Comprehensive risk assessment method and electronic equipment for highway freight channels

Abstract

The present invention discloses a comprehensive risk assessment method and electronic device for highway freight corridors, comprising the following steps: S1, obtaining raw trajectory data, road network data, and other data including OD data, China's administrative division map base map data, and weather data, and preprocessing them to obtain processed data; S2, identifying highway freight corridors based on the processed data; S3, selecting evaluation indicators and establishing a highway freight corridor risk assessment indicator system; S4, establishing a highway freight corridor risk assessment model based on the highway freight corridor risk assessment indicator system; and S5, using the established highway freight corridor risk assessment model to identify corridor risks. The present invention can utilize existing objective data to determine highway risks in advance, focus on monitoring high-risk freight corridors, and prevent accidents; provide safety warnings and references to drivers; and add safety-related impacts to the operation of freight information platforms.

Description

Technical Field

[0001] The present invention belongs to the technical field of traffic safety, and specifically includes a comprehensive risk assessment method for a highway freight channel and electronic equipment. Background Art

[0002] Road transport has long been the primary form of freight transportation in my country. Compared to other vehicles, freight vehicles carry more cargo, have greater inertia, require longer driving times, and carry potentially hazardous materials, all of which undoubtedly increase the risks of road freight. Truck accidents can seriously endanger people's lives and property and cause significant economic losses to society. Nowadays, through freight information platforms, truck drivers can easily access cargo information, recommended routes to different destinations, travel times, and tolls using mobile apps, allowing them to estimate the time and financial costs of different routes. However, there is currently no solution that allows truck drivers to identify freight corridors and assess their overall risk over a specific period of time.

[0003] With the development of information technology in the logistics industry, 76.5% of my country's drivers have joined online freight information platforms. To better monitor and analyze truck drivers' behavior, these platforms have accumulated vast amounts of data. By combining this data with road network data for mining, processing, and analysis, a wealth of information related to freight risks can be obtained, opening up the possibility of using big data to analyze risks along highway freight corridors.

[0004] Currently, research on highway freight safety using truck big data primarily focuses on the impact of truck driver behavior on freight safety. Analyzing driver behavior through trajectory data and sensor data, and determining whether standard driving behavior impacts freight safety, constitutes a study of subjective driver factors. Regardless of whether the driver's actions are in compliance with regulations, objective factors such as road conditions, traffic flow, the impact of other vehicles on the road, and weather conditions can all impact highway freight safety. Assessing the risks of highway freight corridors is a common challenge faced by transportation management departments, freight information platforms, and truck drivers. Summary of the Invention

[0005] In response to the above-mentioned deficiencies in the prior art, the present invention provides a comprehensive risk assessment method and electronic equipment for highway freight channels, which solves the problem of difficulty in judging the risk of highway freight channels due to objective factors such as road conditions and the impact of traffic flow.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a comprehensive risk assessment method for highway freight corridors, comprising the following steps:

[0007] S1, obtaining original trajectory data, road network data and other data including OD data, China administrative division map base map data and weather data, and pre-processing them to obtain processed data;

[0008] S2, identifying highway freight corridors based on the processed data;

[0009] S3. Select evaluation indicators and establish a risk evaluation indicator system for highway freight corridors;

[0010] S4. Establish a highway freight channel risk assessment model based on the highway freight channel risk assessment indicator system;

[0011] S5. Use the established highway freight channel risk assessment model to identify channel risks.

[0012] Furthermore, the specific implementation method of step S1 is as follows:

[0013] S1-1. Obtain original trajectory data, road network data, and other data including OD data, China administrative division map base map data, and weather data;

[0014] S1-2: Consider the Earth as a sphere and calculate the geographic distances associated with the original trajectory data. Perform reverse geocoding on the longitude and latitude of the trajectory points by calling the API provided by AutoNavi, converting the longitude and latitude information of the trajectory points into text-formatted addresses. Perform duplicate data removal, abnormal data processing, and data range coordinate conversion and filtering on the acquired original trajectory data to complete preprocessing of the original trajectory data and obtain processed trajectory data.

[0015] S1-3, pre-processing the road network data by screening the geographical range and the road grade to obtain processed road network data;

[0016] S1-4. The processed trajectory data, the processed road network data and other data are combined into processed data.

