A driver behavior evaluation method and system based on online freight platform

Abstract

The present invention relates to a driver behavior evaluation method and system based on an online freight platform, comprising preprocessing and grading a historical behavior dataset of freight drivers; performing dimensionality reduction and projection on the historical behavior dataset using principal component analysis; generating a two-dimensional visualization graph with a credibility region; establishing a driver behavior evaluation model; and determining the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated using the driver behavior evaluation model. The present invention also provides a driver behavior evaluation system based on an online freight platform, comprising: a data preprocessing module, a principal component analysis module, a two-dimensional visualization graph module, a driver behavior evaluation model establishment module, a driver behavior level determination module, a driver behavior guidance module, and a driver behavior stability judgment module. The present invention introduces the behavior data of drivers on the online freight platform into driver behavior evaluation and guidance, comprehensively evaluates and guides the behavior of freight drivers in multiple dimensions, and improves the credit of freight drivers.

Description

Technical Field

[0001] The present invention relates to the field of logistics technology, and in particular to a driver behavior evaluation method and system based on an online freight platform. Background Art

[0002] On September 6, 2019, the Ministry of Transport and the State Administration of Taxation issued the "Interim Measures for the Management of Online Platform Road Freight Transport Operations." This renamed the "truckless carrier" introduced in 2016 to "online platform road freight transporter." This allows technology companies in the logistics sector to rationally dispatch vehicles and drivers through online freight platforms, reducing empty vehicle loads and improving logistics and transportation efficiency. Over the past two years, online freight platforms have not only reduced waste of resources and time, but also lowered shipping costs for shippers and increased driver incomes. Despite this, online freight platforms still have limitations in their driver evaluation and management: existing logistics platforms typically evaluate and score driver performance based solely on shippers' feedback on shipping orders, failing to comprehensively assess and guide driver behavior from multiple perspectives. This results in inconsistent driver credibility and service capabilities, negatively impacting the actual user experience for shippers. Therefore, how to effectively evaluate and guide driver performance has become a hot topic.

[0003] In summary, the existing online freight platforms have the problem of being unable to evaluate and guide the behavioral norms of freight drivers based on comprehensive multi-dimensional information in their assessment and management of freight drivers. Summary of the Invention

[0004] To this end, the technical problem to be solved by the present invention is to overcome the problem that the existing technology cannot evaluate and guide the behavioral norms of freight drivers based on comprehensive multi-dimensional information, resulting in uneven credit and service capabilities of freight drivers and damage to the actual user experience of cargo owners.

[0005] To solve the above technical problems, the present invention provides a driver behavior evaluation method based on an online freight platform, comprising the following steps:

[0006] Obtaining a historical behavior dataset of freight drivers on an online freight platform, preprocessing the historical behavior dataset and dividing the preprocessed historical behavior dataset into different levels;

[0007] Using principal component analysis to reduce the dimension of the hierarchical historical behavior data set and projecting it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph;

[0008] Generate a two-dimensional visualization map with a credibility region according to the credibility region equation;

[0009] Identifying different levels in the two-dimensional visualization map with the credibility region, obtaining a two-dimensional visualization map with the identified levels, and establishing a driver behavior evaluation model that characterizes the mapping relationship between freight driver behavior and different levels;

[0010] The real-time behavior data corresponding to each transport order of the driver to be evaluated is obtained and preprocessed and principal component analysis is performed. The real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis is projected onto a two-dimensional visualization diagram of the identification level using a driver behavior evaluation model to determine the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated.

[0011] Preferably, the historical behavior data set includes multiple groups of historical behavior data samples, each group of historical behavior data samples includes but is not limited to one or more of the freight driver's capital flow data, transportation location data, transportation cargo data, transportation time data, and traffic violation data; the preprocessing includes but is not limited to one or more of repeated observation processing, missing value deletion, missing value replacement, missing value interpolation, outlier detection, outlier deletion, outlier replacement, and data feature combination.

[0012] Preferably, the data feature combination processing of the historical behavior data set is to use statistical indicators to perform feature combination and screening on the historical behavior data set, and the statistical indicators include but are not limited to one or more of the number of transportation times in a single month, average transportation time, average transportation distance, cargo types, and average capital turnover.

