Research on Safety Risk Assessment Technology of Expressway Traffic Operation

Based on the theory of expressway traffic safety and the analysis of accident causes, the risks of expressway traffic operation safety is identified, and a multilevel traffic safety risk assessment system is established. Combined with the advantages of traditional risk qualitative and quantitative analysis methods, the Fuzzy Analysis Theory is introduced in the Analytic Hierarchy Process, the Analytic Hierarchy Process (AHP) is established to weight each related index, and finally the Fuzzy Analytical Hierarchy Process (FAHP) is applied to estimate the risk probability and risk loss. And then according to the risk probability, risk loss assessment and its grade standard, the two-dimensional risk assessment matrix is used to measure the traffic operation risk level and give basic risk management and control criteria. And finally based on the above evaluation system, the traffic safety risk of an important bridge across the Yellow River on the expressway from Beijing to Shanghai is analyzed and assessed.


Analysis and Identification of Influencing Factors of Expressway Operation Safety Risks
Risk identification is the basis of risk assessment and control management. Expressway traffic operation risk identification usually needs to solve two basic problems: one is what are the risk factors affecting expressway traffic operation, and another one is how to describe the characteristics of these risk factors in a comprehensive and detailed way and their adverse consequences on expressway traffic operations.
Commonly used risk identification methods in the field of risk management include Statistical analysis of related data, Safety checklist method, Brainstorming method, Delphi method, Work task analysis method and Fault tree analysis method, or the combination of two or more of them. [3] However, based on the unique attributes of the expressway operation industry and the relative lack of historical data on operational risk management, the problem of how to find relatively independent key risk factors from so many risk factors requires a more targeted risk identification method.
Based on the documentation research and the investigation of the safety risk management and control requirements of expressway operators, the risk factor identification method combining objective safety risk accident data analysis and subjective expert analysis method is adopted to break the limitation of data analysis and subjective analysis of risk factor and thus to improve the rationality of risk factor identification. At the same time, the risk factor identification method emphasizes the dynamic identification of risk factors and follows the rule that risks change with time and objective conditions.
Based on the statistical analysis of the influencing factors of the traffic safety risk accident case and the questionnaire from traffic safety management experts, we obtained the 15 expressway traffic safety risk factors index as shown in Figure 1.

Expressway Operation Safety Risk Assessment
The main task of risk assessment is to analyze the risk probability and risk loss of target indicators based on risk identification results, which is the most complicated and critical core step in the whole risk control technology process. The analysis methods of risk probability and loss are mainly divided into Qualitative methods (Expert investigation method) and Quantitative methods (Probabilistic analysis method, Decision tree method, Monte Carlo stochastic simulation method and Fuzzy comprehensive analysis method), and the result of traffic safety risk analysis with the quantitative method are more in line with the decision-making needs of expressway managers. [4][5]Probabilistic analysis method is difficult to implement due to the limited statistics on safety accidents. And at the same time, traffic safety risk assessment focuses on the impact of potential risks on traffic safety, so it is not necessary to dig deep into the logical relationship between influencing factors and to apply the Fault tree method. The combination of Analytic Hierarchy Process and Fuzzy Comprehensive Evaluation (AHP-FCE) is more suitable for the study of risk factor impact grading and risk decision hierarchy [6], and considering the uncertainty of risk analysis, it is also suitable for expressway traffic safety risk assessment. The main steps of this method include: establishing a hierarchical model of assessment index, constructing a risk index review set, establishing a hierarchy of evaluation index weights, establishing a membership matrix and a multilevel comprehensive evaluation algorithm.

3.1The establishment of Hierarchical model of assessment index
Establishing a multilevel assessment model is one of the main steps of the Analytic Hierarchy Process, and most of its work has been finished in the process of risk identification. Based on the identification of expressway traffic safety risk sources, the target index U of expressway traffic operation safety risk is divided into m sub-indicators U i , i=1, 2, 3, ..., m, and the corresponding first-level indicator set U=( U1 , U 2 , U 3 ..., U m ). And for each first-level indicator, it can also be further divided into n second-level indicators U ij , j=1, 2, 3,...,n, and the corresponding secondary indicator set U i = ( U 1 i , U i2 , U i3 ,..., U ij ). In this way, the risk index is divided into two levels, and a three-level highway operation safety risk assessment index system is established as shown in Figure 1.

3.2The Construction of Risk index review set
Risk index review set is a grading standard used to determine the level of each risk indicator in each risk factor group, V = ( V 1 , V 2 , V 3 ,... , V k ), and vk represents different risk probabilities or loss levels. In the expressway operation safety risk assessment, it is more reasonable to divide the risk probability and loss into five grades (k=5) according to the cognitive law, that is, extremely low, low, medium, high, and extremely high.In order to better measure the final risk assessment results, it is necessary to give a risk probability and risk lossrating criteria. Tables 1 and 2 give a description of the different risk probabilities and risk loss levels [7].  More than ten people die; property losses are more than 30 million yuan; major damage to key road facilities (bridges, tunnels and etc.), requiring long-term closed maintenance; the operator take the primary responsibility, and it causes social repercussions and it damages corporate reputation; permanent and serious impact on environment.