[0017] Furthermore, the specific implementation of step S2 is as follows:

[0018] S2-1. Use the kernel density analysis tool to obtain the distribution of truck track points and the patterns of truck activities;

[0019] S2-2, correcting the processed trajectory data and matching the processed trajectory data with the road network data to obtain a road network matching result;

[0020] S2-3. Based on the road network matching results, the road network data of the trajectory points are sorted according to the frequency of the trajectory points' activities on different roads, the node cities for highway freight are determined, and two or more roads in the node cities are selected as highway freight channels.

[0021] Furthermore, the specific steps for matching the processed trajectory data with the road network data are as follows:

[0022] S2-2-1. Set up a buffer zone with a radius of 50m.

[0023] S2-2-2. Use the cropping tool to crop the data outside the road network buffer;

[0024] S2-2-3, perform nearest neighbor analysis on the processed trajectory data and road network data;

[0025] S2-2-4. Perform table association between the road network serial number of the newly added trajectory point in the processed trajectory data and the road network serial number in the road network data, thereby completing the matching between the road network data and the processed trajectory data.

[0026] Furthermore, the specific implementation of step S3 is as follows:

[0027] S3-1, according to the formula:

[0028]

[0029] Get the corresponding road level risk index of the kth trajectory point h k is the proportion of the total number of accidents of the road grade to which the kth trajectory point belongs in the highway freight channel; p k is the single accident death rate of the road grade to which the kth trajectory point belongs in the highway freight channel; l k is the total mileage of the road grade to which the kth trajectory point belongs in the highway freight corridor;

[0030] S3-2, according to the formula:

[0031]

[0032] The road grade risk index x1 in the highway freight corridor is obtained, where n is the number of trajectory points in the corresponding highway freight corridor per unit time;

[0033] S3-3, according to the formula:

[0034]

[0035] Get the special section index x2 of the highway freight channel, The kth trajectory point corresponds to the special road section risk

[0036] score;

[0037] S3-4, according to the formula:

[0038]

[0039] Obtain the congestion risk index x3 of the highway freight channel, Score the congestion situation of the k-th trajectory point;

[0040] S3-5, according to the formula:

[0041]

[0042] Get the speeding index x4 of the highway freight channel; Score the speeding situation of the k-th trajectory point;

[0043] S3-6, according to the formula:

[0044]

[0045] Get the highway freight channel overload ratio index x5, is the overload ratio score corresponding to the OD data when both the mth departure city and the arrival city are cities passing through the highway freight corridor; n′ is the total number of OD data in the highway freight corridor;

[0046] S3-7, according to the formula:

[0047]

[0048] Get the dangerous goods transport index x6 of the highway freight channel, Indicates the OD data dangerous goods transport score when both the mth departure city and the arrival city are cities that the highway freight corridor passes through;

[0049] S3-8, according to the formula:

[0050]

[0051]

[0052] Get driving time indicators for highway freight corridors h k represents the freight activity distribution value corresponding to the moment of the k-th trajectory point; s k represents the traffic accident distribution value corresponding to the time of the k-th trajectory point; x7 represents the driving period index of the highway freight channel;

[0053] S3-9, according to the formula:

[0054]

[0055] Get the driving time index of the highway freight channel x8, It represents the freight time value in the OD data when both the mth departure city and the arrival city are cities passed by the highway freight corridor within a unit time;

[0056] S3-10: Set the temperature impact risk score of trajectory points with a minimum temperature below 0°C and a maximum temperature above 40°C to 1, and set the temperature impact risk score of trajectory points with an all-day temperature between 0°C and 40°C to 0;

[0057] S3-11, according to the formula:

[0058]

[0059] Get the weather impact index x9 of the highway freight channel, represents the weather impact risk score corresponding to the k-th trajectory point in the highway freight corridor;

[0060] S3-12, according to the formula:

[0061]

[0062] Get the positive indicator value after deviation standardization of the processed data Set the a-th highway freight channel as the a-th evaluation object; x ab Represents the values ​​of x1, x2...x9, that is, the value of the bth indicator in the ath evaluation object, 1≤a≤g,1≤b≤s, min b is the minimum value of the b-th indicator, max b is the maximum value of the bth indicator, s means there are s indicators in total, and g means there are g evaluation objects in total;

[0063] S3-13, according to the formula:

[0064]

[0065]

[0066]

[0067]

[0068] Perform z-score standardization on the indicator data to obtain a standard indicator with a variance of 1 and a mean of 0 The indicator data is processed by z-score standardization to obtain the dimensionless matrix X * ; where x ij represents the value of the jth indicator at the i-th evaluation object, 1≤i≤g,1≤j≤s; is the mean of the jth indicator; σ jis the standard deviation of the j-th indicator.

