Dynamic bayesian-based road reconstruction and expansion safety situation analysis method and device

Abstract

The application discloses a highway reconstruction and expansion safety situation analysis method and device based on dynamic Bayesian, and relates to the technical field of traffic facilities. The method comprises the following steps: acquiring data of a to-be-predicted highway reconstruction and expansion project; inputting the data into a constructed dynamic Bayesian analysis model; and obtaining evolution prediction results of a traffic safety state of the to-be-predicted highway reconstruction and expansion project over time, safety situation change prediction results, influence factor sensitivity analysis results and accident cause diagnosis results according to the data and the dynamic Bayesian analysis model. The application constructs a DBN model suitable for traffic safety situation analysis of a highway reconstruction and expansion project, predicts traffic safety situation changes, analyzes the sensitivity of influence factors, diagnoses accident causes, and provides a scientific basis for taking targeted traffic control measures.

Description

Technical Field

[0001] This invention relates to the field of traffic safety facilities technology, and in particular to a method and device for analyzing the safety situation of highway reconstruction and expansion based on dynamic Bayesian methods. Background Technology

[0002] By the end of 2020, my country's total expressway mileage exceeded 160,000 kilometers, continuing to rank first in the world. However, with the annual increase in traffic volume, the existing expressway network can no longer meet the growing traffic demand, leading to an increasing number of reconstruction and expansion projects. Compared with new construction projects, expressway reconstruction and expansion projects adopt a "construction while traffic is open" approach, resulting in significant differences in traffic safety influencing factors and safety conditions, making the safety situation exceptionally severe. Conducting traffic safety analysis of expressway reconstruction and expansion projects plays a crucial role in improving the safety situation.

[0003] Analyzing the patterns of accident occurrence through mathematical models is an effective way to improve traffic safety. Studies show that traffic accidents have a spatial effect, exhibiting a high incidence rate on certain road sections. For example, Sara et al. proposed a binary choice model, and Zeng Qiang et al. established a Bayesian model to analyze the risk probability of accident locations. Some scholars, considering the spatial effect of accidents, have introduced a temporal effect to conduct spatiotemporal coupling analysis, which can more fully explain changes in traffic safety conditions. For instance, Yu Guizhen et al., based on traffic flow theory, analyzed the aggregation and dissipation process of queuing vehicles upstream of highway accidents and established a spatiotemporal impact model for highway traffic accidents. Ma et al. considered the spatiotemporal correlation of accidents and analyzed the frequency of accidents with and without injuries on a certain highway. In recent years, data-driven machine learning modeling techniques have become an important method for road traffic safety analysis. For example, Li et al. used a multiple linear regression model to comprehensively assess and score the risk of driver steering and acceleration / deceleration. Li Guiyang et al. used a support vector machine model to assess the impact of potential risk factors on the severity of accidents on mountain highways. Huang Ling et al. constructed a long short-term memory neural network to analyze the lane-changing behavior of unmanned vehicles. However, because machine learning techniques are based on deterministic data and do not consider the sporadic and highly random uncertainties of traffic accidents, their fitting and explanatory effects on relevant factors are poor. Therefore, to explain and predict the evolution of traffic safety trends, a series of probability-based traffic safety risk analysis models have emerged, mainly including the negative binomial (Poisson-Gamma) model, the Poisson-log-normal model, the conditional autoregressive model, the random effects model, the stochastic parameter model, and the generalized event counting model. Bayesian methods, as an important probabilistic approach, have been widely applied in traffic safety analysis. For example, Tong Yao used Bayesian networks (BN) analysis to evaluate and analyze the traffic safety status of urban river-crossing tunnels from the system perspective of the three elements of people, vehicles, and environment. Liu Zhiqiang et al. used Bayesian networks to analyze the mechanism of highway traffic accidents under hazy weather. Xie Kun et al. designed a hierarchical Bayesian model to analyze the safety status of signalized intersections. Guo et al.'s research within the Bayesian framework shows that considering the spatial heterogeneity of unobserved accident locations can significantly improve the model's fitting accuracy. Studies have shown that time factors have a significant impact on safety analysis, and DBN (Dynamic Bayesian networks) is an effective means of describing the characteristics of factors changing over time. For example, Wu et al. established a dynamic Bayesian network model to analyze the changes in safety status over time during subway construction. Zhang Jinglei et al. used a dynamic Bayesian network model to identify the safety status of traffic flow. Wang Kaiming et al. used a dynamic Bayesian network to analyze the reliability of high-speed railway traction substations.KIM et al. developed a decision-making process that combines dynamic probabilistic risk assessment and dynamic Bayesian networks with functional modeling, enabling accurate prediction and analysis of risk situations.

[0004] Current research primarily utilizes machine learning and probabilistic models to analyze and explain the causes of traffic accidents under normal operation and maintenance conditions. However, due to the complex traffic conditions during highway reconstruction and expansion projects, the traffic safety characteristics differ significantly from those of highways in normal operation, especially the evolution mechanism of traffic safety situations under time influence remains unclear. Therefore, this paper constructs a DBN model suitable for analyzing traffic safety situations in highway reconstruction and expansion projects. This model predicts changes in traffic safety situations, conducts sensitivity analysis on influencing factors, and diagnoses accident causes, providing a scientific basis for implementing targeted traffic control measures. Summary of the Invention

[0005] This invention addresses the problem of how to construct a model suitable for traffic safety situation analysis in highway reconstruction and expansion projects, predict changes in traffic safety situation, conduct sensitivity analysis on influencing factors, and diagnose accident causes, thereby providing a scientific basis for taking targeted traffic control measures.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] On the one hand, this invention provides a method for safety situation analysis of highway reconstruction and expansion based on dynamic Bayesian methods. This method is implemented by electronic devices and includes:

[0008] S1. Obtain data on highway reconstruction and expansion projects to be predicted.

[0009] S2. Input the data into the constructed dynamic Bayesian analysis model.

[0010] S3. Based on the data and the dynamic Bayesian analysis model, the predicted results of the evolution of traffic safety status over time, the predicted results of safety situation changes, the results of sensitivity analysis of influencing factors, and the results of accident cause diagnosis are obtained for the highway reconstruction and expansion project to be predicted.