Establishment of a Hierarchy of evaluation index weights
Weight analysis is another important content of the Analytic Hierarchy Process method. The hierarchy of evaluation index weights is a standard that indicates the importance of each level of risk index in each set of factors, and its establishment process includes: the discriminant matrix construction of the hierarchical index priority relationship, the fuzzy consistency discriminant matrix construction and the weight level calculation.

Construction of Fuzzy consistency matrix.
The judgment matrix of the medium-transitiveness of the fuzzy consistency matrix is more suitable for the mental characteristics of people's decision-making thought, and the fuzzy consistency judgment matrix is effective in its function. The introduction of the fuzzy consistency judgment matrix can avoid the consistency check procedure caused by the subjectivity of judgment. Many scholars have studied and proved the consistency of fuzzy judgment matrixes , see Li Yong, Hu Xianghong [9], Zhang Sen, et al. [10]. The fuzzy consistent judgment matrix A=( Ai j )n×n, where aij is a priority judgment matrix.

Calculation of the Weight coefficient.
There are several methods for solving the weight coefficient vector from the relational judgment matrix [8]. In this paper, the weight consistency judgment matrix A is applied to calculate the weight based on the row normalization method. Firstly, the elements The weight coefficient vector W of the first-level indicator set U can be calculated by the same method. For some specific hierarchical structures, the weight of the first-level indicator can also be directly specified through expert investigation opinions.

Establishment of Membership matrix
The Membership matrix is a matrix used to describe the degree of membership of each risk index for each level of the comment set, and it is the basis of the evaluation result. from the index feature values through the membership function [11]; but for most systemic risk analysis problems, the index factors are often difficult to quantify, so it can be obtained by statistical means using Expert survey methods. Table 3 gives the results of the Expert survey of the driver's factor membership probability for the driver's indicator set U1 = ( U11 , U12 , U13 ).

Multilevel comprehensive evaluation
The multi-level comprehensive evaluation refers to the process of calculating the evaluation sets of each level from the bottom layer to the target layer based on the obtained membership degree matrix and weight set, and finally obtaining the risk estimation result of the target indicators.

Comprehensive evaluation of secondary index.
From the second-level index weight set and the membership degree matrix Ri , the second-level index evaluation set Bi can be obtained: In the formula: "•" is a fuzzy operator. In this paper , the Zadeh operator, which is determined by the main factor, highlights the influence of the main factors and ignores other secondary factors [12].
The Dimensional normalization of the Bi .

Comprehensive evaluation of primary indicators.
From the second-level index evaluation set Bi , the first-level membership degree matrix R can be obtained.
Considering the first-level index weight set W, the first-level index evaluation set B can be obtained.

Comprehensive assessment of target index.
For some complex problems, the hierarchical model may have multiple intermediate layers. According to the above steps, the assessment is performed from the bottom layer to the upper layer, and the evaluation set B can be obtained. The score set G = (0.1, 0.3, 0.5, 0.6, 0.9) is established from the middle value of the comment set V score interval, and the risk estimation result D of the target index U can be obtained as in the formula (12).

Expressway Operation Safety Risk Assessment and Control
Based on the risk identification and estimation, the risk assessment used the comprehensive risk probability estimation result and the risk loss estimation result, combined with the corresponding risk assessment model, comprehensively evaluates the risk, determine the overall level and severity level of the system risk. And finally gives corresponding risk control measures. The expressway operation safety risk assessment model applies to a two-dimensional evaluation model, as shown in Table 4, and Table 5 gives the control measures for different risk levels. 0.4~0.6 Ⅰ Ⅱ Ⅱ Ⅲ Ⅲ 0.6~0.8 Ⅰ Ⅱ Ⅲ Ⅲ Ⅳ 0.8~1.0 Ⅱ Ⅲ Ⅲ Ⅳ Ⅳ 0.2~0.4 Ⅲ Ⅲ Ⅳ Ⅳ Ⅳ Not expected High risk, must take risk response and monitoring measures to reduce risk level Ⅳ Cannot be accepted Extremely high risk, must attach great importance to risk aversion, otherwise should reduce risk to undesired levels at all cos

Case Analysis
A Yellow River Highway Bridge, as a prime large bridge on the Beijing-Shanghai expressway, is an critical cross section of the Beijing-Shanghai and Beijing-Taibei Expressway, with a total length of 5,750 meters, a main bridge length of 947.66 meters, and an approach bridge length of 4,152.98 meters. It is 35.5 meter in width, 6 lanes in both directions, and it has a design speed of 120 km/h. It is an expressway bridge with high design standards and large building scale on the Yellow River. At present, it has more than 67,600 traffic flows. With such a huge traffic flow, it always causes frequent traffic safety accidents,