Subway shield construction collapse accident risk analysis method and system

Abstract

The invention provides a subway shield construction collapse accident risk analysis method and system, belongs to the technical field of collapse accident risk analysis, and aims to solve the problem of lack of accident statistical analysis for subway shield construction collapse accidents in the prior art. Comprising the steps of 1, establishing a subway shield construction collapse accident risk assessment index system; 2, constructing an initial fuzzy BNs structure diagram; 3, optimizing the initial fuzzy BNs structure chart in the step 2 to obtain an optimized fuzzy BNs structure chart; 4, determining a root node prior probability based on an expert evaluation method, aggregating opinions of experts based on a similarity aggregation method to obtain an optimized root node prior probability, and re-optimizing the optimized root node prior probability to obtain a final risk factor prior probability; step 5, obtaining the conditional probability of the intermediate node based on a Noisy-Or gate model; and step 6, obtaining a shield construction collapse accident risk probability based on the final risk factor prior probability and the conditional probability.

Description

Technical Field

[0001] The invention belongs to the technical field of collapse accident risk analysis, and in particular relates to a method and system for analyzing the risk of collapse accidents in subway shield construction. Background Art

[0002] With the rapid advancement of domestic urbanization and the rapid development of the economy, the population of major cities is increasing rapidly, and the problem of ground traffic congestion is becoming increasingly serious. Traditional public transportation can no longer meet the needs of modern urban development. As an efficient and convenient means of transportation in modern cities, the subway has the advantages of strong transportation capacity, high punctuality, high speed, high comfort, high safety, full use of ground and underground space, and less pollution. It has become the first choice for solving traffic congestion problems in densely populated cities. According to data from the Ministry of Transport, as of December 31, 2022, 53 cities in my country have opened 290 rail transit lines with an operating mileage of 9,584 kilometers.

[0003] In recent years, subway construction technology has been continuously developed and updated. However, subway construction often faces harsh geological conditions and complex construction sites, and subway construction accidents occur from time to time. Domestic and foreign scholars have achieved certain results in subway construction risk management theory, risk identification, analysis and evaluation methods, and management countermeasures. However, from an engineering perspective, there are still some shortcomings:

[0004] There is currently a lack of statistical analysis on subway shield construction collapse accidents, which means that this type of related research lacks data support and makes it difficult to conduct accurate risk management for specific accident types. Summary of the invention

[0005] In view of this, the present invention provides a risk analysis method and system for subway shield construction collapse accidents to solve the problem in the prior art that there is a lack of accident statistical analysis for subway shield construction collapse accidents, which makes such related research lack of data support and difficult to conduct accurate risk management for specific accident types.

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

[0007] A method for analyzing the risk of collapse accidents in subway shield construction, comprising:

[0008] Step 1: Establish a subway shield construction collapse accident risk assessment index system, the accident risk assessment index system includes a target layer, a criterion layer and an indicator layer, wherein the target layer is the probability of occurrence of accident risk factors, the criterion layer is the factors that reflect the target layer, and the indicator layer is the accident risk factors;

[0009] Step 2: construct an initial fuzzy BNs structure diagram, in which the target layer subway shield construction collapse risk is taken as a leaf node of the fuzzy BNs structure, the criterion layer is taken as an intermediate node of the fuzzy BNs structure, and the indicator layer is taken as a root node of the fuzzy BNs;

[0010] Step 3: Optimize the initial fuzzy BNs structure diagram in step 2 by adding auxiliary nodes to obtain the optimal fuzzy BNs structure diagram;

[0011] Step 4 (there is no order between step 4 and step 5): Determine the prior probability of the root node based on the expert evaluation method, aggregate the opinions of the experts based on the similarity aggregation method to obtain the optimized prior probability of the root node, and re-optimize the optimized prior probability of the root node based on the subjective and objective weighting method to obtain the final prior probability of the risk factor;

[0012] In step 4, the determination of the prior probability of the root node based on the expert evaluation method specifically includes:

[0013] Step A1: Collect the evaluations of the probability of occurrence of risk factors by various experts. i Experts on risk factors i Give natural language evaluation;

[0014] Step A2: Convert the natural language evaluation into triangular fuzzy numbers P ij =(a, b, c).