[0011] Optionally, the construction process of the dynamic Bayesian analysis model in S2 includes:

[0012] S21. Obtain risk factors that affect the safety of highway reconstruction and expansion projects.

[0013] S22. Based on the risk factors, construct an indicator system for the risk factors.

[0014] S23. Based on the indicator system, construct the network structure of the dynamic Bayesian network model.

[0015] S24. Calculate the conditional probability distribution of each node in the network structure to obtain a dynamic Bayesian analysis model.

[0016] Optionally, the risk factors in S21 include road conditions and traffic conditions themselves.

[0017] The road traffic conditions include the length of the work area, temporary traffic facilities, roadside clearance, and emergency lanes.

[0018] Traffic conditions themselves include traffic composition, traffic flow, speed limits, and traffic conflicts.

[0019] Optionally, the conditional probability distribution of each node in the computational network structure in S24 includes:

[0020] S241. Obtain the linguistic value evaluation of the root node of the network structure and the uncertainty of the linguistic value evaluation based on expert knowledge.

[0021] S242. Based on the linguistic value evaluation and the Gaussian membership function, the linguistic value evaluation is converted into a fuzzy interval value.

[0022] S243. Based on the upper and lower limits of the fuzzy interval values, an interval fuzzy set is introduced. The uncertainty of the language value evaluation is quantitatively represented by the interval fuzzy set to obtain the quantified language evaluation value.

[0023] S244. Based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, the conditional probability distribution of each node in the network structure is obtained.

[0024] Optionally, in S244, the conditional probability distribution of each node in the network structure, based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, includes:

[0025] S2441. Calculate the conflict coefficient K of classical evidence theory.

[0026] S2442. Perform conflict detection on the conflict coefficient K according to the preset threshold ε.

[0027] S2443. If the threshold ε is greater than or equal to the conflict coefficient K, then the improved hybrid interval evidence synthesis rule and the quantified language evaluation value are used to obtain the conditional probability distribution of each node in the network structure.

[0028] S2444. If the threshold ε is less than the conflict coefficient K, then the classical evidence theory synthesis rules and the quantified language evaluation values ​​are used to obtain the conditional probability distribution of each node in the network structure.

[0029] The construction process of the improved hybrid interval evidence synthesis rule includes: redefining the allocation of conflicting information based on the credibility of the evidence to obtain the improved hybrid interval evidence synthesis rule.

[0030] Optionally, the predicted results in S3 for the evolution of the traffic safety status of the highway reconstruction and expansion project over time include...

[0031] Identify state transition risk factors among the risk factors; among them, state transition risk factors are risk factors that change over time and have a significant impact on traffic safety status.

[0032] The state transition probability of state transition risk factors is calculated using the Markov chain method.

[0033] Based on the state transition probability, the prediction results of the evolution of traffic safety status of the highway reconstruction and expansion project over time are obtained.

[0034] Optionally, the calculation method for obtaining the predicted safety situation change results of the highway reconstruction and expansion project in S3 is as shown in the following formula (1):

[0035] P(R=r)=P(R=r|X1=x1,…,X N =x N )×P(X1=x1,…,x i ,…,X N =x N (1)

[0036] Where P(R=r) is the probability distribution of the security situation R; r={r1,…,r P},{r1,…,r P} represents the P state ranges of the security situation R; Risk factor X i Q i A range of states, i = 1, ..., N; P(R = r|X1 = x1, ..., X N =x N Let P(X1=x1,…,x) be the conditional probability distribution of the security situation R; i ,…,X N =x N Risk factor X i The joint probability distribution.

[0037] Optionally, the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted obtained in S3 include:

[0038] The sensitivity performance measure of each risk factor is calculated, and the contribution of the risk factor to the safety situation is measured based on the sensitivity performance measure. Key risk factors are identified based on the contribution, and the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted are obtained.

[0039] Optionally, the accident cause diagnosis results obtained in S3 for the highway reconstruction and expansion project to be predicted include:

[0040] Obtain the posterior probability distribution of each risk factor in the highway reconstruction and expansion project to be predicted at the time of the accident; obtain the accident cause diagnosis result of the highway reconstruction and expansion project to be predicted based on the posterior probability distribution.

[0041] On the other hand, the present invention provides a highway reconstruction and expansion safety situation analysis device based on dynamic Bayesian methods. This device is used to implement a highway reconstruction and expansion safety situation analysis method based on dynamic Bayesian methods. The device includes:

[0042] The acquisition module is used to acquire data on highway reconstruction and expansion projects to be predicted.

[0043] The input module is used to input data into the constructed dynamic Bayesian analysis model.

[0044] The output module is used to obtain, based on data and a dynamic Bayesian analysis model, prediction results of the evolution of traffic safety status over time for the highway reconstruction and expansion project, prediction results of safety situation changes, sensitivity analysis results of influencing factors, and accident cause diagnosis results.

[0045] Optionally, the input module is further used for:

[0046] S21. Obtain risk factors that affect the safety of highway reconstruction and expansion projects.

[0047] S22. Based on the risk factors, construct an indicator system for the risk factors.

[0048] S23. Based on the indicator system, construct the network structure of the dynamic Bayesian network model.

[0049] S24. Calculate the conditional probability distribution of each node in the network structure to obtain a dynamic Bayesian analysis model.

[0050] Alternatively, risk factors include road conditions and traffic conditions themselves.

[0051] The road traffic conditions include the length of the work area, temporary traffic facilities, roadside clearance, and emergency lanes.

[0052] Traffic conditions themselves include traffic composition, traffic flow, speed limits, and traffic conflicts.

[0053] Optionally, the input module is further used for:

[0054] S241. Obtain the linguistic value evaluation of the root node of the network structure and the uncertainty of the linguistic value evaluation based on expert knowledge.

[0055] S242. Based on the linguistic value evaluation and the Gaussian membership function, the linguistic value evaluation is converted into a fuzzy interval value.

[0056] S243. Based on the upper and lower limits of the fuzzy interval values, an interval fuzzy set is introduced. The uncertainty of the language value evaluation is quantitatively represented by the interval fuzzy set to obtain the quantified language evaluation value.