[0013] Preferably, dividing the pre-processed historical behavior data set into different levels includes:

[0014] Based on each set of historical behavior data samples in the preprocessed historical behavior dataset, the corresponding freight driver's transportation efficiency, transportation quality, and transportation cost are calculated;

[0015] According to the preset weights corresponding to the freight driver's transportation efficiency, transportation quality, and transportation cost, a weighted sum is performed on the freight driver's transportation efficiency, transportation quality, and transportation cost to determine the total score of each group of historical behavior data samples in the preprocessed historical behavior data set;

[0016] The total scores of each group of historical behavior data samples in the historical behavior data set are sorted, and the pre-processed historical behavior data set is graded according to the sorting results.

[0017] Preferably, the specific steps of using principal component analysis to reduce the dimension of the hierarchical historical behavior data set and projecting it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph are:

[0018] Set the preprocessed historical behavior dataset as X, and calculate the covariance matrix based on the preprocessed historical behavior dataset X

[0019] Use eigenvalue decomposition method to find The eigenvalues ​​and eigenvectors of

[0020] Arrange the obtained eigenvalues ​​in descending order, select the two largest eigenvalues, namely λ1 and λ2, and use their corresponding two eigenvectors as row vectors to form the eigenvector matrix P;

[0021] Calculate the coordinate Y=PX of the historical behavior dataset in the two-dimensional coordinate system and construct a two-dimensional visualization graph.

[0022] Preferably, generating a two-dimensional visualization graph having a credibility region according to the credibility region equation comprises:

[0023] Calculate the credibility region equation, the calculation formula is:

[0024]

[0025] Where λ1 and λ2 are the first and second eigenvalues ​​selected after principal component analysis, θ is the credibility level, x is the horizontal coordinate of the historical behavior data point in the two-dimensional visualization graph, and y is the vertical coordinate of the historical behavior data point in the two-dimensional visualization graph;

[0026] Delete historical behavior data points outside the credibility zone.

[0027] Preferably, the method further comprises obtaining the real-time behavior data corresponding to each transport order of the driver to be evaluated and performing preprocessing and principal component analysis, projecting the real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis onto a two-dimensional visualization graph using a driver behavior evaluation model, and determining the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated.

[0028] According to the time spent on each transport order of the driver to be evaluated, a motion trajectory diagram of the real-time behavior data points corresponding to each transport order of the driver to be evaluated changing over time is obtained;

[0029] Calculate the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior point in the motion trajectory diagram, and guide the behavior of the driver to be evaluated based on the relationship between the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior pattern point and the preset distance threshold;

[0030] Calculate the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated, and judge the behavioral stability of the driver to be evaluated based on the relationship between the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the preset stability threshold.

[0031] Preferably, the calculation formula for the distance index d between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior point is:

[0032]

[0033] Among them, x i is the horizontal coordinate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory graph, i x is the ordinate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory graph, m is the horizontal coordinate of the highest level behavior point in the motion trajectory diagram, y m It is the vertical coordinate of the highest level behavior point in the motion trajectory diagram.

[0034] Preferably, the level change rate v of the real-time behavior data points corresponding to each transport order of the driver to be evaluated is calculated as follows:

[0035]

[0036] Where T is the total time taken by the driver to be evaluated for n transport orders,

[0037] The real-time behavior data points corresponding to the n transport orders of the driver to be evaluated are expressed as follows:

[0038] (x1,y1),(x2,y2),...,(x n ,y n )

[0039] The distance between the real-time behavior data points corresponding to the nth transport order of the driver to be evaluated is D n , and its calculation formula is:

[0040]

[0041] The sum of the distance changes of the real-time behavior data points corresponding to the n transport orders of the driver to be evaluated is:

[0042] D1+D2+...+D n

[0043] The present invention also provides a driver behavior evaluation system based on an online freight platform, comprising:

[0044] Data preprocessing module: used to obtain the historical behavior dataset of freight drivers on the online freight platform, preprocess the historical behavior dataset and divide the preprocessed historical behavior dataset into different levels;

[0045] Principal component analysis module: used to reduce the dimension of the hierarchical historical behavior data set according to the principal component analysis method and project it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph;

[0046] Two-dimensional visualization module: used to generate a two-dimensional visualization map with a credibility region according to the credibility region equation;

[0047] Driver Behavior Assessment Model Building Module: This module is used to identify different levels in a two-dimensional visualization with credibility regions, generate a two-dimensional visualization of the identified levels, and establish a driver behavior assessment model that characterizes the mapping relationship between freight driver behavior and different levels;

[0048] Driver Behavior Level Determination Module: This module is used to obtain the real-time behavior data corresponding to each transport order of the driver to be evaluated and perform preprocessing and principal component analysis. The real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis is projected onto a two-dimensional visualization map indicating the level using the driver behavior evaluation model to determine the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated.

