Analysis Method of Risk Factors of Urban Public Transport Accidents

Abstract

The present invention belongs to the technical field of urban public transportation safety monitoring, and specifically relates to a method for analyzing risk factors of urban public transportation accidents. The public transportation accident information processing system used in the method includes a central processing unit, a data conversion module, a database, a clock module, a gateway module, an information display module, an Internet module, a satellite positioning module, an encoder and an automatic camera integrated and installed on the smart light pole of the urban road; the method includes data collection and preprocessing, selecting risk factors from four aspects of people, vehicles, roads and environment to determine the characteristics of public transportation accidents, establishing a logistic model of public transportation accidents, determining the significant factors affecting urban public transportation safety, and using the Apriori algorithm to associate the risk factors of public transportation accidents. The present invention can quickly and accurately analyze and evaluate the risk factors of urban public transportation accidents.

Description

Technical Field

[0001] The invention belongs to the technical field of urban public transportation safety monitoring, and in particular relates to a method for analyzing risk factors of urban public transportation accidents. Background Art

[0002] The master's thesis "Urban Traffic Accident Risk Analysis Method Based on Dynamic Fault Tree" written by Ma Yixin, a graduate student of Harbin Institute of Technology, conducts risk analysis research on urban road traffic accidents based on dynamic fault tree theory. The basic content of the urban traffic accident risk analysis method disclosed in the thesis is as follows:

[0003] (1) The original accident data was sorted and counted in an orderly manner. According to the distribution characteristics of the accident location, type and form, the research object, main accident type and main accident form were determined. At the same time, the direct causes recorded in the accident identification book were counted, and the basic probability was given. Based on the binomial distribution statistical test, the potential factors with weak correlation were eliminated from the four aspects of people, vehicles, roads and environment, and the prominent potential influencing factors with significant influence were identified. The main influencing factors and prominent potential influencing factors were used as the risk factors of urban road traffic accidents, laying the foundation for the establishment of subsequent models.

[0004] (2) Construct an effective risk analysis model to conduct risk analysis on traffic accidents. First, based on the comparison of the advantages and disadvantages and applicable scope of conventional risk analysis methods, the dynamic fault tree method is selected in a targeted manner. Then, based on the mechanism of accident occurrence, a dynamic fault tree model of urban road traffic accidents is constructed.

[0005] (3) The model is preprocessed and simplified, and then the simplified model is modularized and decomposed to identify all module subtrees. For the static subtree, the corresponding ITE structure expression is written by recursion, and the subtree is converted into a binary decision diagram to determine the minimum cut set and the probability of occurrence. For the dynamic subtree, it is converted into a Markov chain based on the Markov process, and then its failure mode is determined and the probability importance is calculated. Then each subtree is modularly synthesized to obtain the main accident cause chain.

[0006] (4) The Bow-tie model is introduced to describe the whole process from the cause of the accident to the occurrence of the accident and then to the consequences of the accident. First, the event tree model is constructed by analyzing the process of the consequences of the accident. The fuzzy probability of each event chain is determined by using the expert scoring method combined with triangular fuzzy numbers. Combined with the risk matrix, the high-risk event chain is obtained. The main accident cause chain in the fault tree is connected with the high-risk event chain in the event tree to construct the Bow-tie model. The left side of the Bow-tie model proposes corresponding prevention and solutions for the five main accident cause chains; the right side of the Bow-tie model proposes targeted accident consequence response plans based on the two high-risk event chains.

[0007] The above methods play a positive role in the analysis and evaluation of urban traffic accident risks, but they also have certain defects, which are mainly manifested in the following three aspects:

[0008] (1) The data related to urban traffic accidents targeted by the above methods are mainly historical data, and lack the collection and processing of real-time data.

[0009] (2) Encoding is difficult to implement during data processing, especially complex protocol configuration and parsing.

[0010] (3) With regard to the specific issue of analyzing the risk factors of urban bus accidents, the above methods are not targeted enough. Summary of the invention

[0011] The purpose of the present invention is to provide an efficient and convenient analysis method that can timely reflect the dynamic changes of urban public transportation accident risk factors, thereby overcoming the above-mentioned defects of the prior art. The purpose of the invention is achieved through the following technical solutions:

[0012] A method for analyzing risk factors of urban public transportation accidents. The public transportation accident information processing system used in the method comprises a central processing unit, a data conversion module, a database, a clock module, a gateway module, an information display module, an Internet module, a satellite positioning module, an encoder and an automatic camera which are integrated and installed on a smart light pole on an urban road; the satellite positioning module and the encoder are connected to the central processing unit through the data conversion module; the wired Internet module and the information display module are connected to the central processing unit through the gateway module, and the database, the clock module and the automatic camera are directly connected to the central processing unit; the central processing unit is connected to the control center of the urban public transportation management department through the gateway module and the Internet module; the method comprises the following steps:

[0013] Step 1: Collect road traffic accident data information related to buses, create a sample data set based on relevant risk factors and accident types, and perform data preprocessing;

[0014] Step 2: Select risk factors from four aspects: people, vehicles, roads, and environment, and use correlation statistical analysis to intuitively analyze the impact of each risk factor, so as to determine the characteristics of bus accidents;

[0015] Step 3: Using the risk factors in the sample data set as independent variables and the accident types as dependent variables, a logistic model of public transportation accidents is established to determine the significant factors affecting urban public transportation safety;

[0016] Step 4: Use the Apriori algorithm to correlate the risk factors of public transportation accidents and explore the impact of the correlation between the risk factors of people, vehicles, roads, and environment on the types of public transportation accidents;

[0017] Step 5: Combine the impact of single factors on bus accident types analyzed by the Logistic regression model and the impact of the correlation of multiple factors on accident types mined by the Apriori algorithm to analyze the risk factors affecting urban bus safety.

[0018] The basic concept of the present invention is to install a bus accident information processing system on smart lamp posts on both sides of urban roads (especially accident-prone areas or intersections) or on roundabouts to collect, analyze and process bus accident information in real time, combine the historical information of bus accidents stored in the control center of the urban bus management department, use the Logistic model and Apriori algorithm to quickly and accurately analyze and evaluate the risk factors of urban bus accidents.

[0019] On the basis of the above technical solution, the present invention can add the following technical means to better achieve the purpose of the present invention:

[0020] When executing step 1, information on personnel characteristics, vehicle characteristics, road conditions, accident time, environmental information and accident type is obtained through data screening.

[0021] Furthermore, when executing step 2, the current status of bus accidents is analyzed from four aspects: time of accident, location, driver, road and environment, so as to understand the occurrence and development trend of bus accidents as a whole and discover the spatiotemporal distribution characteristics of accidents.

[0022] Furthermore, when executing step 3, when executing step 3, variables in the risk factors of driver attributes, vehicle, road and environment are selected from the bus accident data as independent variables, and the accident type is used as the dependent variable. The accident type is divided into four categories: no casualties, minor injuries, serious injuries and death.

[0023] Furthermore, step 3 includes step 301 and step 302. Step 301 uses an ordered multivariate logistic regression model to establish the relationship between the accident type and each variable, thereby identifying the risk factors that affect the safety of urban public transportation operations; step 302 uses Matlab software to solve the model and solve the regression coefficient of each variable. .

[0024] The present invention has the following beneficial effects:

[0025] (1) Using existing smart light poles (such as the smart light poles described in the invention patent application specification of CN112378373A) to set up a public transportation accident information processing system can not only collect, analyze and process urban public transportation accident information in real time, but also has low implementation costs.

[0026] (2) By establishing a logistic model of bus accidents and adopting the Apriori algorithm, the present invention greatly reduces the difficulty of coding implementation in the process of urban bus accident data processing, effectively solves the configuration and analysis of complex protocols in the process of urban bus accident data processing, and can quickly and accurately analyze and evaluate the risk factors of urban bus accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 The figure is a structural block diagram of a public transportation accident information processing system in one embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to facilitate those skilled in the art to understand the technical solution of the present invention, an embodiment of the present invention is described below in conjunction with the accompanying drawings.

[0029] like Figure 1 As shown, the bus accident information processing system used in the urban bus accident risk factor analysis method of the present invention includes a central processing unit, a data conversion module, a database, a clock module, a gateway module, an information display module, an Internet module, a satellite positioning module, an encoder and an automatic camera integrated and installed on the smart light pole of the urban road; the satellite positioning module and the encoder are communicated with the central processing unit through the data conversion module; the wired Internet module and the information display module are communicated with the central processing unit through the gateway module, and the database, the clock module and the automatic camera are directly communicated with the central processing unit; the central processing unit is communicated with the control center (not shown in the figure) of the urban bus management department through the gateway module and the Internet module.

[0030] The method for analyzing risk factors of urban public transportation accidents of the present invention comprises the following steps:

[0031] Step 1: Collect road traffic accident data information related to buses, screen and obtain information such as personnel characteristics, vehicle characteristics, road conditions, accident time, environmental information and accident type in the data, and make a sample data set according to relevant risk factors and accident types, and perform data preprocessing.