[0057] S244. Based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, the conditional probability distribution of each node in the network structure is obtained.

[0058] Optionally, the input module is further used for:

[0059] S2441. Calculate the conflict coefficient K of classical evidence theory.

[0060] S2442. Perform conflict detection on the conflict coefficient K according to the preset threshold ε.

[0061] S2443. If the threshold ε is greater than or equal to the conflict coefficient K, then the improved hybrid interval evidence synthesis rule and the quantified language evaluation value are used to obtain the conditional probability distribution of each node in the network structure.

[0062] S2444. If the threshold ε is less than the conflict coefficient K, then the classical evidence theory synthesis rules and the quantified language evaluation values ​​are used to obtain the conditional probability distribution of each node in the network structure.

[0063] The construction process of the improved hybrid interval evidence synthesis rule includes: redefining the allocation of conflicting information based on the credibility of the evidence to obtain the improved hybrid interval evidence synthesis rule.

[0064] Optionally, the output module is further used for:

[0065] Identify state transition risk factors among the risk factors; among them, state transition risk factors are risk factors that change over time and have a significant impact on traffic safety status.

[0066] The state transition probability of state transition risk factors is calculated using the Markov chain method.

[0067] Based on the state transition probability, the prediction results of the evolution of traffic safety status of the highway reconstruction and expansion project over time are obtained.

[0068] Optionally, the calculation method for obtaining the predicted results of the safety situation changes of the highway reconstruction and expansion project is as shown in the following formula (1):

[0069] P(R=r)=P(R=r|X1=x1,…,X N =x N )×P(X1=x1,…,x i ,…,X N =x N (1)

[0070] Where P(R=r) is the probability distribution of the security situation R; r={r1,…,r P},{r1,…,r P} represents the P state ranges of the security situation R; Risk factor X i Q i A range of states, i = 1, ..., N; P(R = r|X1 = x1, ..., X N =x N Let P(X1=x1,…,x) be the conditional probability distribution of the security situation R; i ,…,X N =x N Risk factor X i The joint probability distribution.

[0071] Optionally, the output module is further used for:

[0072] The sensitivity performance measure of each risk factor is calculated, and the contribution of the risk factor to the safety situation is measured based on the sensitivity performance measure. Key risk factors are identified based on the contribution, and the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted are obtained.

[0073] Optionally, the output module is further used for:

[0074] Obtain the posterior probability distribution of each risk factor in the highway reconstruction and expansion project to be predicted at the time of the accident; obtain the accident cause diagnosis result of the highway reconstruction and expansion project to be predicted based on the posterior probability distribution.

[0075] On the one hand, an electronic device is provided, which includes a processor and a memory, wherein the memory stores at least one instruction, which is loaded and executed by the processor to implement the above-mentioned method for analyzing the safety situation of highway reconstruction and expansion based on dynamic Bayes.

[0076] On the one hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-mentioned method for analyzing the safety situation of highway reconstruction and expansion based on dynamic Bayes.

[0077] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0078] The above scheme constructs a DBN model suitable for traffic safety situation analysis in highway reconstruction and expansion projects, predicts changes in traffic safety situation, conducts sensitivity analysis on influencing factors, and diagnoses the causes of accidents, providing a scientific basis for taking targeted traffic control measures. Attached Figure Description

[0079] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0080] Figure 1 This is a schematic diagram of the method for analyzing the safety situation of highway reconstruction and expansion based on dynamic Bayes, provided in an embodiment of the present invention.

[0081] Figure 2 This invention provides a traffic organization diagram and an accident scene diagram for a single-lane two-way road section in a reconstruction and expansion project.

[0082] Figure 3 This is a schematic diagram of the DBN structure for traffic safety situation analysis in highway reconstruction and expansion projects provided in this embodiment of the invention;

[0083] Figure 4 This is a traffic safety situation prediction map for a certain road section provided in an embodiment of the present invention;

[0084] Figure 5 This is a sensitivity analysis chart of various factors under different safety status levels provided in the embodiments of the present invention;

[0085] Figure 6 This is a block diagram of the highway reconstruction and expansion safety situation analysis device based on dynamic Bayes provided in the embodiments of the present invention;

[0086] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0087] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0088] like Figure 1 As shown, this embodiment of the invention provides a method for analyzing the safety situation of highway reconstruction and expansion based on dynamic Bayesian methods, which can be implemented by electronic devices. Figure 1 The flowchart shown is for a safety situation analysis method for highway reconstruction and expansion based on dynamic Bayesian methods. The processing flow of this method may include the following steps:

[0089] S1. Obtain data on highway reconstruction and expansion projects to be predicted.

[0090] In one feasible implementation, acquiring data on the highway reconstruction and expansion project to be predicted can involve acquiring data corresponding to risk factors. Furthermore, this patent applies not only to highways but also to ordinary roads.

[0091] S2. Input the data into the constructed dynamic Bayesian analysis model.

[0092] Optionally, the construction process of the dynamic Bayesian analysis model in S2 includes:

[0093] S21. Obtain risk factors that affect the safety of highway reconstruction and expansion projects.

[0094] Optionally, the risk factors in S21 include road conditions and traffic conditions themselves.

[0095] The road traffic conditions include the length of the work area, temporary traffic facilities, roadside clearance, and emergency lanes.

[0096] Traffic conditions themselves include traffic composition, traffic flow, speed limits, and traffic conflicts.

[0097] In one feasible implementation, risk factor identification includes determining which risk factors may affect project safety and recording their characteristics. Risk factor identification mainly includes acquiring prior knowledge, conducting risk mechanism analysis on target risk events, and identifying relevant influencing factors representing the system's safety status, process, and performance status, as well as their explicit relationships and correlations. For complex systems such as highway reconstruction and expansion traffic, numerous influencing factors are involved. Based on relevant project research, combined with standards and specifications, expert experience, and previous research results, it mainly includes two levels: road traffic conditions B1 and traffic conditions themselves B2.

