Dynamic evaluation method for safety risk of maintenance construction of village-town crossing road

Abstract

The invention discloses a village-town crossing road maintenance construction safety risk dynamic evaluation method, and belongs to the field of artificial intelligence modeling and road engineering safety, and the method comprises the steps: building a risk index system comprising multi-level indexes; based on the risk indicator system, a dynamic Bayesian network model is constructed, and the model comprises a root node and a conditional dependency relationship between the nodes; determining the fuzzy probability of each root node in the dynamic Bayesian network model by adopting a fuzzy set theory in combination with an expert scoring method; the conditional probability of a non-root node in the dynamic Bayesian network model is corrected on the basis of a Leak Noise-or Gate expansion model; forward reasoning is carried out through a dynamic Bayesian network model, and a dynamic probability curve of the town-crossing highway maintenance construction safety risk is generated; backward reasoning is carried out through a dynamic Bayesian network model, and the posterior probability of the key risk factors is calculated. According to the method, the safety in the village-town-penetrating maintenance construction process can be guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence modeling and road engineering safety technology, and in particular relates to a dynamic evaluation method for safety risks in the maintenance and construction of rural highways. Background Art

[0002] As important transportation arteries connecting towns and villages, the safety and efficiency of rural roads have long attracted research attention. Since Pates first defined rural roads from a functional perspective and analyzed their safety issues in 1998, related research has gradually expanded to multi-dimensional risk analysis. In 2012, Zhang Tiejun further explored the risk composition of rural roads from a road space perspective, providing a theoretical framework for subsequent research. Ji Xiaofeng and others focused on two-lane rural roads in mountainous areas, identifying the impact of terrain and traffic flow on accident risk. Song Zhenglin constructed a traffic safety risk assessment model for national highways through rural areas based on indicators such as people, vehicles, roads, and the environment, combined with a gray cloud model. In addition, for highway construction safety, scholars have proposed methods such as improved accident tree models, hierarchical analysis-extension models, game theory-TOPSIS hybrid evaluation models, and indicator system methods, providing technical support for the quantification and control of construction risks.

[0003] While existing research has made some progress in safety analysis during the operational phase of rural highways, its limitations are significant. First, existing methods are mostly based on static perspectives (e.g., accident trees and indicator systems), which make it difficult to capture the dynamic evolution of risk factors during construction, resulting in assessment results lagging behind actual risk changes. Second, there is a lack of research on maintenance and construction scenarios for rural highways. Existing models do not fully consider the specific characteristics of these scenarios (e.g., road occupation, complex alignments, pedestrian interference, etc.), and are unable to accurately identify key risk factors in dynamic environments. Furthermore, traditional methods are unable to adequately handle fuzzy and uncertain factors (e.g., expert experience and unknown risks), which restricts the reliability and applicability of risk assessments. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes a dynamic evaluation method for safety risks of maintenance construction of rural highways to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a method for dynamically evaluating safety risks in the maintenance and construction of rural highways, comprising:

[0006] According to the characteristics of maintenance and construction of highways through villages and towns and the theory of accident causes, a risk indicator system with multiple levels of indicators is established;

[0007] Based on the risk indicator system, a dynamic Bayesian network model is constructed, wherein the model includes a root node and conditional dependencies between nodes; fuzzy set theory is combined with an expert scoring method to determine the fuzzy probability of each root node in the dynamic Bayesian network model; and the conditional probabilities of non-root nodes in the dynamic Bayesian network model are corrected based on the Leak Noisy-or Gate extended model;

[0008] The steps to build a dynamic Bayesian network model include:

[0009] Generate a network structure based on the causal relationship between the indicators in the risk indicator system;

[0010] Based on the network structure, determining the node transition probability between adjacent time slices by using a transition probability formula;

[0011] The dynamic Bayesian network model is used for forward reasoning to generate a dynamic probability curve of the safety risk of maintenance and construction of rural roads; the dynamic Bayesian network model is used for reverse reasoning to calculate the posterior probability of key risk factors.

[0012] Preferably, the steps of establishing a risk indicator system comprising multiple levels of indicators include:

[0013] Risk factors are divided into six primary indicators: drivers, engineers, vehicles, roads, environment, construction areas and facilities;

[0014] For each first-level indicator, set the corresponding second-level indicator.

[0015] Preferably, the secondary indicators of the driver index include driving age, gender, speeding and fatigue driving;

[0016] The secondary indicators of the engineering personnel indicators include age, health status, educational level, fatigue, safety awareness, safety clothing and professional technical capabilities.