[0015] In step 4, the optimized root node prior probability obtained by aggregating the opinions of the experts based on the similarity aggregation method specifically includes:

[0016] Step B1: Calculate the O of two different experts i (i=1, 2,...θ) and O j (j=1, 2, ...θ) opinion P i =(a i , b i , c i ) and P j =(a j , b j , c j ) similarity S(P i , P j ), as shown below:

[0017]

[0018] Where: θ is the total number of experts; S(P i , P j )∈[0,1] is a similarity function, P i With P jExpert O i With O j The opinions are triangular fuzzy numbers:

[0019] Step B2: Calculate the average agreement measure AA (P i ), as shown below:

[0020]

[0021] Step B3: Calculate the relative consistency measure RA (P i ), as shown below:

[0022]

[0023] Step B4: Calculate the expert consistency coefficient CC (P i ), as shown below:

[0024] CC(P i )=β·ω(O i )+(1-β)·RA(P i )

[0025] Where: ω(O i ) is the expert weight, here is the normalized weight, β is the relaxation factor, showing the expert weight ω(O i ) relative to RA(P i ) importance. When β = 0, it means that the expert's weight is not considered. When β = 1, it means that the expert consistency measure is the same as the expert weight. The recommended value of β is 0.5. In this paper, β = 0.5 is used to calculate the expert's consistency coefficient.

[0026] Step B5: Aggregate the expert opinions to obtain the fuzzy aggregate value P AG , as shown below:

[0027] P AG =CC(O1)·P1+CC(O2)·P2+…+CC(O θ )·P θ

[0028] Step B6: Perform triangular fuzzy averaging on the opinion aggregation results to obtain the optimized root node prior probability P′ j , as shown below:

[0029]

[0030] In step B4, the step of re-optimizing the optimized root node prior probability based on the subjective and objective weighting method to obtain the final risk factor prior probability specifically includes the following steps:

[0031] The P′ that has been aggregated j The final risk factor prior probability is calculated with the specified ratio of the probability of occurrence of the risk factor in step 1, with μ taking the value of 0.7 and γ taking the value of 0.3, as shown in the following formula:

[0032]

[0033] Step 5: Obtain the conditional probability of the intermediate nodes based on the Noisy-Or gate model;

[0034] In step 5, the conditional probability of the intermediate node is obtained based on the Noisy-Or gate model and calculated by the following formula:

[0035]

[0036] Step 6: Obtain the risk probability of collapse accident in shield construction based on the prior probability and conditional probability of the final risk factor.

[0037] A subway shield construction collapse accident risk analysis system, comprising:

[0038] Index system construction module: Establish an index system for risk assessment of subway shield construction collapse accidents. The accident risk assessment index system includes a target layer, a criterion layer and an index layer. The target layer is the probability of occurrence of accident risk factors, the criterion layer is the factors that reflect the target layer, and the index layer is the accident risk factors.

[0039] BNs structure diagram construction module: construct an initial fuzzy BNs structure diagram, in which the target layer subway shield construction collapse risk is taken as a leaf node of the fuzzy BNs structure, the criterion layer is taken as an intermediate node of the fuzzy BNs structure, and the indicator layer is taken as a root node of the fuzzy BNs;

[0040] Optimization module: Optimize the initial fuzzy BNs structure diagram in step 2 by adding auxiliary nodes to obtain the optimal fuzzy BNs structure diagram;

[0041] Prior probability acquisition module: Determine the prior probability of the root node based on the expert evaluation method, aggregate the opinions of experts based on the similarity aggregation method to obtain the optimized prior probability of the root node, and re-optimize the optimized prior probability of the root node based on the subjective and objective weighting method to obtain the final prior probability of the risk factor;

[0042] Conditional probability acquisition module: obtains the conditional probability of the intermediate nodes based on the Noisy-Or gate model;

[0043] Accident risk probability acquisition module: The risk probability of shield construction collapse accident is obtained based on the prior probability and conditional probability of the final risk factor.

[0044] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0045] 1. This paper aims to understand the current research status of BNs, fuzzy theory and subway shield construction technology development at home and abroad, and to identify the risk factors that may lead to collapse accidents in subway construction, laying a foundation for subsequent research.

[0046] 2. The present invention obtains fuzzy BNs prior probability and risk factor loss comment set through questionnaire survey, and deeply understands the experts' cognition and understanding of risk factors and potential risks, thereby improving the accuracy and reliability of the model.