[0098] Road traffic conditions mainly include four factors: work zone length (X1), temporary traffic facilities (X2), roadside clearance (X3), and emergency lane (X4). The length of the work zone directly affects the road's service level; the longer the work zone, the greater the impact on traffic. Temporary traffic facilities play a crucial role in reconstruction and expansion projects; their absence directly affects vehicle speed changes, direction adjustments, and route changes. Roadside clearance directly impacts normal vehicle movement within the lanes; smaller clearances increase the likelihood of collisions with roadside facilities. Emergency lanes directly affect the ability to stop vehicles promptly in the event of an accident, minimizing disruption to other vehicles.

[0099] The main factors influencing traffic conditions include traffic composition (proportion of large vehicles) x5, traffic flow x6, speed limit x7, and traffic conflict x8. Traffic composition (proportion of large vehicles) has a significant impact on traffic safety; the higher the proportion of large vehicles, the more likely traffic accidents are to occur. Increased traffic flow reduces road service levels, increases the probability of traffic accidents, and increases the difficulty of traffic management on road sections. Speed ​​limits are a crucial direct factor affecting traffic safety; higher speed limits lead to higher accident rates and greater accident severity. Traffic conflict is an important factor in assessing the stability of traffic flow; in areas with severe traffic conflict, such as traffic flow conversion points in the central median of highway reconstruction and expansion projects, traffic accidents are more frequent.

[0100] For example, in 2021, a highway in a certain province, specifically Highway A, was undergoing a "four-lane to eight-lane" expansion project. The traffic organization was a two-lane, two-way traffic system on one side, with movable steel guardrails separating opposing traffic flows. At 8:50 AM on November 18th, at kilometer marker K254+500 on Highway A, a point where traffic flow changes at the opening of the central median, with a speed limit of 60 km / h, a high-speed truck veered off course and entered the expansion construction area. It collided with temporary barriers and braked suddenly, causing no injuries. After nearly an hour of emergency response, the situation was resolved at 9:50 AM, and traffic resumed. The accident resulted in damage to approximately 50 meters of movable steel guardrails, nearly totaling the truck and destroying its cargo. It also caused varying degrees of damage to five private cars, with total economic losses exceeding 600,000 yuan. At the time of the accident, the traffic diversion was not yet complete, the left-hand lane was not yet open, and the relevant temporary traffic safety facilities were not yet fully installed. Traffic organization diagram and accident details at the stage of the accident are as follows: Figure 2 As shown in the figure, the locations marked with stars are the accident sites.

[0101] Furthermore, the node states are classified. At the time of the accident, the traffic organization implementation section had an existing central median on one side and movable steel guardrails / fences with Class A protection on the other, effectively isolating the traffic lanes from the construction area. This ensures that the accident will not harm construction workers or machinery. Therefore, assuming only vehicles and passengers are injured in the accident, the accident level can be determined according to the "Notice of the Ministry of Public Security on Revising the Standards for Classifying Road Traffic Accident Levels." The safety situation R is divided into four levels: safe (Ⅰ), relatively safe (Ⅱ), relatively dangerous (Ⅲ), and dangerous (Ⅳ). Simultaneously, based on the safety level classification of each risk factor, the root and intermediate node states in the Bayesian network are divided into three levels: good, medium, and poor.

[0102] S22. Based on the risk factors, construct an indicator system for the risk factors.

[0103] In one feasible implementation, based on the results of risk identification, an indicator system is constructed, as shown in Table 1 below, to classify the indicator safety status levels:

[0104] Table 1

[0105]

[0106] For example, based on the constructed indicator system, combined with the DBN model structure design ideas and specific practices, and integrating prior knowledge such as fault trees and expert experience, the causal relationships between nodes were determined, and a DBN network for traffic safety status in highway reconstruction and expansion projects was established, with the structure as follows: Figure 3 As shown.

[0107] S23. Based on the indicator system, construct the network structure of the dynamic Bayesian network model.

[0108] In one feasible implementation, the goal of structure learning in constructing the network structure of a dynamic Bayesian network model is to find a suitable directed acyclic graph (DBN) and determine the relationships between nodes. Every variable in the real world can be represented by a Bayesian node. Based on the risk identification results, DBN nodes are first created. Then, directed edges are created using existing methods such as fault trees to establish the network structure.

[0109] S24. Calculate the conditional probability distribution of each node in the network structure to obtain a dynamic Bayesian analysis model.

[0110] Optionally, the conditional probability distribution of each node in the computational network structure in S24 includes:

[0111] S241. Obtain the linguistic value evaluation of the root node of the network structure and the uncertainty of the linguistic value evaluation based on expert knowledge.

[0112] In one feasible implementation, experts in the field typically provide linguistic value evaluations and uncertainties regarding the risk level of the root node in the DBN model based on their engineering experience and knowledge. The experts' linguistic value evaluations can be converted into fuzzy interval values ​​using a membership function, thus fuzzifying the expert evaluations. The Gaussian membership function reflects the characteristics of a nonlinear normal distribution and can effectively represent linguistic value evaluations in a fuzzy manner. The calculation method is shown in equation (1):

[0113]

[0114] In the formula, a is the quantified value corresponding to the expert evaluation language value; μ is the center of the function; σ is the standard deviation, representing the root mean square width of the function.

[0115] S242. Based on the linguistic value evaluation and the Gaussian membership function, the linguistic value evaluation is converted into a fuzzy interval value.

[0116] In one feasible implementation, to better represent uncertainty, an interval fuzzy set is introduced based on the upper and lower limits of the membership degree to quantify the uncertainty. The Gaussian membership function u... A Membership degree and It can be determined by equation (2):

[0117]

[0118] In the formula, α is a constant, and α∈[1,+∞).

[0119] S243. Based on the upper and lower limits of the fuzzy interval values, an interval fuzzy set is introduced. The uncertainty of the language value evaluation is quantitatively represented by the interval fuzzy set to obtain the quantified language evaluation value.

[0120] S244. Based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, the conditional probability distribution of each node in the network structure is obtained.

[0121] Optionally, the quantified expert language evaluation values ​​are fused using evidence theory to obtain conditional probabilities. However, it is worth noting that the classic evidence theory synthesis rules have certain shortcomings. To minimize the negative impact of highly conflicting evidence, an improved hybrid interval evidence synthesis rule is proposed based on evidence credibility. In S244, based on the quantified language evaluation values ​​and the improved hybrid interval evidence synthesis rule, the conditional probability distribution of each node in the network structure is obtained as follows:

[0122] S2441. Calculate the conflict coefficient K of classical evidence theory.