[0049] Driver Behavior Guidance Module: This module is used to calculate the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior pattern point in the motion trajectory diagram, and guide the behavior of the driver to be evaluated based on the relationship between the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior pattern point and the preset distance threshold;

[0050] Driver behavior stability judgment module: used to calculate the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated in the motion trajectory diagram, and judge the stability of the driver's behavior based on the relationship between the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the preset stability threshold.

[0051] Compared with the existing technology, this solution has the following advantages:

[0052] The present invention preprocesses the historical behavior data set of freight drivers on the network freight platform, reduces low-quality data such as abnormal data values, and improves the accuracy of the data; uses the principal component analysis method to reduce the dimension of the historical behavior data set and projects it to a plane, so that the multi-dimensional historical behavior data set that is difficult to process is converted into a low-dimensional and visualized graph, which is convenient for establishing a driver behavior evaluation model; generates a credibility area in a two-dimensional visualization graph according to the credibility area equation, so that the historical behavior data points with the lowest level and abnormalities fall outside the credibility area, and the historical behavior data points of other levels fall within the credibility area; identifies different levels in the two-dimensional visualization graph with credibility areas, so that when judging the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated, the judgment result can be presented more intuitively and conveniently; the driver behavior evaluation model constructed based on the above method can guide the behavior evaluation of freight drivers more accurately and quickly from a comprehensive multi-dimensional perspective, improve the credit and service capabilities of the drivers on the freight platform, and ensure the actual user experience of the cargo owners. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings, wherein

[0054] Figure 1 is a flow chart of the present invention;

[0055] Figure 2 is a two-dimensional visualization diagram of the present invention;

[0056] Figure 3 A two-dimensional visualization of the credibility region of the present invention;

[0057] Figure 4 It is a two-dimensional visualization diagram after the level division of the present invention;

[0058] Figure 5 is the motion trajectory diagram of the present invention;

[0059] Figure 6 This is a schematic diagram of the driver behavior assessment system provided by the present invention. DETAILED DESCRIPTION

[0060] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0061] Embodiment 1:

[0062] Reference Figure 1 As shown, the present invention is a driver behavior evaluation method based on an online freight platform, comprising the following steps:

[0063] S1: Obtain a historical behavior dataset of freight drivers on an online freight platform, preprocess the historical behavior dataset, and divide the preprocessed historical behavior dataset into different levels;

[0064] The historical behavior data set includes one or more of the following data: the freight driver's capital flow data, transportation location data, transportation cargo data, transportation time data, traffic violation data, etc.

[0065] The obtained historical behavior data set can be preprocessed by selecting one or more of the following methods: repeated observation processing, missing value deletion, missing value replacement, missing value interpolation, outlier detection, outlier deletion, outlier replacement, data feature combination, etc.

[0066] The data feature combination processing of the historical data set includes: using statistical indicators to perform feature combination and screening on the historical behavior data set, and the statistical indicators include but are not limited to one or more of the number of transportation times in a single month, average transportation time, average transportation distance, cargo types, and average capital turnover.

[0067] S2: Use principal component analysis to reduce the dimension of the hierarchical historical behavior dataset and project it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph;

[0068] S3: Generate a two-dimensional visualization map with credibility region according to the credibility region equation;

[0069] S4: Identifying different levels in the two-dimensional visualization map with the credibility region, obtaining a two-dimensional visualization map with the identified levels, and establishing a driver behavior evaluation model that characterizes the mapping relationship between freight driver behavior and different levels;

[0070] S5: Obtain the real-time behavior data corresponding to each transport order of the driver to be evaluated and perform preprocessing and principal component analysis. Use the driver behavior evaluation model to project the real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis onto a two-dimensional visualization diagram that identifies the level, and determine the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated.

[0071] This embodiment preprocesses the historical behavior data set of freight drivers on the online freight platform, reduces low-quality data such as abnormal data values, and improves the accuracy of the data; uses the principal component analysis method to reduce the dimension of the historical behavior data set and project it onto a plane, so that the multi-dimensional and difficult-to-process historical behavior data set is converted into a low-dimensional and visual graph, which is convenient for establishing a driver behavior evaluation model; generates a credibility area in the two-dimensional visualization graph according to the credibility area equation, so that the historical behavior data points with the lowest level and abnormalities fall outside the credibility area, and the historical behavior data points of the remaining levels fall within the credibility area; identifies different levels in the two-dimensional visualization graph with credibility areas, so that when judging the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated, the judgment result can be presented more intuitively and conveniently; in summary, the driver behavior evaluation model built based on the above method can guide the behavior of freight drivers more accurately and quickly from a comprehensive multi-dimensional perspective, improve the credit and service capabilities of the drivers on the freight platform, and ensure the actual user experience of the shippers.