[0047] 3. The present invention combines the risk factors of subway shield construction collapse accidents to carry out qualitative identification, quantitative analysis and evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The present invention will now be described by way of example with reference to the accompanying drawings, in which:

[0049] Figure 1 It is a risk indicator hierarchy diagram of the present invention;

[0050] Figure 2 This is a risk index system diagram of subway shield collapse accidents of the present invention;

[0051] Figure 3 A flow chart for the risk analysis study of the present invention;

[0052] Figure 4 It is the initial fuzzy BNs structure diagram of the present invention;

[0053] Figure 5 This is a fuzzy BNs structure diagram after adding auxiliary nodes in the present invention;

[0054] Figure 6 It is the expert natural language fuzzy number graph of the present invention;

[0055] Figure 7 It is the Genie fuzzy BNs causal reasoning graph of the present invention;

[0056] Figure 8 is the fuzzy BNs diagnostic reasoning graph of the present invention;

[0057] Fig. 9 It is the diagnostic reasoning analysis diagram of the present invention;

[0058] Fig.10 It is the posterior probability change diagram of the intermediate nodes and leaf nodes of the present invention;

[0059] Fig.11 This is a sensitivity analysis diagram of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0061] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0063] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0064] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0065] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0066] Example 1

[0067] like Figure 3 As shown, an embodiment of the present invention discloses a method for analyzing the risk of collapse accidents in subway shield construction, including:

[0068] Step 1: Establish a subway shield construction collapse accident risk assessment index system, which includes a target layer, a criterion layer and an indicator layer, wherein the target layer is the probability of occurrence of accident risk factors, the criterion layer is the factor that reflects the target layer, and the indicator layer is the accident risk factor; specifically, Figure 1 Referring to relevant literature and the standards issued by the corresponding countries and the hierarchical relationship between the objectives and indicators of risk assessment, under the guidance of the principles of risk assessment indicator system construction, the main risk factors were finally selected from five aspects, including human factors, mechanical factors, material factors, environmental factors, and construction technology factors, to form the risk assessment indicator system for subway shield construction collapse accidents. Figure 2 shown.

[0069] Depend on Figure 2 It can be seen that due to the different functions of workers and managers, the identification of personnel risk factors is mainly carried out from these two aspects. Mechanical risk factors are mainly identified from the shield machine itself and other supporting facilities. Material factors are mainly identified based on the quality of materials used in shield construction. Environmental risk factors mainly include construction geology, hydrogeology and natural factors. Construction technology risk factors are mainly identified based on shield construction technology.

[0070] Step 2: Construct an initial fuzzy BNs structure diagram, in which the target layer subway shield construction collapse risk is taken as a leaf node of the fuzzy BNs structure, the criterion layer is taken as an intermediate node of the fuzzy BNs structure, and the indicator layer is taken as a root node of the fuzzy BNs; specifically, it includes:

[0071] The main subway shield construction collapse accident risk factors determined by the construction management method are used as the root nodes of the fuzzy BNs. The criterion layer in the index system is the middle layer of the fuzzy BNs structure, and the target layer subway shield construction collapse risk is the leaf node of the fuzzy BNs structure. Each node state is described by State0 and State1. State0 indicates that the node situation occurs, and State1 indicates that the node situation does not occur. According to the risk assessment system determined in Section 3.3, the fuzzy BNs node factors are determined, as shown in Table 1.

[0072] Table 1 Fuzzy BNs node description table

[0073]

[0074] The determined subway shield construction risk index system is transformed into the initial fuzzy BNs network structure, such as Figure 4 shown.

[0075] Step 3: Optimize the initial fuzzy BNs structure diagram in step 2 by adding auxiliary nodes to obtain the optimal fuzzy BNs structure diagram; specifically:

[0076] Depend on Figure 3 It can be seen that when constructing a fuzzy BNs network structure, if the intermediate nodes contain too many root nodes, the amount of calculation may be too large. For example, for node H, there are 6 root nodes and each root node has 2 states, then the conditional probability distribution table of node H needs to determine 2 6 = 64 parameters, which will result in too many parameters to be estimated, which is prone to errors and will affect the accuracy of the estimation work. In order to simplify the calculation, the method of adding auxiliary nodes can be adopted

[119] For example, if two auxiliary nodes are added to node H, only 16 parameters need to be calculated, that is, 2 3 +2 3 This will greatly reduce the number of parameters that need to be calculated, thereby improving calculation efficiency and evaluation accuracy. The specific implementation of the structure diagram after adding auxiliary nodes is as follows Figure 5 shown.