[0123] In one feasible implementation, according to classical evidence theory, the degree of conflict among n pieces of evidence is represented by the conflict coefficient K as follows:

[0124]

[0125] In the formula A jn (j = 1, 2, ..., n) is the j-th focal element of the n-th piece of evidence; This represents the empty set.

[0126] S2442. Perform conflict detection on the conflict coefficient K according to the preset threshold ε.

[0127] In one feasible implementation, to differentiate the degree of conflict between pieces of evidence and thus apply different fusion rules, a threshold ε can be set to detect conflicts in the conflict coefficient K. That is, when K ≥ ε, it is considered a high-conflict situation, and an improved fusion rule is used for fusion; otherwise, classical evidence theory is used. Based on relevant literature, the threshold ε can be set to 0.95.

[0128] S2443. If the threshold ε is greater than or equal to the conflict coefficient K, then the improved hybrid interval evidence synthesis rule and the quantified language evaluation value are used to obtain the conditional probability distribution of each node in the network structure.

[0129] In one feasible implementation, conflict between pieces of evidence is acceptable when the conflict coefficient K < ε. For the interval basic probability assignment function m1,…,m n The lower limit of the synthesized interval value and upper limit Determined by equation (4):

[0130]

[0131] S2444. If the threshold ε is less than the conflict coefficient K, then the classical evidence theory synthesis rules and the quantified language evaluation values ​​are used to obtain the conditional probability distribution of each node in the network structure.

[0132] The construction process of the improved hybrid interval evidence synthesis rule includes: redefining the allocation of conflicting information based on the credibility of the evidence to obtain the improved hybrid interval evidence synthesis rule.

[0133] In one feasible implementation, when the conflict coefficient K≥ε, in order to avoid the defect that the classical evidence synthesis rules are prone to contradicting the facts under highly conflicting evidence, the allocation of conflicting information is redefined based on the credibility of the evidence, resulting in a new evidence synthesis rule. The specific algorithm is as follows:

[0134] The credibility of evidence is obtained by calculating the support level of the evidence. The evidence m is calculated according to equation (5). i and mj The distance d between ij Then calculate the evidence m i Support Sup(m) i Evidence m i Credibility of CRDE i via Sup(m) i Normalization yields the result, as shown in equation (6):

[0135]

[0136]

[0137] The higher the similarity of a piece of evidence to other evidence, the more supported it is by the other evidence, and the more credible it is. Conflicting information is distributed proportionally to the evidence based on its credibility, for the interval basic probability assignment function m1,…,m n The lower and upper limits of the synthesized interval values ​​are determined by equation (7):

[0138]

[0139] For example, since there is a lack of accurate traffic accident statistics for highway reconstruction and expansion projects, five experienced experts (see Table 2 for relevant information) were invited to complete a questionnaire to establish quantitative connections between nodes and conditional probability distribution tables for the nodes. Based on the root node status in Table 1, the quantitative values ​​of the expert language evaluations at three levels were defined as 0.1, 0.5, and 0.9, respectively. The Gaussian μ values ​​for the three status levels were set to 0, 0.5, and 1, respectively, thus converting the expert language evaluations into fuzzy interval values. The improved fuzzy evidence theory was then used to perform information fusion processing on the expert knowledge, and the interval evidence fusion results are shown in Table 3. Using the inference function of DBN, the probabilities of intermediate nodes and leaf nodes at each situational level can be calculated given the prior probability of the root node. The average probabilities of intermediate nodes B1 and B2 and leaf node R are shown in Table 4.

[0140] Table 2

[0141]

[0142] Table 3

[0143]

[0144]

[0145] Table 4

[0146]

[0147] S3. Based on the data and the dynamic Bayesian analysis model, the predicted results of the evolution of traffic safety status over time, the predicted results of safety situation changes, the results of sensitivity analysis of influencing factors, and the results of accident cause diagnosis are obtained for the highway reconstruction and expansion project to be predicted.

[0148] Optionally, the predicted results in S3 for the evolution of the traffic safety status of the highway reconstruction and expansion project over time include...

[0149] Identify the state transition risk factors within the risk factors.

[0150] Among them, state transition risk factors are risk factors that change over time and have a significant impact on traffic safety status.

[0151] The state transition probability of state transition risk factors is calculated using the Markov chain method.

[0152] Based on the state transition probability, the prediction results of the evolution of traffic safety status of the highway reconstruction and expansion project over time are obtained.

[0153] In one feasible implementation, the state transition probabilities are typically learned using a Markov chain method. Discrete-time Markov chains assume that the next state x[t-1] depends only on the current state x[t] and not on the sequence of events preceding it, as shown in equation (8):

[0154]

[0155] In the formula, X = {x[0],…,x[T-1]} represents the variables of T+1 hidden states, Y = {y[0],…,y[T-1]} represents the variables of T manifest states, Pr(x[t]|x[t-1]) is the state transition probability density function, Pr(y[t]|x[t]) is the observation probability density function, and Pr(x0) is the initial state distribution.

[0156] For example, in the established DBN, X6 changes over time and has a significant impact on traffic safety status. Establishing a state transition probability table for X6 can predict the evolution of traffic safety status over time. Based on the state transition probability learning method in formula (8), the state transition probability table for X6 can be calculated, as shown in Table 5:

[0157] Table 5

[0158]

[0159] Optionally, the calculation method for obtaining the predicted change in the safety situation of the highway reconstruction and expansion project in S3 is shown in the following formula (9):

[0160] P(R=r)=P(R=r|X1=x1,…,X N =x N )×P(X1=x1,…,x i ,…,X N =x N (9)

[0161] Where P(R=r) is the probability distribution of the security situation R; r={r1,…,r P},{r1,…,r P} represents the P state ranges of the security situation R; Risk factor X i Q i A range of states, i = 1, ..., N; P(R = r|X1 = x1, ..., X N =x N Let P(X1=x1,…,x) be the conditional probability distribution of the security situation R; i ,…,X N =x N Risk factor X i The joint probability distribution.