[0017] Preferably, the transition probability formula is:

[0018]

[0019] in, represents the value of the i-th variable at time t, for The parent node of , N represents the number of variables.

[0020] Preferably, the step of determining the fuzzy probability of each root node in the dynamic Bayesian network model includes:

[0021] Convert the experts' linguistic evaluation of nodes into triangular fuzzy numbers;

[0022] The mean area method is used to convert the triangular fuzzy number into a probability value.

[0023] Preferably, the membership function of the triangular fuzzy number is:

[0024]

[0025] Among them, a, m, and b represent the three values ​​of the triangular fuzzy number interval; express The degree of membership in .

[0026] Preferably, the step of modifying the conditional probability of the non-root node comprises:

[0027] Introducing the Leak Noisy-or Gate extended model, defining the uncovered risk factors as unknown nodes and adding them to the dynamic Bayesian network model;

[0028] Based on the unknown node and the known parent node, the conditional probability of node Y is calculated using the conditional probability formula;

[0029] Preferably, the conditional probability formula is:

[0030] P(Y|X i )=P i +P all -P i P all ;

[0031]

[0032]

[0033] Among them, Xi is a known parent node, To eliminate the unknown factor X l All factors except X all For other parent nodes, P i For X i The corresponding connection probability, P all For X all The corresponding connection probability.

[0034] Preferably, the steps of reverse reasoning include:

[0035] When the probability of a security risk reaching a preset threshold, the posterior probability of each node is calculated;

[0036] Calculate the proportional change value of each node according to the proportional change formula;

[0037] Ranking the risk factors by sensitivity based on the proportional change values, and marking nodes with proportional change values ​​greater than a preset threshold as key risk factors;

[0038] Key risk factors include speeding, vehicle overload, road speed limits, construction area maintenance and safety facilities, weather, age of construction workers, driver fatigue, worker fatigue, traffic volume, and driver experience.

[0039] Wherein, the ratio change formula is:

[0040]

[0041] Among them, V RO (C i ) is node C i ROV,π(C i ) is the posterior probability, θ(C i ) is the prior probability.

[0042] In a second aspect, the present invention further discloses a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described in the first aspect when executed by a processor.

[0043] Compared with the prior art, the present invention has the following advantages and technical effects:

[0044] The present invention provides a dynamic evaluation method for the safety risk of maintenance and construction of roads through villages and towns, comprising: first, establishing a risk indicator system including multi-level indicators based on the characteristics of maintenance and construction of roads through villages and towns and accident causation theory; second, constructing a dynamic Bayesian network model based on the risk indicator system, wherein the model includes root nodes and conditional dependencies between nodes; adopting fuzzy set theory combined with an expert scoring method to determine the fuzzy probability of each root node in the dynamic Bayesian network model; correcting the conditional probability of non-root nodes in the dynamic Bayesian network model based on a Leak Noisy-or Gate extended model; finally, performing forward reasoning through the dynamic Bayesian network model to generate a dynamic probability curve for the safety risk of maintenance and construction of roads through villages and towns; and performing reverse reasoning through the dynamic Bayesian network model to calculate the posterior probability of key risk factors.

[0045] In response to the technical problem that "traditional static methods (such as accident trees and indicator systems) are unable to track the dynamic evolution of risks", the present invention constructs a dynamic Bayesian network model and introduces a time slice transfer probability formula, which can reflect the dynamic changes of risk factors in the construction process in real time (forward reasoning generates safety risk probability curves under different time slices), thereby providing real-time early warning and decision support.

[0046] To address the problem of "traditional methods' insufficient ability to process subjective experience and fuzzy information", the present invention adopts fuzzy set theory combined with expert scoring method to convert expert language evaluation into triangular fuzzy numbers, thereby improving the reliability and adaptability of model parameters.

[0047] In response to the defect of "lack of adaptability of existing models to scenarios", a risk indicator system was established specifically for the particularities of maintenance and construction of roads through villages and towns (such as road occupation, complex alignment, pedestrian interference, etc.), accurately covering the core risk sources in the construction of roads through villages and towns.