[0077] In the case of a large number of nodes, the fuzzy BNs model needs to calculate enough parameters for causal and diagnostic reasoning, which requires the use of corresponding software. Currently, BNs software includes Netica, Genie, Matlab toolkit, BayesianLab, etc.

[0078] Step 4: Determine the prior probability of the root node based on the expert evaluation method, aggregate the opinions of each expert based on the similarity aggregation method to obtain the optimized prior probability of the root node, and re-optimize the optimized prior probability of the root node based on the subjective and objective weighting method to obtain the final prior probability of the risk factor;

[0079] In step 4, seven natural language comments of experts are introduced into the expert evaluation method, namely, very low (VL), low (L), low (ML), medium (M), high (MH), high (H) and very high (VH). The membership functions and triangular fuzzy numbers corresponding to these language variables are as follows: Figure 6 And as shown in Table 2.

[0080] Table 2 Triangular fuzzy number table corresponding to expert natural language

[0081]

[0082] The determination of the prior probability of the root node based on the expert evaluation method specifically includes:

[0083] Step A1: Collect the evaluations of the probability of occurrence of risk factors by various experts. i Experts on risk factorsi Give natural language evaluation;

[0084] Step A2: Convert the natural language evaluation into triangular fuzzy numbers P ij =(a, b, c).

[0085] Since different experts have different perceptions of the same risk factors, there are different opinions, which leads to inconsistency in the collected triangular fuzzy numbers. In order to integrate the opinions of experts and reduce the influence of the subjectivity of a single expert, this paper uses similarity aggregation method to process the fuzzy opinions of experts in natural language.

[0086] In step 4, the optimized root node prior probability obtained by aggregating the opinions of the experts based on the similarity aggregation method specifically includes:

[0087] Step B1: Calculate the O of two different experts i (i=1, 2,...θ) and O j (j=1, 2, ...θ) opinion P i =(a i , b i , c i ) and P j =(a j , b j , c j ) similarity S(P i , P j ), as shown below:

[0088]

[0089] Where: θ is the total number of experts; S(P i , P j )∈[0,1] is a similarity function, P i With P j Expert O i With O j The opinion is a triangular fuzzy number;

[0090] Step B2: Calculate the average agreement measure AA (P i ), as shown below:

[0091]

[0092] Step B3: Calculate the relative consistency measure RA (P i ), as shown below:

[0093]

[0094] Step B4: Calculate the expert consistency coefficient CC (Pi ), as shown below:

[0095] CC(P i )=β·ω(O i )+(1-β)·RA(P i )

[0096] Where: ω(O i ) is the expert weight, here is the normalized weight, β is the relaxation factor, showing the expert weight ω(O i ) relative to RA(P i ), when β = 0, it means that the expert's weight is not considered, and when β = 1, it means that the expert consistency measure is the same as the expert weight. The recommended value of β is 0.5. In this paper, β = 0.5 is taken to calculate the expert's consistency coefficient. The expert weight score table is shown in Table 3.

[0097] Table 3 Expert weight score table

[0098]

[0099] Step B5: Aggregate the expert opinions to obtain the fuzzy aggregate value P AG , as shown below:

[0100] P AG =CC(O1)·P1+CC(O2)·P2+…+CC(O θ )·P θ

[0101] Step B6: Perform triangular fuzzy averaging on the opinion aggregation results to obtain the optimized root node prior probability P′j ， As shown below:

[0102]

[0103] In step B4, the step of re-optimizing the optimized root node prior probability based on the subjective and objective weighting method to obtain the final risk factor prior probability specifically includes the following steps:

[0104] The P′ that has been aggregated j The final risk factor prior probability is calculated with the specified ratio of the probability of occurrence of the risk factor in step 1, with μ taking the value of 0.7 and γ taking the value of 0.3, as shown in the following formula:

[0105] P″=μP′ j +γP.