[0072] Example 2:

[0073] Based on the above embodiment, this embodiment further explains the process of building a driver behavior evaluation model. The specific steps of building a driver behavior evaluation model are as follows:

[0074] S20: Obtain a historical behavior dataset of freight drivers on the online freight platform, and preprocess the historical behavior dataset.

[0075] S21: Calculate the transportation efficiency, transportation quality, and transportation cost of the corresponding freight driver based on each set of historical behavior data samples in the preprocessed historical behavior data set.

[0076] S22: Based on the preset weight ratios of transport efficiency, transport quality, and transport cost, the transport efficiency, transport quality, and transport cost of the freight driver are weighted and summed to determine the total score of each group of historical behavior data samples.

[0077] S23: Sort the total scores of each group of historical behavior data samples in the historical behavior data set, and classify the historical behavior data set according to the ranking.

[0078] In step S21, the transportation efficiency, transportation quality, and transportation cost of the freight driver corresponding to each set of historical behavior data samples in the preprocessed historical behavior data set are calculated. The calculation methods include but are not limited to the following two methods:

[0079] S231: Calculate the freight driver's transportation efficiency, transportation quality, and transportation cost using the shipper's evaluation score, the consignee's evaluation score, and the carrier's system evaluation score in the historical behavior data set:

[0080]

[0081] S232: Using the transportation distance / total time, cargo damage degree / actual damage rate, fuel costs + highway fees + traffic fines + vehicle maintenance fees + cargo damage compensation fees in the historical behavior dataset, calculate the freight driver's transportation efficiency, transportation quality, and transportation cost:

[0082]

[0083] The method for grading the historical behavior data set according to the sorting described in step S23 includes but is not limited to arranging the historical behavior data set of freight drivers from high to low according to the total score, and taking the top 10%, 10% to 20%, 20% to 40%, and 40% to 100% as the grades of excellent, good, qualified, and unqualified, respectively.

[0084] S24: Set the historical behavior dataset after level division as X, and calculate the covariance matrix based on the historical behavior dataset X

[0085] S25: Use eigenvalue decomposition method to find The eigenvalues ​​and eigenvectors of .

[0086] S26: Arrange the obtained eigenvalues ​​in descending order, select the two largest eigenvalues, namely λ1 and λ2, and use their corresponding two eigenvectors as row vectors to form an eigenvector matrix P.

[0087] S27: Calculate the coordinate Y=PX of the historical behavior dataset in the two-dimensional coordinate system and construct a two-dimensional visualization graph.

[0088] S28: Generating a two-dimensional visualization graph having a credibility region according to the credibility region equation includes:

[0089] S281: Calculate the credibility region equation, the calculation formula is:

[0090]

[0091] Where λ1 and λ2 are the first and second eigenvalues ​​selected after principal component analysis, θ is the credibility level, x is the horizontal coordinate of the historical behavior data point in the two-dimensional visualization graph, and y is the vertical coordinate of the historical behavior data point in the two-dimensional visualization graph;

[0092] S282: Delete historical behavior data points outside the credibility area.

[0093] S29: Identify different levels in the two-dimensional visualization map with credibility areas, obtain a two-dimensional visualization map with identified levels, and establish a driver behavior evaluation model that characterizes the mapping relationship between freight driver behavior and different levels.

[0094] This embodiment uses a multi-dimensional historical behavior data set to calculate the freight driver's transportation efficiency, transportation quality, and transportation cost when establishing a driver behavior evaluation model that characterizes the mapping relationship between freight driver behavior and different levels, thereby improving the comprehensiveness and accuracy of freight driver evaluation; the principal component analysis method is used to reduce the dimension of the historical behavior data set and project it onto a plane, so that the multi-dimensional and difficult-to-process historical behavior data set is converted into a low-dimensional and visual graph, which can more intuitively present the freight driver behavior level and facilitate the establishment of a driver behavior evaluation model.

