most senior expert to be solved, is the subjective seniority weight of the most junior expert to be solved; is the maximum absolute error variable for measuring the consistency degree of the preference scale of the meta-decision maker; The consistency check calculates a consistency ratio wherein is the consistency index corresponding to the scale; When the logic consistency is judged to be satisfied.

4. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 1, characterized in that, In step S2, the objective weights are calculated using the inverse entropy weight method, and the specific process includes: extracting bit experts The original semantic comments given by the experts are converted into interval-valued numbers by using the preset mapping rules, and then substituted into the interval-valued score function de-buzzing in: and Let these represent the lower and upper bounds of the proper membership function of the numbers in the interval, respectively. and Let these represent the lower and upper bounds of the hesitant membership function of the intelligence number in the interval, respectively. and Let represent the lower and upper bounds of the pseudo-membership function of the intelligence numbers in the interval, respectively; A real matrix of evaluations is constructed with risk indicators as rows and evaluation experts as columns wherein represents the precise quantified score of the th expert for the th risk indicator . The score of the first bit expert is the maximum value and the normalized value of the first indicator in the score sequence of the expert is , and the calculation formula is: Computing a probability weight : Computing information entropy : Computing the objective weight : Adopting the anti-entropy value to measure the information value, the greater the value, the higher the weight.

5. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 1, characterized in that, In step S2, the combined weight Using linear weighting method: Wherein: ; is the subjective weight distribution coefficient of the expert, the value range is 0 <1; is the objective weight distribution coefficient of the expert, the value range is 0< <1 ; is the subjective weight of the first expert calculated by the BWM method, is the objective weight of the first expert calculated by the inverse entropy weight method.

6. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 1, characterized in that, In step S3, the three independent components of the intelligent number in the interval are respectively: the true membership interval representing the determination degree of the existence of risk , the uncertain membership interval representing the degree of hesitation caused by cognitive limitations , and the false membership interval representing the determination degree of the non-existence of risk ; wherein: and respectively represent the lower bound and the upper bound of the true membership function of the interval-valued intuitionistic number, for representing the confidence degree of the evaluation information; and respectively represent the lower bound and the upper bound of the hesitancy membership function of the interval-valued intuitionistic number, for representing the uncertainty degree or the fuzzy degree of the evaluation information; and respectively represent the lower bound and the upper bound of the false membership function of the interval-valued intuitionistic number, for representing the unconfidence degree of the evaluation information; and satisfy: , and .

7. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 6, characterized in that, In step S4, the ideal solution is the number of the interval of the minimum value in the semantic set corresponding to the ideal value of the risk-type indicator, and the anti-ideal solution is the number of the interval of the interval of the maximum value in the semantic set corresponding to the anti-ideal value of the risk-type indicator.

8. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 7, characterized in that, In step S5, the calculation formula for the intelligent weighting operator in the interval is: wherein: m the number of risk evaluation indexes, denotes the number of risk evaluation indexes, ; n the number of experts participating in the evaluation, denotes the number of evaluation experts wherein n the total number of experts participating in the evaluation, denotes the comprehensive combination weight of the th evaluation expert, and respectively denote the lower limit and the upper limit of the true membership function of the th expert for the th index evaluation, and respectively denote the lower limit and the upper limit of the hesitant membership function of the th expert for the th index evaluation, and respectively denote the lower limit and the upper limit of the false membership function of the th expert for the th index evaluation.

9. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 8, characterized in that, In step S6, the risk evaluation index is de-fuzzified to obtain an absolute comprehensive score function ) : in: Indicates the first The absolute comprehensive score is obtained after defuzzifying each risk assessment indicator using a scoring function. and They represent the first The lower and upper limits of the comprehensive true membership degree after aggregating various risk assessment indicators. and They represent the first The lower and upper limits of the overall hesitation membership degree after aggregating various risk assessment indicators. and They represent the first The lower and upper limits of the composite pseudo-membership degree after aggregating various risk assessment indicators.

10. The improved INS-MARCOS-based safety risk assessment method for hydraulic engineering construction according to claim 9, characterized in that, In step S7, the formula for calculating the comprehensive utility value is: In the formula, , and These are the individual utility function values ​​of the solution relative to the ideal solution and the anti-ideal solution, respectively. Indicates the first The ratio of the utility of each risk assessment indicator to the ideal solution. Indicates the first The ratio of the utility of a risk assessment indicator to the anti-ideal solution.

Images