[0106] Step 5: Obtain the conditional probability of the intermediate node based on the Noisy-Or gate model; the intermediate node usually has many parent nodes, and even if the parent node has only two states, the intermediate node requires 2n conditional probability values ​​to model the model. When the Noisy-Or gate model is introduced, the 2n conditional probabilities can be reduced to 2n, which greatly reduces the amount of calculation and increases the operability of the model, improving the accuracy of the model. The Noisy-OR gate model is mainly used to describe the intrinsic relationship between multiple variables and their related child nodes. This model can be used when there are only two states of the fuzzy BNs node;

[0107] In step 5, the conditional probability of the intermediate node is obtained based on the Noisy-Or gate model and calculated by the following formula:

[0108]

[0109] Step 6: Based on the prior probability and conditional probability of the final risk factor, the risk probability of collapse accident in shield construction is obtained, including:

[0110] Establish the probability and level standard of risk factors to more intuitively determine the possibility of the risk factor. By sorting out and analyzing the historical construction risk situation and combining the relevant national standards for the industry, the risk probability level is divided into five intervals, as shown.

[0111] Table 4 Probability level correspondence table

[0112]

[0113] Among them, levels 1 to 5 represent that risk factors may occur frequently, may occur, occasionally occur, rarely occur, and are unlikely to occur.

[0114] Example 2

[0115] This embodiment proposes a subway shield construction collapse accident risk analysis system based on the embodiment 1, including:

[0116] Index system construction module: Establish an index system for risk assessment of subway shield construction collapse accidents. The accident risk assessment index system includes a target layer, a criterion layer and an index layer. The target layer is the probability of occurrence of accident risk factors, the criterion layer is the factors that reflect the target layer, and the index layer is the accident risk factors.

[0117] BNs structure diagram construction module: construct an initial fuzzy BNs structure diagram, in which the target layer subway shield construction collapse risk is taken as a leaf node of the fuzzy BNs structure, the criterion layer is taken as an intermediate node of the fuzzy BNs structure, and the indicator layer is taken as a root node of the fuzzy BNs;

[0118] Optimization module: Optimize the initial fuzzy BNs structure diagram in step 2 by adding auxiliary nodes to obtain the optimal fuzzy BNs structure diagram;

[0119] Prior probability acquisition module: Determine the prior probability of the root node based on the expert evaluation method, aggregate the opinions of experts based on the similarity aggregation method to obtain the optimized prior probability of the root node, and re-optimize the optimized prior probability of the root node based on the subjective and objective weighting method to obtain the final prior probability of the risk factor;

[0120] Conditional probability acquisition module: obtains the conditional probability of the intermediate nodes based on the Noisy-Or gate model;

[0121] Accident risk probability acquisition module: The risk probability of shield construction collapse accident is obtained based on the prior probability and conditional probability of the final risk factor.

[0122] Example 3

[0123] This embodiment proposes a specific implementation case of embodiment 1, and verifies it by taking the section between Fengtan Road Station and Wenhua Road Station of Hangzhou Metro Line 2 as an example; the mileage range between Fengtan Road Station and Wenhua Road Station is SDK30+181.036 (XDK30+181.036) to SDK31+740.300 (XDK31+740.300). The short chain of the up line is 0.739m, with a total length of 1558.525m; the long chain of the down line is 12.729m, with a total length of 1571.993m; the interval between the lines is 13.4 to 15.6m. The specific details of the project are shown in Table 5.

[0124] Table 5 Project Introduction

[0125]

[0126] 1. Get the root node parameters

[0127] 1) Expert evaluation method

[0128] Based on the comprehensive coverage of the risk factors considered above, combined with the actual situation of the Fengtan Road Station to Wenhua Road Station section, the constructed fuzzy BNs structure is used to determine the node parameters. This paper distributed questionnaires to 15 relevant industry experts, university teachers, and front-line construction personnel, and finally collected 12 questionnaires, 10 valid questionnaires, and selected the 6 most objective questionnaire results to obtain the possible comments on risk factors. The weight coefficients of the 6 experts were determined according to the expert scoring criteria in Table 3 as shown in Table 6.

[0129] Table 6 Expert weight table

[0130]

[0131] According to Table 2, the expert evaluation results are subjected to triangular fuzzy processing, and the results are shown in Table 7.