[0048] To address the problem of "evaluation bias caused by ignoring unknown factors in traditional models", the present invention introduces the LeakNoisy-or Gate extended model, defines uncovered risk factors as unknown nodes and integrates them into the dynamic Bayesian network, thereby enhancing the model's adaptability to complex construction environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0050] Figure 1 Schematic diagram of the safety risk analysis process for maintenance and construction of rural highways based on dynamic Bayesian analysis according to an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of the security risk structure of a Dynamic Bayesian Network (DBN) according to an embodiment of the present invention;

[0052] Figure 3 This is a dynamic change diagram of the probability of safety risks occurring during maintenance and construction of rural roads according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0054] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0055] Example 1

[0056] The safety risks of road maintenance and construction through rural towns are dynamic, changeable, and complex. They are affected by multiple factors such as traffic flow, rural environment, construction conditions, and operators, making accident risks difficult to predict. DBN can dynamically evaluate and predict these changing risks by integrating real-time data and multi-dimensional factors, providing real-time warnings and decision support, thereby improving the efficiency and adaptability of construction safety management. Therefore, DBN is used to conduct dynamic risk analysis of road maintenance and construction through rural towns. The main steps are as follows: Figure 1 This embodiment provides a method for dynamically evaluating the safety risks of road maintenance and construction through villages and towns, including:

[0057] S1. Based on the characteristics of maintenance and construction of roads through villages and towns and the theory of accident causes, a risk indicator system with multiple levels of indicators is established;

[0058] Furthermore, the steps of establishing a risk indicator system including multi-level indicators include:

[0059] Risk factors are divided into six primary indicators: drivers, engineers, vehicles, roads, environment, construction areas and facilities;

[0060] For each first-level indicator, set the corresponding second-level indicator.

[0061] Furthermore, the secondary indicators of the driver index include driving age, gender, speeding and fatigue driving;

[0062] The secondary indicators of the engineering personnel indicators include age, health status, educational level, fatigue, safety awareness, safety clothing and professional technical capabilities.

[0063] Specifically, this embodiment collects accident data, analyzes safety accident cases during maintenance and construction of rural highways, and integrates relevant technical specifications for rural highway maintenance and construction safety. It categorizes safety risk factors for rural highway maintenance and construction into drivers, engineers, vehicles, roads, environments, and construction areas and facilities. Based on expert opinion, this embodiment identifies the safety risk factors for rural highway maintenance and construction. The safety risk assessment indicator system for rural highway maintenance and construction is shown in Table 1.

[0064] Table 1

[0065]

[0066] S2. Based on the risk indicator system, construct a dynamic Bayesian network model, wherein the model includes a root node and conditional dependencies between nodes;

[0067] Furthermore, the steps of constructing a dynamic Bayesian network model include:

[0068] Generate a network structure based on the causal relationship between the indicators in the risk indicator system;

[0069] Based on the network structure, the node transition probability between adjacent time slices is determined by a transition probability formula.

[0070] Specifically, the transition probability formula between two adjacent time slices in a dynamic Bayesian network (DBN) is:

[0071]

[0072] Where, represents the value of the i-th variable at time t, for The parent node of , N represents the number of variables.

[0073] Based on the conditional probability of the initial network and the transition probability of the transition network, the joint probability of any node in the DBN can be calculated as follows:

[0074]

[0075] The construction principle of the dynamic Bayesian network model is based on probability theory and graph theory. By representing the conditional dependency relationship between variables, a dynamic model is established to reflect the changes in risk factors over time. It is mainly divided into two parts: determining the network structure and determining the network parameters.

[0076] Based on the index system in Table 1 and combined with the characteristics of maintenance and construction of rural highways, the possible correlation between factors is taken into account, and the expert experience is converted into the causal relationship between risk factors, and a reasonable safety risk DBN structure is constructed, such as Figure 2 shown.

[0077] Furthermore, the fuzzy set theory is combined with the expert scoring method to determine the fuzzy probability of each root node in the dynamic Bayesian network model;

[0078] Furthermore, the step of determining the fuzzy probability of each root node in the dynamic Bayesian network model includes:

[0079] Convert the experts' linguistic evaluation of nodes into triangular fuzzy numbers;

[0080] The mean area method is used to convert the triangular fuzzy number into a probability value.

[0081] Specifically, fuzzy set theory (FST) is often used to deal with uncertainty problems caused by fuzziness. In this embodiment, the node fuzzy probability of the DBN model adopts triangular fuzzy numbers. The membership function is shown in the following formula:

[0082]

[0083] In the formula, a, m, and b represent the three values ​​of the triangular fuzzy number interval respectively; the membership function express The membership degree in , x∈[0,1].