[0032] Step 2: Select risk factors from four aspects: people, vehicles, roads, and environment, and use correlation statistical analysis to intuitively analyze the impact of each risk factor, so as to determine the characteristics of bus accidents.

[0033] Step 3, select the variables of the risk factors of driver attributes, vehicles, roads and environment from the bus accident data as independent variables, take the accident type as the dependent variable, establish the bus accident logistic model, and determine the significant factors affecting the safety of urban public transportation; the accident type is divided into four categories: no casualties, minor injuries, serious injuries and death. Step 3 includes step 301 and step 302. Step 301 uses an ordered multivariate logistic regression model to establish the relationship between the accident type and each variable, so as to identify the risk factors affecting the safety of urban public transportation operations; step 302 uses Matlab software to solve the model and solve the regression coefficient β of each variable.

[0034] Step 4: Use the Apriori algorithm to correlate the risk factors of public transportation accidents and explore the impact of the correlation between the risk factors of people, vehicles, roads, and environment on the types of public transportation accidents;

[0035] Step 5: Combine the impact of single factors on bus accident types analyzed by the Logistic regression model and the impact of the correlation of multiple factors on accident types mined by the Apriori algorithm to analyze the risk factors affecting urban bus safety.

[0036] The above-mentioned road traffic accident data information includes the historical information of public transportation accidents stored in the control center of the urban public transportation management department and the information collected in real time by the public transportation accident information processing system. The risk factor analysis of public transportation accidents in local sections of the city is completed by the central processor in the public transportation accident information processing system, and the risk factor analysis of public transportation accidents in the city as a whole is completed by the control center of the urban public transportation management department. During this process, the control center of the urban public transportation management department communicates information with the central processor in the public transportation accident information processing system.

[0037] The above describes the public transportation accident information processing system and basic working steps used in one embodiment of the present invention. The following further describes the data processing method used in this embodiment.

[0038] When executing step 301, the first or last category is generally selected as the base category. If the Kth category is used to represent the base category, the multivariate logistic regression model can be represented as K-1 binary logistic regression models:

[0039]

[0040]

[0041] …

[0042] (1)

[0043] In the formula, For the i explanatory variables for observations, including J explanatory variables, the parameters of the k-th category Logistics regression model are ,only J + 1 parameter, the first of which is the intercept term. From this we can see that the coefficients of the multinomial logistic regression model The power index It is an important indicator to measure the influence of explanatory variables on dependent variables. It is the ratio of the probability of an event occurring and the probability of it not occurring, and the relative risk can be judged based on it. Explanation: Under the condition of controlling other explanatory variables, the explanatory variable The impact of the unit change ratio on the occurrence ratio of the category and the benchmark category. The occurrence ratio indicates the change in the distribution probability of each type when the influencing factor does not increase or decrease by one unit, that is, < 1, the occurrence ratio is reduced; =1 Occurrence ratio remains unchanged; > 1 the occurrence ratio increases.

[0044] When executing step 302 (using Matlab software to solve the model), the exponential transformation is performed on equation (1), and the probability of each category can be expressed by the probability of the benchmark category, that is:

[0045] (2)

[0046] Since the sum of the probabilities of all categories is 1, the probability of the benchmark category can be obtained from formula (2):

[0047] (3)

[0048] Substituting the probability of the benchmark category in equation (3) into equation (2), we can obtain the probability of each category:

[0049] (4)

[0050] The multinomial Logistic regression model is solved using the maximum likelihood method, and its likelihood function is expressed as:

[0051]

[0052] The parameter estimates can be obtained as:

[0053] When executing step 4, the present invention uses the Apriori algorithm. The Apriori algorithm is a classic method for association rule data mining. The algorithm uses layer-by-layer search iteration and uses k-item sets to explore (k+1)-item sets. First, determine the geometry of the frequent 1-item set, denoted as L 1, use L1 to determine the set L of frequent 2-item sets 2 , using L 2 Determine L 3 , ..., and so on, iterating layer by layer until no frequent k-item sets can be found. Each iteration determines L k A database scan is required. All non-empty subsets of frequent itemsets must also be frequent, which can be used to compress the search space, thereby improving the efficiency of generating frequent itemsets layer by layer. Secondly, the Apriori algorithm needs to perform three specific steps of connection, pruning and generating association rules to generate association rules. Therefore, in this embodiment, step 4 specifically includes the following three steps:

[0054] Step 401, connect, to find L k (All frequent k Itemsets), by L k-1 (All frequent k -1 item set) is connected with itself to generate candidate k The set of item sets, the candidate set is denoted by C k .set up l 1 and l 2 Yes Yes L k-1 Members of l i [ j ] express l i The j Items. Assume that the Apriori algorithm sorts the items in the transaction or item set in lexicographic order, that is, for ( k -1) Itemset l i , l i [1] < l i [2] < l i [ k -1], L k-1 Connect to itself, if ( l 1 [1] = l 2 [1])&&( l 1 [2] = l 2 [2])&&…&&( l 1 [k -2] = l 2 [ k -2])&&( l 1 [ k -1] = l 2 [ k -1]), then it is considered l 1 and l 2 is connectable. Connect l 1 and l 2 The result is { l 1 [1], l 1 [1], … , l 1 [1], l 1 [1]}

[0055] Step 402, pruning, C k yes L k is a superset of C k The members of may or may not be frequent. By scanning all transactions, determine C k The count of each candidate in is judged whether it is less than the minimum support count. If it is greater than the minimum support count, the candidate member is considered frequent.

[0056] Step 403, for each frequent item set l ,produce l For all non-empty subsets of l Every non-empty subset of s ,if , then the output rule .in, Is the minimum confidence threshold. When using the Apriori algorithm for data mining, you need to set a minimum confidence and minimum support first. The confidence and support of the rule are both between 0% and 100%. You need to focus on debugging these two parameters to achieve the goal of reducing the number of rules.

[0057] The above describes the technical solution of an embodiment of the present invention. The following two tables further describe the results obtained when the present invention was tested in Chancheng District, Foshan City:

[0058] Table 1 shows the results of the Logistic model

[0059]

[0060] Table 2 shows the combined results of factors affecting public transport accidents

[0061]

Claims

1. A method for analyzing risk factors of urban public transportation accidents. Features: The public transportation accident information processing system used in the method includes a central processing unit, a data conversion module, a database, a clock module, a gateway module, an information display module, an Internet module, a satellite positioning module, an encoder and an automatic camera which are integrated and installed on a smart light pole on a city road; the satellite positioning module and the encoder are connected to the central processing unit through the data conversion module; the Internet module and the information display module are connected to the central processing unit through the gateway module, and the database, the clock module and the automatic camera are directly connected to the central processing unit; the central processing unit is connected to the control center of the city public transportation management department through the gateway module and the Internet module; the method includes the following steps: Step 1: Collect road traffic accident data information related to buses, including historical bus accident information stored in the control center of the urban bus management department and information collected in real time by the bus accident information processing system, and make a sample data set according to relevant risk factors and accident types, and perform data preprocessing; Step 2: Select risk factors from four aspects: people, vehicles, roads, and environment. Use correlation statistical analysis to intuitively analyze the impact of each risk factor. The risk factor analysis of bus accidents on local sections of the city is completed by the central processor in the bus accident information processing system. The risk factor analysis of bus accidents in the city as a whole is completed by the control center of the city bus management department. In this process, the control center of the city bus management department communicates with the central processor in the bus accident information processing system to determine the characteristics of the bus accident. Step 3: Using the risk factors in the sample data set as independent variables and the accident types as dependent variables, a logistic model of public transportation accidents is established to determine the significant factors affecting urban public transportation safety; Step 4: Use the Apriori algorithm to correlate the risk factors of public transportation accidents and explore the impact of the correlation between the risk factors of people, vehicles, roads, and environment on the types of public transportation accidents; Step 5: Combine the impact of single factors on bus accident types analyzed by the Logistic regression model and the impact of the correlation of multiple factors on accident types mined by the Apriori algorithm to analyze the risk factors affecting urban bus safety.

2. The method for analyzing risk factors of urban public transportation accidents according to claim 1, Features: When executing step 1, information on personnel characteristics, vehicle characteristics, road conditions, accident time, environmental information and accident type is obtained through data screening.

3. The method for analyzing risk factors of urban public transportation accidents according to claim 1, Features: When executing step 2, the current status of bus accidents is analyzed from four aspects: time, location, driver, road and environment of the accident, so as to understand the occurrence and development trend of bus accidents as a whole and discover the spatiotemporal distribution characteristics of the accidents.

4. The method for analyzing risk factors of urban public transportation accidents according to claim 1, Features: When executing step 3, variables in the risk factors of driver attributes, vehicle, road and environment are selected from the bus accident data as independent variables, and the accident type is used as the dependent variable. The accident types are divided into four categories: no casualties, minor injuries, serious injuries and death.

5. The method for analyzing risk factors of urban public transportation accidents according to claim 4, Features: Step 3 includes step 301 and step 302. Step 301 uses an ordered multivariate logistic regression model to establish the relationship between the accident type and each variable, so as to identify the risk factors affecting the safety of urban public transportation operations; step 302 uses Matlab software to solve the model and solve the regression coefficient β of each variable.

Images