[0132] Table 7 Correspondence between expert natural language variables and triangular fuzzy numbers

[0133]

[0134]

[0135]

[0136] 2) Similarity aggregation method calculation

[0137] Based on the results in the above table, taking the root node H1 as an example, similarity aggregation calculation is performed. The specific process is as follows:

[0138] (1) According to the following formula, the similarity between the first and second expert opinions is calculated:

[0139]

[0140] The similarity calculation of other expert opinions is performed as above, and the similarity of the evaluations of the six experts is shown in Table 8.

[0141] Table 8 Expert opinion similarity table

[0142]

[0143] (2) Calculate the average consistency measure of expert opinions according to the following formula:

[0144]

[0145] The average consistency measure calculation of other expert opinions is performed as above, and the average consistency measure of the 6 experts is shown in Table 9.

[0146] Table 9 Experts’ average consistency measure

[0147]

[0148] (3) Calculate the relative consistency measure of each expert according to the following formula:

[0149]

[0150] The relative consistency measures of other experts’ opinions were calculated by referring to the above, and the relative consistency measures of the six experts are shown in Table 10.

[0151] Table 10 Expert relative consistency measurement table

[0152]

[0153] (4) Calculate the expert consistency coefficient CC(P i ):

[0154] CC(P1)=β·ω(O1)+(1-β)·RA(P1)

[0155] =0.5*0.203+0.5*0.181

[0156] =0.192

[0157] The relative consistency coefficients of other expert opinions were calculated by referring to the above, and the consistency coefficients of the six experts are shown in Table 11.

[0158] Table 11 Expert relative consistency coefficient table

[0159]

[0160] (5) The aggregation result of fuzzy opinions is obtained according to the following formula:

[0161] P AG =CC(O1)·P1+CC(O2)·P2+…+CC(O θ )·P θ

[0162] =0.192·P1+0.181·P2+0.177·P3+0.17·P4+0.147·P5+0.133·P6

[0163] =(0.428,0.628,0.797)

[0164] (6) The aggregated triangular fuzzy numbers are homogenized according to the following formula:

[0165]

[0166] 3) Subjective and objective empowerment method

[0167] Combining expert opinions with actual data according to the following formula, we get the final prior probability of H1 risk factor:

[0168] P=μP j γP

[0169] =0.7×0.62+0.3×0.084

[0170] =0.459

[0171] According to the calculation process of H1, the prior probabilities of all root nodes are obtained as shown in Table 12.

[0172] Table 12 Root node prior probability table

[0173]

[0174] 2. Determination of intermediate node parameters

[0175] Taking auxiliary node X1 as an example, node conditional probability calculation is performed. As shown in Table 13:

[0176] Table 13 X1 node conditional probability table

[0177]

[0178] This conditional probability is input into the Genie software, and the probability result of the X1 node is 0.386. Similarly, the conditional probability table of all intermediate nodes is obtained and input into the Genie software.

[0179] Instance-based Fuzzy BNs Reasoning

[0180] 1. Causal Reasoning

[0181] By inputting the root node and intermediate node parameters into the Genie software, the risk probability of collapse accident in shield construction of this project can be obtained, such as Figure 7 shown.

[0182] Depend on Figure 7 It can be seen that the probability of human factors, mechanical factors, environmental factors, material factors, and construction technology factors are 26%, 27%, 31%, 13%, and 29% respectively. The probability of the risk of collapse accident in shield construction of this project is 15%.

[0183] 2. Diagnostic Reasoning

[0184] Diagnostic reasoning is the process of solving the cause when the result is known. When the probability of a subway shield collapse accident is set to 100%, that is, when the state of the T node State0 is 1, the posterior probability calculation results of each node can be obtained, which indicates the probability of each risk factor appearing when the accident occurs. The calculation process at this time is a posterior estimation, and the calculation results are shown in Figure 8 shown.

[0185] Depend on Figure 8 It can be seen that environmental risk factors have the largest posterior probability, and when an accident occurs, it is usually accompanied by environmental risk factors.

[0186] The root node posterior probability and the rate of change compared to the prior probability are shown in Table 14 and Fig. 9 shown.

[0187] Table 14 Diagnostic reasoning results

[0188]

[0189]

[0190] From the above table 14, Fig. 9 It can be seen that the root node under environmental factors has the largest change rate. The risk factor with the largest change rate is groundwater abundance E4, with a change rate of 35.8%. The change rates of other environmental risk factors are all over 25%. In the actual construction process, environmental risk factors often become the cause of accidents, and corresponding measures must be taken.