[0162] In one feasible implementation, predictive reasoning analysis predicts the security posture defined by R under the influence of factors (X1,…,X). N The probability distribution of R under the combined effects of ) is denoted by P(R=r), which can be used as an indicator to evaluate R.

[0163] For example, based on the established DBN model, the probability of a road segment's safety situation level occurring and its trend over time can be predicted. To analyze the relationship between accidents and traffic safety situation, the changes in traffic safety situation are predicted and analyzed within 30 minutes before and after an accident, as well as during the accident handling process. This time period is divided into 15-minute segments, and the influencing factor X is... i Evidence information from (i = 1, ..., 8) is input into the DBN model to predict the security situation level. The results are shown below. Figure 4The accident occurred in time slot 2, and the incident was resolved in time slot 6. As shown in the graph, the overall traffic safety status of this road segment is between levels III and IV. Level IV occurs in time slots 0-3 and 6-9, with the highest probability value approaching 0.6. Level III occurs in time slots 3-6, and the probabilities of levels III and IV are extremely close, with only a significant difference in probability between level III and level IV in time slot 6. The probabilities of levels I and II are relatively low, generally below 0.2. These results indicate that the traffic safety status of this road segment is relatively dangerous. The main reason for the trend change is that, starting from time slice 0, the probability of a Level IV safety state rises to around 0.6, indicating a deterioration in safety. In time slice 2, a traffic accident occurs, disrupting traffic and bringing the traffic safety state towards a safer level. However, as vehicles gradually accumulate upstream of the accident site, rear-end collisions are likely to occur at the rear of the queue, resulting in the highest predicted probability of Level III, which effectively reflects the traffic safety risk during this period. After the accident is resolved and traffic resumes, the traffic flow gradually increases, leading to the highest predicted probability of a Level IV safety state again, indicating a dangerous traffic safety state. Under the control of on-site emergency personnel, the accumulated vehicles gradually leave, traffic flow decreases, and the overall road safety state returns to a safe level. The model prediction results can accurately characterize the trend of safety state changes over time.

[0164] Predictive analysis indicates that the traffic safety status of this road section is at Level III or above, suggesting a relatively serious traffic safety situation. Corresponding measures should be taken, such as traffic diversion through the road network and controlling the proportion of large vehicles. Traffic diversion through the road network can significantly reduce traffic flow on this section; controlling the proportion of large vehicles can not only reduce traffic flow but also improve traffic composition and promote an improvement in the traffic safety status of the road section.

[0165] Optionally, the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted obtained in S3 include:

[0166] The sensitivity performance measure of each risk factor is calculated, and the contribution of the risk factor to the safety situation is measured based on the sensitivity performance measure. Key risk factors are identified based on the contribution, and the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted are obtained.

[0167] In one feasible implementation, sensitivity analysis refers to how sensitive the model's performance is to small changes in input parameters. Sensitivity analysis is particularly useful when investigating the impact of each risk factor on the security posture. The most straightforward way to perform sensitivity analysis is to change the values ​​of the input parameters and then monitor the impact of the changes on the output probabilities. The Indicator Sensitivity Performance Measure (SPM) is typically used to measure the impact of each influencing factor X. iThe contribution to the security posture R, thereby identifying key risk factors that influence it, helps decision-makers determine key checkpoints during the construction phase. Based on actual observed events, such as X... i It is observed in the state sequence q i ,Right now Then SPM(X) i It can be calculated using equation (10):

[0168]

[0169] In the formula, r represents the security situation R with P states; Indicates risk factor X i With Q i The state in; Indicates risk factor X i The j-th state. In general, SPM(X) i ) can be used as a measure of the root node X i Sensitivity indicators during risk events. When SPM(X) i When X approaches 1, i It is more likely to become a sensitive factor for risk events and should be given more attention.

[0170] For example, sensitivity analysis can help managers of participating units identify key influencing factors so as to take targeted control measures. For the DBN model, a specific time slice can be selected, and the actual observed results can be input into the model as evidence. Then, the X of each factor can be calculated using formula (10). i The sensitivity of (i = 1, ..., 8) is shown in the results. Figure 5 In the relatively safe Level I state, X5, X8, and X1 are considered relatively sensitive factors; in the relatively safe Level II state, X5 and X6 are considered relatively sensitive factors; in the relatively poor Level III and IV states, X5 and X8 are considered the most sensitive influencing factors due to their highest sensitivity. X5 exhibits high sensitivity across all four states and should be given special attention. Typically, the observation status of influencing factors differs across scenarios. In practical applications, targeted attention and control should be implemented for different influencing factors based on the safety status of different time slices of traffic organization sections in highway reconstruction and expansion projects.

[0171] Optionally, the accident cause diagnosis results obtained in S3 for the highway reconstruction and expansion project to be predicted include:

[0172] Obtain the posterior probability distribution of each risk factor in the highway reconstruction and expansion project to be predicted at the time of the accident; obtain the accident cause diagnosis result of the highway reconstruction and expansion project to be predicted based on the posterior probability distribution.

[0173] In one feasible implementation, the purpose of diagnostic analysis is to obtain the posterior probability distribution of each influencing factor at the time of the accident. Using the posterior probability distribution, suspicious causes can be detected in a very short time. Influencing factor X i The posterior probability distribution is given by P(X) i =x i |R=r) represents, which can be calculated by equation (11):

[0174]

[0175] In the formula, r represents the security situation R with P states; x i Indicates influencing factor X i With Q i State; typically, when P(X) i =x i The closer |R=r) is to 1, the better X i It is more likely to be the direct cause of the security situation R.