[0084] Furthermore, the conditional probabilities of non-root nodes in the dynamic Bayesian network model are modified based on the Leak Noisy-or Gate extended model;

[0085] Furthermore, the step of correcting the conditional probability of the non-root node includes:

[0086] Introducing the Leak Noisy-or Gate extended model, defining the uncovered risk factors as unknown nodes and adding them to the dynamic Bayesian network model;

[0087] Based on the unknown node and the known parent node, the conditional probability of node Y is calculated using the conditional probability formula;

[0088] Specifically, in practical applications, there are many factors that influence risk events, and building a DBN model may not fully cover all risk factors. This means that even if all risk factors in the model do not occur, a risk event may still occur. To this end, this embodiment uses the LeakNoisy-orGate extended model to adjust and correct the probability to improve the accuracy of the calculation.

[0089] Assume that node Y has two parent nodes, namely X i and X all , and its corresponding conditional probability is P i and P all Here, X all Represents division by X i For other nodes except

[0090] P(Y|X i )=P i +P all -P i P all ;

[0091]

[0092] Where, To eliminate the unknown factor X l All factors except

[0093] We can further obtain the connection probability P of all parent nodes of node Y i , the formula is:

[0094]

[0095] Combine all unaccounted risk factors into one unknown factor X l , let its connection probability be P l , then the conditional probability of node Y is:

[0096]

[0097] In the safety risk analysis model for road maintenance construction through rural towns, given the uncertainty and ambiguity of various parameters, even randomized statistical experiments cannot accurately determine the risk of the same accident occurring in adjacent time slices. Therefore, the transfer probability should be determined based on the current situation, reference to relevant information, and consultation with engineering experts.

[0098] This example uses a method that combines fuzzy theory with expert scoring to determine indicator probabilities, reducing overreliance on historical statistical data and thus avoiding significant deviations. It also incorporates expert expertise and actual project conditions. The resulting parameter results reflect the dynamic dependencies and transition probabilities of each variable, making the model more accurate in predicting and assessing future risks.

[0099] Because it is difficult for experts to accurately assess probability values, the expert scoring method uses linguistic variables. Five levels of linguistic variables are used, and their corresponding relationships with triangular fuzzy numbers are given (see Table 2). In addition, for projects with large differences in linguistic variables, experts were invited to re-score to reduce the differences.

[0100] Table 2

[0101]

[0102] Experts were invited to evaluate the probability of each node in the DBN structure of the rural highway maintenance and construction, and all experts were given the same weight. After the fuzzy value was averaged, it was converted into an exact value P using the mean area method. The formula is:

[0103]

[0104] According to the data obtained by the expert scoring method, the conditional probability table (CPT) of other nodes still needs to be further determined to represent the conditional dependency relationship between nodes in the DBN model.

[0105] S3. Perform forward reasoning through the dynamic Bayesian network model to generate a dynamic probability curve of the safety risk of maintenance and construction of rural roads; perform reverse reasoning through the dynamic Bayesian network model to calculate the posterior probability of key risk factors.

[0106] Furthermore, the reverse reasoning step includes:

[0107] When the probability of a security risk reaching a preset threshold, the posterior probability of each node is calculated;

[0108] Calculate the proportional change value of each node according to the proportional change formula;

[0109] Ranking the risk factors by sensitivity based on the proportional change values, and marking nodes with proportional change values ​​greater than a preset threshold as key risk factors;

[0110] The key risk factors include speeding by the driver, exceeding the limit or overloading of the vehicle, speed limit of the road, maintenance and safety facilities in the construction area, weather, age of the construction personnel, driver fatigue, fatigue level of the construction personnel, traffic volume and driving experience of the driver.

[0111] Specifically, the calculated node probability is input into the constructed DBN structure, and forward reasoning is performed to predict the dynamic change trend of the safety risk of maintenance and construction of rural roads, which helps decision makers implement early intervention strategies. In addition, the reverse reasoning function of the DBN model is used to calculate the posterior probability of each risk factor when a risk occurs at a specific time point, so as to identify and diagnose the key factors leading to risk events in real time. When the node R state is known, reverse reasoning can obtain the posterior probability of each node and further calculate the ratio of variation (ROV) of each node. The nodes are sorted according to the size of ROV. A higher ROV value indicates that the node has a higher sensitivity. The calculation formula of ROV is:

[0112]

[0113] Where V RO (C i ),π(C i ) and θ(C i ) represent nodes C i ROV, posterior probability and prior probability.