[0191] Among the material factors, the risk factor with the largest change rate is poor segment quality C1, with a change rate of 12.07%. The posterior probability change rates of the remaining material factors also exceed 10%.

[0192] There was no significant change in the posterior probabilities of personnel factors, construction technology factors, and mechanical factors. The risk factor with the largest change rate was the inadequate risk identification in the early stage of construction, with a change rate of 4.86%.

[0193] The posterior probability of the intermediate nodes and leaf nodes changes as follows Fig.10 shown.

[0194] Depend on Fig.10 It can be seen that the posterior probability of environmental factors changes significantly, with a change rate of 135.67%, from the prior probability of 0.314 to 0.74, and from the risk possibility level 4 to level 3. This shows that when an accident occurs, it is usually accompanied by the emergence of environmental risk factors.

[0195] 3. Sensitivity analysis

[0196] Sensitivity analysis is a technique that can help verify the probability parameters of BNs. This is done by studying the impact of small changes in the model's numerical parameters (i.e., prior probabilities and conditional probabilities) on the output parameters (e.g., posterior probabilities). Highly sensitive parameters have a greater impact on the inference results. That is, small probability changes in certain risk factors may cause the occurrence of subway shield collapse accidents, so effective measures need to be taken against these risk factors in order to effectively reduce the possibility of accidents.

[0197] With the help of Genie software, the accident node T is set as the target node. The result is as follows: Fig.11 shown.

[0198] Nodes with higher sensitivity have been distinguished by color depth, where the darker the color, the higher the sensitivity of the node. From the figure, we can see that the factors with higher sensitivity are environmental factors and material factors, which is basically consistent with the change rate of the posterior probability. The specific sensitivity of the root node is shown in Table 15.

[0199] Table 15 Sensitivity analysis table

[0200]

[0201]

[0202] According to the above table, the sensitivity ranking results are similar to the diagnostic reasoning results. The risk factors under environmental risk factors are ranked high, among which the highest-ranked risk factor is the groundwater abundance risk factor. In the actual construction process, this risk needs to be fully explored. The risk factors under material risk factors are also ranked high, among which the poor quality of the pipe segment has the highest sensitivity. In the shield construction process, the pipe segment is an indispensable construction material, and its quality needs to be strictly controlled and the corresponding work needs to be done well.

[0203] Example risk analysis results

[0204] According to the node prior probability and combined with the risk possibility level evaluation table in Table 4, the risk factor possibility evaluation results are shown in Table 16:

[0205] Table 16 Risk probability estimation table

[0206]

[0207] The material risk factors and the collapse risk level of the shield construction of this project are level 5, which is "impossible to occur", and the levels of other risk factors are level 4, which is "rare", which is consistent with the actual project situation.

[0208] The circuits, electronic components and modules involved are all prior art and can be fully implemented by those skilled in the art. Needless to say, the content protected by the present invention does not involve improvements to software and methods.

[0209] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0210] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for analyzing the risk of collapse accidents in subway shield construction, characterized in that: include: Step 1: Establish a subway shield construction collapse accident risk assessment index system, the accident risk assessment index system includes a target layer, a criterion layer and an indicator layer, wherein the target layer is the probability of occurrence of accident risk factors, the criterion layer is the factors that reflect the target layer, and the indicator layer is the accident risk factors; Step 2: construct an initial fuzzy BNs structure diagram, in which the target layer subway shield construction collapse risk is taken as a leaf node of the fuzzy BNs structure, the criterion layer is taken as an intermediate node of the fuzzy BNs structure, and the indicator layer is taken as a root node of the fuzzy BNs; Step 3: Optimize the initial fuzzy BNs structure diagram in step 2 by adding auxiliary nodes to obtain the optimal fuzzy BNs structure diagram; Step 4: Determine the root node prior probability based on the expert evaluation method, aggregate the opinions of each expert based on the similarity aggregation method to obtain the optimized root node prior probability, and re-optimize the optimized root node prior probability based on the subjective and objective weighting method to obtain the final risk factor prior probability; Step 5: Obtain the conditional probability of the intermediate nodes based on the Noisy-Orgate model; Step 6: Obtain the risk probability of collapse accident in shield construction based on the prior probability and conditional probability of the final risk factor.