[0176] For example, the example accident, which resulted in the total loss and damage of multiple vehicles, can be classified as a particularly serious accident according to the "Notice of the Ministry of Public Security on Revising the Standards for Classifying Road Traffic Accidents," thus placing the safety status of this road section at Level IV. Using the established DBN model to diagnose and infer the cause of this accident, the inference results for the time slice 2 are shown in Table 6. The cause of this traffic accident is likely due to the inadequate installation of temporary traffic safety facilities. Simultaneously, the posterior probabilities of speed limits and traffic conflicts are relatively high, serving as secondary causes. The accident occurred at a traffic flow conversion point at the opening of the central median, and was under traffic diversion implementation. Temporary traffic safety facilities such as speed limits and warning signs were not yet fully installed, resulting in significant differences in vehicle speeds and severe traffic conflicts. Therefore, the DBN inference results can effectively explain the cause and main influencing factors of this accident. During the implementation of highway reconstruction and expansion projects, participating units should promptly implement temporary traffic safety facility installation and traffic management in accordance with the traffic organization plan. Meanwhile, the posterior probability analysis reveals that the posterior probability of factor X6 in time slice 3 is 0.435, and this factor has the highest posterior probability when it is in state III across all time slices. This indicates that traffic flow also had a certain impact on the accident. During highway reconstruction and expansion construction, participating units should actively implement traffic flow control measures, such as road network diversion and restrictions on large vehicles, to reduce traffic flow on construction sections.

[0177] Table 6

[0178]

[0179]

[0180] In this embodiment of the invention, a DBN model suitable for traffic safety situation analysis of highway reconstruction and expansion projects was constructed to predict changes in traffic safety situation, conduct sensitivity analysis on influencing factors, and diagnose the causes of accidents, providing a scientific basis for taking targeted traffic control measures.

[0181] like Figure 6 As shown, this embodiment of the invention provides a highway reconstruction and expansion safety situation analysis device 600 based on dynamic Bayesian methods. This device 600 is used to implement a highway reconstruction and expansion safety situation analysis method based on dynamic Bayesian methods. The device 600 includes:

[0182] The acquisition module 610 is used to acquire data on the highway reconstruction and expansion projects to be predicted.

[0183] Input module 620 is used to input data into the constructed dynamic Bayesian analysis model.

[0184] The output module 630 is used to obtain, based on data and a dynamic Bayesian analysis model, the predicted results of the evolution of traffic safety status over time for the highway reconstruction and expansion project, the predicted results of safety situation changes, the results of sensitivity analysis of influencing factors, and the results of accident cause diagnosis.

[0185] Optionally, the input module 620 is further used for:

[0186] S21. Obtain risk factors that affect the safety of highway reconstruction and expansion projects.

[0187] S22. Based on the risk factors, construct an indicator system for the risk factors.

[0188] S23. Based on the indicator system, construct the network structure of the dynamic Bayesian network model.

[0189] S24. Calculate the conditional probability distribution of each node in the network structure to obtain a dynamic Bayesian analysis model.

[0190] Alternatively, risk factors include road conditions and traffic conditions themselves.

[0191] The road traffic conditions include the length of the work area, temporary traffic facilities, roadside clearance, and emergency lanes.

[0192] Traffic conditions themselves include traffic composition, traffic flow, speed limits, and traffic conflicts.

[0193] Optionally, the input module 620 is further used for:

[0194] S241. Obtain the linguistic value evaluation of the root node of the network structure and the uncertainty of the linguistic value evaluation based on expert knowledge.

[0195] S242. Based on the linguistic value evaluation and the Gaussian membership function, the linguistic value evaluation is converted into a fuzzy interval value.

[0196] S243. Based on the upper and lower limits of the fuzzy interval values, an interval fuzzy set is introduced. The uncertainty of the language value evaluation is quantitatively represented by the interval fuzzy set to obtain the quantified language evaluation value.

[0197] S244. Based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, the conditional probability distribution of each node in the network structure is obtained.

[0198] Optionally, the input module 620 is further used for:

[0199] S2441. Calculate the conflict coefficient K of classical evidence theory.

[0200] S2442. Perform conflict detection on the conflict coefficient K according to the preset threshold ε.

[0201] S2443. If the threshold ε is greater than or equal to the conflict coefficient K, then the improved hybrid interval evidence synthesis rule and the quantified language evaluation value are used to obtain the conditional probability distribution of each node in the network structure.

[0202] S2444. If the threshold ε is less than the conflict coefficient K, then the classical evidence theory synthesis rules and the quantified language evaluation values ​​are used to obtain the conditional probability distribution of each node in the network structure.

[0203] The construction process of the improved hybrid interval evidence synthesis rule includes: redefining the allocation of conflicting information based on the credibility of the evidence to obtain the improved hybrid interval evidence synthesis rule.

[0204] Optionally, the output module 630 is further used for:

[0205] Identify state transition risk factors among the risk factors; among them, state transition risk factors are risk factors that change over time and have a significant impact on traffic safety status.

[0206] The state transition probability of state transition risk factors is calculated using the Markov chain method.

[0207] Based on the state transition probability, the prediction results of the evolution of traffic safety status of the highway reconstruction and expansion project over time are obtained.

[0208] Optionally, the calculation method for obtaining the predicted results of the safety situation changes of the highway reconstruction and expansion project is as shown in the following formula (1):

[0209] P(R=r)=P(R=r|X1=x1,…,X N =x N )×P(X1=x1,…,x i ,…,X N =x N (1)

[0210] Where P(R=r) is the probability distribution of the security situation R; r={r1,…,r P},{r1,…,r P} represents the P state ranges of the security situation R; Risk factor X i Q i A range of states, i = 1, ..., N; P(R = r|X1 = x1, ..., X N =x N Let P(X1=x1,…,x) be the conditional probability distribution of the security situation R; i ,…,X N =x N Risk factor X i The joint probability distribution.

[0211] Optionally, the output module 630 is further used for:

[0212] The sensitivity performance measure of each risk factor is calculated, and the contribution of the risk factor to the safety situation is measured based on the sensitivity performance measure. Key risk factors are identified based on the contribution, and the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted are obtained.

[0213] Optionally, the output module 630 is further used for:

[0214] Obtain the posterior probability distribution of each risk factor in the highway reconstruction and expansion project to be predicted at the time of the accident; obtain the accident cause diagnosis result of the highway reconstruction and expansion project to be predicted based on the posterior probability distribution.

[0215] In this embodiment of the invention, a DBN model suitable for traffic safety situation analysis of highway reconstruction and expansion projects was constructed to predict changes in traffic safety situation, conduct sensitivity analysis on influencing factors, and diagnose the causes of accidents, providing a scientific basis for taking targeted traffic control measures.