[0114] Example:

[0115] The research object selected is a special maintenance section of a township road passing through villages and towns, with a total length of 3.2km, a speed limit of 60km / h, and two lanes in both directions.

[0116] Field research conducted during road maintenance work revealed that it was the rainy season in the region, and the maintenance area featured complex terrain, numerous slopes, and frequent vehicle traffic. These factors posed significant challenges during the construction process. Since maintenance work was mostly conducted outdoors, rainy weather not only hindered workers and equipment but also posed safety risks to drivers. Rainy weather and slippery road conditions impaired drivers' judgment, increasing the risks of maintenance work.

[0117] To more accurately assess the safety risks of the rural highway maintenance project, this example invited five experts to score and evaluate the project's construction safety risks. These experts included one expert with extensive experience in construction site management, two professors with long-term experience in related assessments, and two senior engineers with professional titles and extensive practical experience. The fuzzy values ​​and initial probabilities of each root node are shown in Table 3, the transition probabilities in Table 4, and the conditional probabilities of each root node in Table 5. When calculating the conditional probability, in order to improve the accuracy of the conditional probability, this embodiment adopts the LeakyNoisy-or Gate extended model to calculate and correct the conditional probability of the non-root node, and obtains P(G2=1|H2=1)=0.810, P(G2=1|H2=0)=0.5; P(S7=1|D1=1, G5=1)=0.917, P(S7=1|D1=1, G5=0)=0.800, P(S7=1|D1=0, G5=1)=0.750, P(S7=1|D1=0, G5=0)=0.660; P(J3=1|D4=1)=0.855, P(J3=1|D4=0)=0.355.

[0118] Table 3

[0119]

[0120] Table 4

[0121]

[0122] Table 5

[0123]

[0124] Forward reasoning:

[0125] Input the probability of each node into the DBN model, set the time slice to 10, and obtain the dynamic probability change curve of the safety risk of maintenance construction under the village and town in different time slices, as shown in the following figure: Figure 3As shown in the figure, the probability of a safety risk occurring during maintenance construction through rural towns in the initial time slice is 0.200. This probability increases with time. During time slices 0-4, the probability of the risk occurring shows a clear upward trend; in subsequent time slices, the probability of the risk occurring levels off, stabilizing at around 0.600. However, this is a relatively high risk level. Therefore, to ensure the safe and reliable completion of maintenance construction on rural roads, measures such as strengthening traffic diversion and management and regularly checking for hidden dangers should be implemented to reduce the likelihood of accidents.

[0126] Reverse reasoning:

[0127] Assuming that the probability of occurrence of safety risks in pavement maintenance construction of rural highways is 100%, the posterior probability of each node is calculated through reasoning. The posterior probability of the first-level indicator is shown in Table 6, and the posterior probability of the second-level indicator and ROV ranking are shown in Table 7. Based on this, the ROV value of each node is calculated according to the formula, and the results are shown in Table 7.

[0128] Table 6 shows that the posterior probabilities of the six risk factors—engineering personnel, driver, construction area and facilities, environmental, road, and vehicle—decline in descending order. This indicates that human factors are the most likely contributor to safety risks in the maintenance and construction of rural highways. Therefore, in actual projects, it is important to strengthen the management and training of engineering personnel, as well as provide guidance and reminders to drivers, to effectively mitigate these risk factors.

[0129] Based on the ROVs at each node, we can determine that driver speeding J3, vehicle overload or overload C3, road speed limit D4, construction area maintenance safety facilities S5, weather H2, engineer age G1, driver fatigue J4, engineer fatigue level G4, traffic volume H3, and driver experience J1 are key factors influencing the safety risk of road maintenance and construction through rural towns. Even small changes in these factors can alter the probability of a safety risk occurring during road maintenance and construction through rural towns. Although the ROV for construction area layout S7 ranks relatively low, its posterior factor is high. Therefore, this factor is also a key factor that cannot be ignored in influencing the safety risk of road maintenance and construction through rural towns. These factors should be given special attention during routine road maintenance work through rural towns.