[0069] Furthermore, the specific implementation of step S4 is as follows:

[0070] S4-1. Perform min-max normalization on the indicator data and record the normalized data matrix as Using entropy method Calculate the risk assessment value and obtain the comprehensive score of the risk assessment;

[0071] S4-2. Based on the comprehensive score of risk assessment, the dynamic risk level of the channel is defined as "low risk", "medium-low risk", "medium risk", "medium-high risk" and "high risk";

[0072] S4-3, perform principal component analysis on the evaluation indicators to obtain principal component scores;

[0073] S4-4. Using the results of principal component analysis as the input matrix of the BP neural network and the comprehensive score of risk assessment obtained by the entropy method as the training target matrix, the BP neural network is trained to obtain a trained BP neural network;

[0074] S4-5. Use the trained BP neural network as a risk assessment model for highway freight channels.

[0075] Furthermore, the calculation method of the entropy method in step S4-1 is as follows:

[0076] S4-1-1, according to the formula:

[0077]

[0078] Get the proportion p of the ath evaluation object under the bth evaluation index ab ; Among them, c means there are c evaluation objects in total, and d means there are d evaluation indicators in total;

[0079] S4-1-2, according to the formula:

[0080]

[0081] Get the entropy value e of the bth evaluation index b ;in,

[0082] S4-1-3, according to the formula:

[0083] f b =1-e b

[0084] Get the information entropy redundancy f of the bth evaluation index b ;

[0085] S4-1-4, according to the formula:

[0086]

[0087] Get the weight value h of the bth evaluation index b ;

[0088] S4-1-5, according to the formula:

[0089]

[0090] Get the comprehensive score z of the a-th evaluation object a .

[0091] Furthermore, the principal component analysis algorithm in step S4-3 is implemented as follows:

[0092] S4-3-1. According to the formula:

[0093]

[0094] Obtain the processed trajectory data matrix represented by X; where v is the number of channels in the processed trajectory data; u is the number of evaluation indicators in the processed trajectory data;

[0095] S4-3-2, according to the formula:

[0096]

[0097]

[0098] Get the correlation coefficient r of any two evaluation indicators αβ ; and the correlation coefficient r αβ The correlation coefficient matrix R is formed; where α and β represent the αth indicator and the βth evaluation indicator respectively; γ represents the γth channel;

[0099] S4-3-3. Characteristic equation for the correlation coefficient matrix R |λI u -R|=0 to solve and obtain all eigenvalues ​​and corresponding eigenvectors:

[0100]

[0101] Among them, the eigenvector t (1) , t (2) ,…,t (u) The corresponding eigenvalues ​​are λ1, λ2, ..., λ u ;

[0102] S4-3-4, according to the formula:

[0103]

[0104]

[0105] Get the yth principal component Y y Contribution rate b y , the cumulative contribution rate b of 1 to τ principal components μ ; Obtain the value of τ when the cumulative contribution rate reaches more than 90%; where θ represents the θth eigenvalue, 1≤y≤u, 1≤τ≤u;

[0106] S4-3-5. The dimensionless matrix X is obtained by the entropy calculation formula in step S4-1. * , get the first τ principal component scores, expressed as:

[0107]

[0108] Among them, a 11 、a 12 …a τu is the matrix X * The coefficient of .

[0109] Furthermore, the specific method of the nearest neighbor analysis in step S2-2-3 is:

[0110] The processed trajectory data and road network data are input into the neighbor analysis tool to match the processed trajectory data with the road network data; the latitude and longitude coordinates before and after road network matching are placed in the base map, and the position changes of the matched trajectory points are compared; among them, three new fields, namely latitude and longitude and the road network number where the trajectory point is located, are added to the attribute table after matching.