2. A subway shield construction collapse accident risk analysis method according to claim 1, characterized in that: In step 4, determining the prior probability of the root node based on the expert evaluation method specifically includes: Step A1: Collect the evaluations of the probability of occurrence of risk factors by various experts, among which the first i Experts on risk factors i Give natural language evaluation; Step A2: Convert the natural language evaluation into triangular fuzzy numbers P ij =(a, b, c), where a is the minimum possible boundary, b is the maximum possible boundary, and c is the maximum possible boundary. ij Indicates the Oth i Risk factors given by experts i The triangular fuzzy number of the occurrence probability.

3. The method for analyzing the risk of collapse accidents in subway shield construction according to claim 1 is characterized in that: In step 4, the optimized root node prior probability obtained by aggregating the opinions of the experts based on the similarity aggregation method specifically includes: Step B1: Calculate the O of two different experts i and O j Opinion i =(a i , b i , c i ) and P j =(a j , b j , c j ) similarity S(P i , P j ), as shown below: Where: S(P i , P j )∈[0,1] is a similarity function, P i With P j Expert O i With O j The opinion is a triangular fuzzy number; Step B2: Calculate the average agreement measure AA (P i ), as shown below: Step B3: Calculate the relative consistency measure RA (P i ), as shown below: Step B4: Calculate the expert consistency coefficient CC (P i ), as shown below: CC(P i )=β·ω(O i )+(1-β)·RA(P i ) Where: ω(O i ) is the expert weight, here is the normalized weight, β is the relaxation factor, showing the expert weight ω(O i ) relative to RA(P i ), when β = 0, it means that the expert's weight is not taken into account, and when β = 1, it means that the expert consistency measure is the same as the expert weight; Step B5: Based on steps B1-B4, aggregate the expert opinions to obtain the fuzzy aggregate value P AG , as shown below: P AG =CC(O1)·P1+CC(O2)·P2+…+CC(O θ )·P θ Among them, CC(O1) represents the consistency coefficient of experts’ evaluation on the first influencing factor; Step B6: Perform triangular fuzzy averaging on the opinion aggregation results to obtain the optimized root node prior probability P' j , as shown below:

4. A subway shield construction collapse accident risk analysis method according to claim 3, characterized in that: In step B4, the step of re-optimizing the optimized root node prior probability based on the subjective and objective weighting method to obtain the final risk factor prior probability specifically includes the following steps: The P' j The final risk factor prior probability is calculated with the specified ratio of the probability of occurrence of the risk factor in step 1, with μ taking the value of 0.7 and γ taking the value of 0.3, as shown in the following formula: P”=μP' j +γP。 5. The method for analyzing the risk of collapse accidents in subway shield construction according to claim 1 is characterized in that: In step 5, the conditional probability of the intermediate node is obtained based on the Noisy-Or gate model and calculated by the following formula: Among them, P i represents the connection probability of all parent nodes of node Y, P i (Y) represents the conditional probability of node Y.

6. A subway shield construction collapse accident risk analysis system, used to implement a subway shield construction collapse accident risk analysis method according to claims 1-5, characterized in that: include: Index system construction module: Establish an index system for risk assessment of subway shield construction collapse accidents. The accident risk assessment index system includes a target layer, a criterion layer and an index layer. The target layer is the probability of occurrence of accident risk factors, the criterion layer is the factors that reflect the target layer, and the index layer is the accident risk factors. BNs structure diagram construction module: construct an initial fuzzy BNs structure diagram, in which the target layer subway shield construction collapse risk is taken as a leaf node of the fuzzy BNs structure, the criterion layer is taken as an intermediate node of the fuzzy BNs structure, and the indicator layer is taken as a root node of the fuzzy BNs; Optimization module: Optimize the initial fuzzy BNs structure diagram in step 2 by adding auxiliary nodes to obtain the optimal fuzzy BNs structure diagram; Prior probability acquisition module: Determine the prior probability of the root node based on the expert evaluation method, aggregate the opinions of experts based on the similarity aggregation method to obtain the optimized prior probability of the root node, and re-optimize the optimized prior probability of the root node based on the subjective and objective weighting method to obtain the final prior probability of the risk factor; Conditional probability acquisition module: obtains the conditional probability of the intermediate nodes based on the Noisy-Or gate model; Accident risk probability acquisition module: The risk probability of shield construction collapse accident is obtained based on the prior probability and conditional probability of the final risk factor.

Images