[0130] Table 6

[0131]

[0132] Table 7

[0133]

[0134] Beneficial effects of this embodiment:

[0135] This embodiment fully considers the dynamic variability and uncertainty of different parameters, provides a model based on dynamic Bayesian networks (DBNs), and qualitatively analyzes the safety issues in the maintenance and construction of roads through villages and towns, thus breaking the limitations of traditional Bayesian networks. Combined with actual engineering cases, the DBN risk analysis model was applied to calculate the safety risk probability of the maintenance and construction of roads through villages and towns, and its dynamically changing probability curve was obtained. The factors that have the greatest impact on the risk in the case projects were identified as the factors of drivers and engineers. The secondary indicators of driver speeding J3, vehicle overload or overload C3, road speed limit D4, maintenance safety facilities in the construction area S5, weather H2, engineer age G1, driver fatigue J4, engineer fatigue, traffic volume H3, and driver driving experience J1 are the key factors affecting the safety risk of the maintenance and construction of roads through villages and towns.

[0136] This model provides a new approach and method for dynamic system risk analysis for projects like rural highway maintenance and construction. It can identify key factors and weaknesses that lead to accidents during construction, helping to improve accident prevention and safety management. In the future, more factors can be taken into account based on actual conditions, leading to more accurate hazard detection and improved safety during rural highway maintenance and construction.

[0137] Example 2

[0138] This embodiment further discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first embodiment are implemented.

[0139] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A dynamic evaluation method for safety risk of maintenance construction of rural highways, characterized by: The following steps are involved: According to the characteristics of maintenance and construction of highways through villages and towns and the theory of accident causes, a risk indicator system with multiple levels of indicators is established; Based on the risk indicator system, a dynamic Bayesian network model is constructed, wherein the model includes root nodes and conditional dependencies between nodes; fuzzy set theory is combined with expert scoring method to determine the fuzzy probability of each root node in the dynamic Bayesian network model; Correcting the conditional probabilities of non-root nodes in the dynamic Bayesian network model based on the Leak Noisy-or Gate extended model; The steps to build a dynamic Bayesian network model include: Generate a network structure based on the causal relationship between the indicators in the risk indicator system; Based on the network structure, determining the node transition probability between adjacent time slices by using a transition probability formula; The dynamic Bayesian network model is used for forward reasoning to generate a dynamic probability curve of the safety risk of maintenance and construction of rural roads; the dynamic Bayesian network model is used for reverse reasoning to calculate the posterior probability of key risk factors.

2. The method according to claim 1, characterized in that The steps to establish a risk indicator system with multiple levels of indicators include: Risk factors are divided into six primary indicators: drivers, engineers, vehicles, roads, environment, construction areas and facilities; For each first-level indicator, set the corresponding second-level indicator.

3. The method according to claim 2, characterized in that The secondary indicators of the driver index include driving experience, gender, speeding and fatigue driving; The secondary indicators of the engineering personnel indicators include age, health status, educational level, fatigue, safety awareness, safety clothing and professional technical capabilities.

4. The method according to claim 1, wherein The transition probability formula is: in, represents the value of the i-th variable at time t, for The parent node of , N represents the number of variables.

5. The method according to claim 1, wherein The step of determining the fuzzy probability of each root node in the dynamic Bayesian network model includes: Convert the experts' linguistic evaluation of nodes into triangular fuzzy numbers; The mean area method is used to convert the triangular fuzzy number into a probability value.

6. The method according to claim 5, characterized in that The membership function of the triangular fuzzy number is: Among them, a, m, and b represent the three values ​​of the triangular fuzzy number interval; express The degree of membership in .

7. The method according to claim 1, characterized in that The steps to correct the conditional probability of non-root nodes include: Introducing the Leak Noisy-or Gate extended model, defining the uncovered risk factors as unknown nodes and adding them to the dynamic Bayesian network model; Based on the unknown node and the known parent node, the conditional probability of node Y is calculated using a conditional probability formula.

8. The method according to claim 7, characterized in that The conditional probability formula is: P(Y|X i )=P i +P all -P i P all ; Among them, Xi is a known parent node, To eliminate the unknown factor X l All factors except X all For other parent nodes, P i For X i The corresponding connection probability, P all For X all The corresponding connection probability.

9. The method according to claim 1, characterized in that The steps of reverse reasoning include: When the probability of a security risk reaching a preset threshold, the posterior probability of each node is calculated; Calculate the proportional change value of each node according to the proportional change formula; Ranking the risk factors by sensitivity based on the proportional change values, and marking nodes with proportional change values ​​greater than a preset threshold as key risk factors; Key risk factors include speeding, vehicle overload, road speed limits, construction area maintenance and safety facilities, weather, age of construction workers, driver fatigue, worker fatigue, traffic volume, and driver experience. Wherein, the ratio change formula is: Among them, V RO (C i ) is node C i ROV,π(C i ) is the posterior probability, θ(C i ) is the prior probability.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images