[0111] An electronic device is provided, comprising:

[0112] a memory storing executable instructions; and

[0113] The processor is configured to execute executable instructions in the memory to implement a comprehensive risk assessment method for highway freight corridors.

[0114] The beneficial effects of the present invention are: it can use existing objective data to judge road risks in advance, focus on monitoring high-risk freight channels, and prevent accidents; it can provide safety warnings and references to drivers; and it can add safety-related influences to the operation of freight information platforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0115] Figure 1 This is a flow chart for comprehensive risk assessment of highway freight corridors. DETAILED DESCRIPTION

[0116] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0117] like Figure 1 As shown, a comprehensive risk assessment method for a highway freight channel includes the following steps:

[0118] S1, obtaining original trajectory data, road network data and other data including OD data, China administrative division map base map data and weather data, and pre-processing them to obtain processed data;

[0119] S2, identifying highway freight corridors based on the processed data;

[0120] S3. Select evaluation indicators and establish a risk evaluation indicator system for highway freight corridors;

[0121] S4. Establish a highway freight channel risk assessment model based on the highway freight channel risk assessment indicator system;

[0122] S5. Use the established highway freight channel risk assessment model to identify channel risks.

[0123] The specific implementation method of step S1 is as follows:

[0124] S1-1. Obtain original trajectory data, road network data, and other data including OD data, China administrative division map base map data, and weather data;

[0125] S1-2: Consider the Earth as a sphere and calculate the geographic distances associated with the original trajectory data. Perform reverse geocoding on the longitude and latitude of the trajectory points by calling the API provided by AutoNavi, converting the longitude and latitude information of the trajectory points into text-formatted addresses. Perform duplicate data removal, abnormal data processing, and data range coordinate conversion and filtering on the acquired original trajectory data to complete preprocessing of the original trajectory data and obtain processed trajectory data.

[0126] S1-3, pre-processing the road network data by screening the geographical range and the road grade to obtain processed road network data;

[0127] S1-4. The processed trajectory data, the processed road network data and other data are combined into processed data.

[0128] The specific implementation of step S2 is as follows:

[0129] S2-1. Use the kernel density analysis tool to obtain the distribution of truck track points and the patterns of truck activities;

[0130] S2-2, correcting the processed trajectory data and matching the processed trajectory data with the road network data to obtain a road network matching result;

[0131] S2-3. Based on the road network matching results, the road network data of the trajectory points are sorted according to the frequency of the trajectory points' activities on different roads, the node cities for highway freight are determined, and two or more roads in the node cities are selected as highway freight channels.

[0132] The specific steps to match the processed trajectory data with the road network data are as follows:

[0133] S2-2-1. Set up a buffer zone with a radius of 50m.

[0134] S2-2-2. Use the cropping tool to crop the data outside the road network buffer;

[0135] S2-2-3, perform nearest neighbor analysis on the processed trajectory data and road network data;

[0136] S2-2-4. Perform table association between the road network serial number of the newly added trajectory point in the processed trajectory data and the road network serial number in the road network data, thereby completing the matching between the road network data and the processed trajectory data.

[0137] The specific implementation of step S3 is as follows:

[0138] S3-1, according to the formula:

[0139]

[0140] Get the corresponding road level risk index of the kth trajectory point h k is the proportion of the total number of accidents of the road grade to which the kth trajectory point belongs in the highway freight channel; p k is the single accident death rate of the road grade to which the kth trajectory point belongs in the highway freight channel; l k is the total mileage of the road grade to which the kth trajectory point belongs in the highway freight corridor;

[0141] S3-2, according to the formula:

[0142]

[0143] The road grade risk index x1 in the highway freight corridor is obtained, where n is the number of trajectory points in the corresponding highway freight corridor per unit time;

[0144] S3-3, according to the formula:

[0145]

[0146] Get the special section index x2 of the highway freight channel, Score the risk of the special road section corresponding to the k-th trajectory point;

[0147] S3-4, according to the formula:

[0148]

[0149] Obtain the congestion risk index x3 of the highway freight channel, Score the congestion situation of the k-th trajectory point;

[0150] S3-5, according to the formula:

[0151]

[0152] Get the speeding index x4 of the highway freight channel; Score the speeding situation of the k-th trajectory point;

[0153] S3-6, according to the formula:

[0154]