[0216] Figure 7This is a schematic diagram of the structure of an electronic device 700 provided in an embodiment of the present invention. The electronic device 700 can vary considerably due to differences in configuration or performance. It may include one or more central processing units (CPUs) 701 and one or more memories 702. The memory 702 stores at least one instruction, which is loaded and executed by the processor 701 to implement the following method for analyzing the safety situation of highway reconstruction and expansion based on dynamic Bayesian methods:

[0217] S1. Obtain data on highway reconstruction and expansion projects to be predicted.

[0218] S2. Input the data into the constructed dynamic Bayesian analysis model.

[0219] S3. Based on the data and the dynamic Bayesian analysis model, the predicted results of the evolution of traffic safety status over time, the predicted results of safety situation changes, the results of sensitivity analysis of influencing factors, and the results of accident cause diagnosis are obtained for the highway reconstruction and expansion project to be predicted.

[0220] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions that can be executed by a processor in a terminal to complete the aforementioned dynamic Bayes-based highway reconstruction and expansion safety situation analysis method. For example, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device.

[0221] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0222] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic Bayesian-based highway reconstruction safety situation analysis method, characterized in that, The method includes: S1, acquiring data of a highway reconstruction and expansion project to be predicted, the data being data corresponding to risk factors, the risk factors including road traffic conditions and traffic conditions, wherein the road traffic conditions include four influencing factors of work zone length , temporary traffic facilities , roadside clearance , and emergency lane , and the traffic conditions include four influencing factors of traffic composition , traffic flow , speed limit , and traffic conflict ; S2. Input the data into the constructed dynamic Bayesian analysis model; The construction process of the dynamic Bayesian analysis model in S2 includes: S21. Obtain risk factors that affect the safety of highway reconstruction and expansion projects; S22. Based on the aforementioned risk factors, construct an indicator system for the risk factors; S23. Based on the aforementioned index system, construct the network structure of the dynamic Bayesian network model; S24. Calculate the conditional probability distribution of each node in the network structure to obtain a dynamic Bayesian analysis model; The calculation of the conditional probability distribution of each node in the network structure in S24 includes: S241. Obtain the linguistic value evaluation of the root node of the network structure and the uncertainty of the linguistic value evaluation based on expert knowledge; S242. Based on the language value evaluation and the Gaussian membership function, convert the language value evaluation into a fuzzy interval value; S243. Based on the upper and lower limits of the fuzzy interval values, an interval fuzzy set is introduced. The uncertainty of the language value evaluation is quantified based on the interval fuzzy set to obtain the quantified language evaluation value. S244. Based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, obtain the conditional probability distribution of each node in the network structure; The conditional probability distribution of each node in the network structure obtained in step S244, based on the quantified language evaluation value and the improved hybrid interval evidence synthesis rule, includes: S2441. Calculate the conflict coefficient of classical evidence theory. ; S2442, According to the preset threshold Regarding the conflict coefficient Perform conflict detection; S2443, if the threshold Greater than or equal to the conflict coefficient Then, the improved hybrid interval evidence synthesis rule and the quantified language evaluation value are used to obtain the conditional probability distribution of each node in the network structure. S2444, if the threshold Less than the conflict coefficient Then, using the classical evidence theory synthesis rules and the quantified language evaluation values, the conditional probability distribution of each node in the network structure is obtained; The construction process of the improved hybrid interval evidence synthesis rule includes: redefining the allocation of conflicting information based on the credibility of the evidence to obtain the improved hybrid interval evidence synthesis rule; S3. Based on the data and the dynamic Bayesian analysis model, the following results are obtained: the prediction results of the evolution of traffic safety status over time, the prediction results of safety situation changes, the sensitivity analysis results of influencing factors, and the diagnosis results of accident causes for the highway reconstruction and expansion project to be predicted. The predicted results of the traffic safety status evolution of the highway reconstruction and expansion project over time obtained in S3 include: Identify state transition risk factors from the risk factors; wherein, the state transition risk factors are risk factors that change over time and have a significant impact on traffic safety status; The state transition probability of the state transition risk factors is calculated using the Markov chain method. Based on the state transition probabilities, the prediction results of the evolution of traffic safety status of the highway reconstruction and expansion project over time are obtained. The calculation method for obtaining the predicted safety status change of the highway reconstruction and expansion project in S3 is shown in the following formula (1): （1） in, For security situation The probability distribution; , For security situation of A range of states; , Risk factors of A range of states, ; For security situation The conditional probability distribution; Risk factors The joint probability distribution.

2. The method according to claim 1, characterized in that, The sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted obtained in S3 include: Calculate the sensitivity performance measure for each risk factor, measure the contribution of the risk factor to the safety situation based on the sensitivity performance measure, identify key risk factors based on the contribution, and obtain the sensitivity analysis results of the influencing factors of the highway reconstruction and expansion project to be predicted. The accident cause diagnosis results obtained in S3 for the highway reconstruction and expansion project to be predicted include: Obtain the posterior probability distribution of each risk factor in the highway reconstruction and expansion project to be predicted at the time of the accident; obtain the accident cause diagnosis result of the highway reconstruction and expansion project to be predicted based on the posterior probability distribution.

3. A highway reconstruction and expansion safety situation analysis device based on dynamic Bayesian methods, used to implement the method described in claim 1 or 2, characterized in that, The device includes: The acquisition module is used to acquire data on the highway reconstruction and expansion project to be predicted. This data includes data corresponding to risk factors, which include road conditions and traffic conditions. Road conditions include the length of the work area. Temporary traffic facilities Roadside clearance and emergency lane Four influencing factors, including the traffic conditions themselves and the traffic composition. Traffic flow Speed ​​limit and traffic conflicts Four influencing factors; The input module is used to input the data into the constructed dynamic Bayesian analysis model; The output module is used to obtain, based on the data and a dynamic Bayesian analysis model, the predicted results of the evolution of traffic safety status over time for the highway reconstruction and expansion project, the predicted results of safety situation changes, the results of sensitivity analysis of influencing factors, and the results of accident cause diagnosis.

Images