[0155] Get the highway freight channel overload ratio index x5, is the overload ratio score corresponding to the OD data when both the mth departure city and the arrival city are cities passing through the highway freight corridor; n′ is the total number of OD data in the highway freight corridor;

[0156] S3-7, according to the formula:

[0157]

[0158] Get the dangerous goods transport index x6 of the highway freight channel, Indicates the OD data dangerous goods transport score when both the mth departure city and the arrival city are cities that the highway freight corridor passes through;

[0159] S3-8, according to the formula:

[0160]

[0161]

[0162] Get driving time indicators for highway freight corridors h k represents the freight activity distribution value corresponding to the moment of the k-th trajectory point; s krepresents the traffic accident distribution value corresponding to the time of the k-th trajectory point; x7 represents the driving period index of the highway freight channel;

[0163] S3-9, according to the formula:

[0164]

[0165] Get the driving time index of the highway freight channel x8, It represents the freight time value in the OD data when both the mth departure city and the arrival city are cities passed by the highway freight corridor within a unit time;

[0166] S3-10: Set the temperature impact risk score of trajectory points with a minimum temperature below 0°C and a maximum temperature above 40°C to 1, and set the temperature impact risk score of trajectory points with an all-day temperature between 0°C and 40°C to 0;

[0167] S3-11, according to the formula:

[0168]

[0169] Get the weather impact index x9 of the highway freight channel, represents the weather impact risk score corresponding to the k-th trajectory point in the highway freight corridor;

[0170] S3-12, according to the formula:

[0171]

[0172] Get the positive indicator value after deviation standardization of the processed data Set the a-th highway freight channel as the a-th evaluation object; x ab Represents the values ​​of x1, x2...x9, that is, the value of the bth indicator in the ath evaluation object, 1≤a≤g,1≤b≤s, min b is the minimum value of the b-th indicator, max b is the maximum value of the bth indicator, s means there are s indicators in total, and g means there are g evaluation objects in total;

[0173] S3-13, according to the formula:

[0174]

[0175]

[0176]

[0177]

[0178] Perform z-score standardization on the indicator data to obtain a standard indicator with a variance of 1 and a mean of 0 The indicator data is processed by z-score standardization to obtain the dimensionless matrix X * ; where x ij represents the value of the jth indicator at the i-th evaluation object, 1≤i≤g,1≤j≤s; is the mean of the jth indicator; σ j is the standard deviation of the j-th indicator.

[0179] The specific implementation of step S4 is as follows:

[0180] S4-1. Perform min-max normalization on the indicator data and record the normalized data matrix as Using entropy method Calculate the risk assessment value and obtain the comprehensive score of the risk assessment;

[0181] S4-2. Based on the comprehensive score of risk assessment, the dynamic risk level of the channel is defined as "low risk", "medium-low risk", "medium risk", "medium-high risk" and "high risk";

[0182] S4-3, perform principal component analysis on the evaluation indicators to obtain principal component scores;

[0183] S4-4. Using the results of principal component analysis as the input matrix of the BP neural network and the comprehensive score of risk assessment obtained by the entropy method as the training target matrix, the BP neural network is trained to obtain a trained BP neural network;

[0184] S4-5. Use the trained BP neural network as a risk assessment model for highway freight channels.

[0185] The calculation method of the entropy method in step S4-1 is as follows:

[0186] S4-1-1, according to the formula:

[0187]

[0188] Get the proportion p of the ath evaluation object under the bth evaluation index ab ; Among them, c means there are c evaluation objects in total, and d means there are d evaluation indicators in total;

[0189] S4-1-2, according to the formula:

[0190]

[0191] Get the entropy value e of the bth evaluation index b ;in,

[0192] S4-1-3, according to the formula:

[0193] f b =1-e b

[0194] Get the information entropy redundancy f of the bth evaluation index b ;

[0195] S4-1-4, according to the formula:

[0196]

[0197] Get the weight value h of the bth evaluation index b ;

[0198] S4-1-5, according to the formula:

[0199]

[0200] Get the comprehensive score z of the a-th evaluation object a .

[0201] The steps for implementing the principal component analysis algorithm in step S4-3 are as follows:

[0202] S4-3-1. According to the formula:

[0203]

[0204] Obtain the processed trajectory data matrix represented by X; where v is the number of channels in the processed trajectory data; u is the number of evaluation indicators in the processed trajectory data;

[0205] S4-3-2, according to the formula:

[0206]

[0207]

[0208] Get the correlation coefficient r of any two evaluation indicators αβ ; and the correlation coefficient r αβ The correlation coefficient matrix R is formed; where α and β represent the αth indicator and the βth evaluation indicator respectively; γ represents the γth channel;

[0209] S4-3-3. Characteristic equation for the correlation coefficient matrix R |λI u -R|=0 to solve and obtain all eigenvalues ​​and corresponding eigenvectors:

[0210]

[0211] Among them, the eigenvector t (1) , t (2) ,…,t (u) The corresponding eigenvalues ​​are λ1, λ2, ..., λ u ;

[0212] S4-3-4, according to the formula:

[0213]

[0214]

[0215] Get the yth principal component Y y Contribution rate b y , the cumulative contribution rate b of 1 to τ principal components μ ; Obtain the value of τ when the cumulative contribution rate reaches more than 90%; where θ represents the θth eigenvalue, 1≤y≤u, 1≤τ≤u;

[0216] S4-3-5. The dimensionless matrix X is obtained by the entropy calculation formula in step S4-1. * , get the first τ principal component scores, expressed as:

[0217]

[0218] Among them, a 11 、a 12 …a τu is the matrix X * The coefficient of .

[0219] The specific method of nearest neighbor analysis in step S2-2-3 is:

[0220] The processed trajectory data and road network data are input into the neighbor analysis tool to match the processed trajectory data with the road network data; the latitude and longitude coordinates before and after road network matching are placed in the base map, and the position changes of the matched trajectory points are compared; among them, three new fields, namely latitude and longitude and the road network number where the trajectory point is located, are added to the attribute table after matching.

[0221] In one embodiment of the present invention, four freight channels including G107, G4, G106 and G45 between the Beijing node and the Handan node on November 1, 2018 were selected as channel samples as research objects, their risk assessment values ​​on a specific date were determined, and recommendations for truck travel channel selection were made based on the channel risk values ​​on that day.

[0222] (1) The trajectory data and other multi-source data after data preprocessing are calculated to obtain the original sample data of each indicator of the case sample as shown in Table 1:

[0223] Table 1: Case original risk indicator data

[0224]

[0225] (2) The data is dedimensionalized and then the channel risk assessment value is calculated using the entropy method. The sample risk assessment values ​​are shown in Table 2:

[0226] Table 2: Sample risk assessment value calculation

[0227]

[0228] (3) Principal component analysis is performed on the evaluation index data; the present invention uses SPSS tools to perform principal component analysis. After the original channel index data is de-dimensionalized and centered using the z-score standardization method, principal component analysis is performed on the data corresponding to 9 indicators and the channel names. After the KMO test, it is found that the KMO value is 0.649, which is greater than 0.6 and meets the requirements of principal component analysis. The number of principal components is set to 5, the characteristic root threshold is set to 1.00, and click to start principal component analysis. It is generally believed that the cumulative contribution rate needs to reach more than 90%. As shown in the variance explanation in Table 3, when the principal component is 5, the characteristic root explained by the total variance is less than 1.0, and the contribution rate of variable explanation reaches 93.378%. Therefore, 5 principal components are selected to meet the requirements.

[0229] Table 3: Principal component scores of case data

[0230]

[0231] (4) The principal component results are used as input values ​​and input into the BP neural network model to obtain the sample prediction results. Then, the sample risk assessment value is used as the expected value to compare the output results and determine the evaluation level corresponding to the neural network output value and the expected value as shown in Table 4:

[0232] Table 4: Neural network output results

[0233]

[0234]

[0235] The results show that the risk assessment scores for the four sample corridors connecting Beijing and Handan that day were ranked, from low to high, as G45, G4, G107, and G106. From a risk perspective, truck drivers are recommended to use the low-risk G45 corridor, while the medium-risk G106 corridor is less recommended. This conclusion can also provide guidance to transportation authorities on risk management and traffic flow management.

[0236] The present invention can use existing objective data to judge highway risks in advance, focus on monitoring high-risk freight channels, and prevent accidents; it can provide safety warnings and references to drivers; and it can add safety-related influences to the operation of freight information platforms.

Claims

1. A comprehensive risk assessment method for highway freight corridors, characterized in that: The following steps are involved: S1, obtaining original trajectory data, road network data and other data including OD data, China administrative division map base map data and weather data, and pre-processing them to obtain processed data; S2, identifying highway freight corridors based on the processed data; S3. Select evaluation indicators and establish a risk evaluation indicator system for highway freight corridors; S4. Establish a highway freight channel risk assessment model based on the highway freight channel risk assessment indicator system; the specific implementation method is as follows: S4-1. Perform min-max normalization on the indicator data and record the normalized data matrix as , using entropy method to Calculate the risk assessment value and obtain the comprehensive score of the risk assessment; S4-2. Based on the comprehensive score of risk assessment, the dynamic risk level of the channel is defined as "low risk", "medium-low risk", "medium risk", "medium-high risk" and "high risk"; S4-3. Perform principal component analysis on the evaluation indicators to obtain principal component scores. The implementation steps are as follows: S4-3-1. According to the formula: Obtain the processed trajectory data matrix represented by X; where v is the number of channels in the processed trajectory data; u is the number of evaluation indicators in the processed trajectory data; S4-3-2, according to the formula: Get the correlation coefficient of any two evaluation indicators ; and the correlation coefficient Composed of correlation coefficient matrix R; where, Respectively represent indicators and evaluation indicators; Indicates the channels; S4-3-3. Correlation coefficient matrix Characteristic equation of Solve and obtain all eigenvalues ​​and corresponding eigenvectors: Among them, the eigenvector , , , The corresponding eigenvalue is , , , ; S4-3-4, according to the formula: Get the first principal components Contribution rate , 1 to The cumulative contribution rate of the principal components ; When the cumulative contribution rate reaches more than 90% The value of ; among them, Indicates the eigenvalues, ; S4-3-5. Dimensionless matrix obtained by the entropy method calculation formula in step S4-1 , get the front The principal component scores are expressed as: in, 、 is a matrix The coefficient of S4-4. Using the results of principal component analysis as the input matrix of the BP neural network and the comprehensive score of risk assessment obtained by the entropy method as the training target matrix, the BP neural network is trained to obtain a trained BP neural network; S4-5, using the trained BP neural network as a risk assessment model for highway freight corridors; S5. Use the established highway freight channel risk assessment model to identify channel risks.

2. A comprehensive risk assessment method for highway freight channels according to claim 1, characterized in that: The specific implementation method of step S1 is as follows: S1-1. Obtain original trajectory data, road network data, and other data including OD data, China administrative division map base map data, and weather data; S1-2: Consider the Earth as a sphere and calculate the geographic distances associated with the original trajectory data. Perform reverse geocoding on the longitude and latitude of the trajectory points by calling the API provided by AutoNavi, converting the longitude and latitude information of the trajectory points into text-formatted addresses. Perform duplicate data removal, abnormal data processing, and data range coordinate conversion and filtering on the acquired original trajectory data to complete preprocessing of the original trajectory data and obtain processed trajectory data. S1-3, pre-processing the road network data by screening the geographical range and the road grade to obtain processed road network data; S1-4. The processed trajectory data, the processed road network data and other data are combined into processed data.

3. A comprehensive risk assessment method for highway freight corridors according to claim 2, characterized in that: The specific implementation of step S2 is as follows: S2-1. Use the kernel density analysis tool to obtain the distribution of truck track points and the patterns of truck activities; S2-2, correcting the processed trajectory data and matching the processed trajectory data with the road network data to obtain a road network matching result; S2-3. Based on the road network matching results, the road network data of the trajectory points are sorted according to the frequency of the trajectory points' activities on different roads, the node cities for highway freight are determined, and two or more roads in the node cities are selected as highway freight channels.

4. A comprehensive risk assessment method for highway freight corridors according to claim 3, characterized in that: The specific steps to match the processed trajectory data with the road network data are as follows: S2-2-1. Set up a buffer zone with a radius of 50m. S2-2-2. Use the cropping tool to crop the data outside the road network buffer; S2-2-3, perform nearest neighbor analysis on the processed trajectory data and road network data; S2-2-4. Perform table association between the road network serial number of the newly added trajectory point in the processed trajectory data and the road network serial number in the road network data, thereby completing the matching between the road network data and the processed trajectory data.

5. A comprehensive risk assessment method for highway freight channels according to claim 3, characterized in that: The specific implementation of step S3 is as follows: S3-1, according to the formula: Get the first The corresponding road level risk index of each trajectory point ; For the first time in the highway freight channel The proportion of total accidents of the road class to which the trajectory point belongs; For the first time in the highway freight channel The single accident fatality rate of the road class to which each trajectory point belongs; For the first time in the highway freight channel The total mileage of the road class to which each trajectory point belongs; S3-2, according to the formula: Obtain road grade risk index within highway freight corridor , n is the number of trajectory points in the corresponding highway freight channel per unit time; S3-3, according to the formula: Get the special section indicators of highway freight corridors , For the Each trajectory point corresponds to a risk score for a special road section; S3-4, according to the formula: Obtain the congestion risk index of highway freight channels , For the Congestion score of each trajectory point; S3-5, according to the formula: Obtain speeding indicators for highway freight channels ; For the Speeding score for each track point; S3-6, according to the formula: Obtain the overload ratio index of highway freight channel , For the The overload ratio score of OD data where both the departure city and the arrival city are cities that the highway freight corridor passes through; is the total number of OD data within the highway freight corridor; S3-7, according to the formula: Get the dangerous goods transport indicators of highway freight corridors , Indicates the OD data dangerous goods transport score for cities where both the departure and arrival cities are cities that the highway freight corridor passes through; S3-8, according to the formula: Get driving time indicators for highway freight corridors ; Indicates the The freight activity distribution value corresponding to the moment of each trajectory point; Indicates the The distribution value of traffic accidents corresponding to the time of each trajectory point; An indicator representing the driving hours of highway freight corridors; S3-9, according to the formula: Get the driving time index of the highway freight corridor , Indicates the number of The freight time value in the OD data of the departure city and the arrival city are both cities that the highway freight corridor passes through; S3-10: Set the temperature impact risk score of trajectory points with a minimum temperature below 0°C and a maximum temperature above 40°C to 1, and set the temperature impact risk score of trajectory points with an all-day temperature between 0°C and 40°C to 0; S3-11, according to the formula: Obtain weather impact indicators for highway freight corridors , Indicates the first Weather impact risk score corresponding to each trajectory point; S3-12, according to the formula: Get the positive indicator value after deviation standardization of the processed data , set the a-th highway freight channel as the a-th evaluation object; express The value of The indicator in The value of the evaluation object, , It is The minimum value of the indicator, It is The maximum value of the indicators, s means there are s indicators in total, and g means there are g evaluation objects in total; S3-13, according to the formula: Perform z-score standardization on the indicator data to obtain a standard indicator with a variance of 1 and a mean of 0 , the indicator data is processed by z-score standardization to obtain a dimensionless matrix ;in, Representative The first indicator The value of the evaluation object, ; It is The mean of the indicators; is the standard deviation of the j-th indicator.

6. A comprehensive risk assessment method for highway freight corridors according to claim 1, characterized in that: The calculation method of the entropy method in step S4-1 is as follows: S4-1-1, according to the formula: Get the first Under the evaluation index The proportion of evaluation objects in this indicator ; Among them, c means there are c evaluation objects in total, Indicates shared evaluation indicators; S4-1-2, according to the formula: Get the first The entropy value of the evaluation index ;in, , ; S4-1-3, according to the formula: Get the information entropy redundancy of the bth evaluation index ; S4-1-4, according to the formula: Get the weight value of the bth evaluation index ; S4-1-5, according to the formula: Get the comprehensive score of the ath evaluation object .

7. A comprehensive risk assessment method for highway freight corridors according to claim 1, characterized in that: The specific method of nearest neighbor analysis in step S2-2-3 is: The processed trajectory data and road network data are input into the neighbor analysis tool to match the processed trajectory data with the road network data; the latitude and longitude coordinates before and after road network matching are placed in the base map, and the position changes of the matched trajectory points are compared; among them, three new fields, namely latitude and longitude and the road network number where the trajectory point is located, are added to the attribute table after matching.

8. An electronic device, characterized in that: The device comprises: a memory storing executable instructions; and A processor configured to execute the executable instructions in the memory to implement the method according to any one of claims 1 to 7.

Images