[0095] Example 3:

[0096] Based on the above embodiment, in this embodiment, the motion trajectory diagram can be used to calculate the real-time behavior data points (x i ,y i ) and excellent behavior pattern points (x m ,y m ) guides the behavior of freight drivers, and the stability of the driver's behavior can be determined by calculating the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated, including:

[0097] S30: In the motion trajectory diagram, calculate the real-time behavior data points (x i ,y i ) and excellent behavior pattern points (x m ,y m ) is calculated as follows:

[0098]

[0099] Among them, x i is the horizontal coordinate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory graph, i x is the ordinate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory diagram; m is the horizontal coordinate of the excellent behavior pattern point in the motion trajectory diagram, y m The vertical coordinate of the excellent behavior pattern point in the motion trajectory diagram;

[0100] S31: Further guide the behavior of the driver to be evaluated based on the distance indicator d, including:

[0101] Based on operational experience, α1 is set as the distance threshold for judging the behavior of the driver to be evaluated as an excellent behavior pattern, and α2 is set as the distance threshold for judging the behavior of the driver to be evaluated as a qualified behavior pattern, where α1 < α2. Therefore, the relationship between the distance index d and the behavioral norm guidance can be expressed as:

[0102]

[0103] S32: In the motion trajectory diagram, based on the total time spent on n transport orders by the driver to be evaluated and the sum of the distances of the real-time behavior data points corresponding to the n transport orders of the driver to be evaluated, calculate the level change rate v of the real-time behavior data points corresponding to each transport order of the driver to be evaluated, and determine the behavioral stability of the driver to be evaluated, including:

[0104] S321: Calculate the level change rate v of the real-time behavior data points corresponding to each transport order of the driver to be evaluated, and the calculation formula is:

[0105]

[0106] Where T is the total time taken by the driver to be evaluated for n transport orders,

[0107] The real-time behavior data points corresponding to the n transport orders of the driver to be evaluated are expressed as follows:

[0108] (x1,y1),(x2,y2),...,(x n ,y n )

[0109] The distance between the real-time behavior data points corresponding to the nth transport order of the driver to be evaluated is D n , and its calculation formula is:

[0110]

[0111] The sum of the distance changes of the real-time behavior data points corresponding to the n transport orders of the driver to be evaluated is:

[0112] D1+D2+...+D n

[0113] S322: Determine the behavioral stability of the driver to be evaluated based on the relationship between the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the stability threshold β:

[0114] When v>β, the driver's behavior is less stable and his credit and service capabilities are weaker;

[0115] When v≤β, the driver's behavior is more stable and his credit and service capabilities are stronger.

[0116] This embodiment uses a motion trajectory diagram to calculate the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the excellent behavior pattern points, and guides the behavior of the driver to be evaluated based on the relationship between the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the excellent behavior pattern points and a preset distance threshold, thereby improving the comprehensiveness and accuracy of guiding the behavior of the driver to be evaluated; by calculating the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated, the stability of the behavior of the driver to be evaluated is judged based on the relationship between the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated and a stability threshold, which helps to improve the credit and service capabilities of the driver to be evaluated and ensure the actual user experience of the cargo owner.

[0117] Embodiment 4:

[0118] refer to Figure 6 As shown, Figure 6 A driver behavior evaluation system based on an online freight platform is provided in an embodiment of the present invention; it specifically includes: a data preprocessing module, a principal component analysis module, a two-dimensional visualization module, a driver behavior evaluation model establishment module, a driver behavior level determination module, a driver behavior guidance module, and a driver behavior stability judgment module.

[0119] The driver behavior evaluation system based on the network freight platform of this embodiment is used to implement the aforementioned driver behavior evaluation method based on the network freight platform. Therefore, the specific implementation method of the driver behavior evaluation system based on the network freight platform can be seen in the embodiment part of the driver behavior evaluation method based on the network freight platform in the previous text. For example, the data preprocessing module 100 is used to obtain the historical behavior data set of freight drivers on the network freight platform, preprocess the historical behavior data set and divide the preprocessed historical behavior data set into different levels; the principal component analysis module 200 is used to use the principal component analysis method to reduce the dimension of the historical behavior data set after the level division and project it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph; the two-dimensional visualization graph module 300 is used to generate a two-dimensional visualization graph with a credibility area according to the credibility area equation; the driver behavior evaluation model establishment module 400 is used to identify different levels in the two-dimensional visualization graph with a credibility area, generate a two-dimensional visualization graph after the identified levels, and establish a driver behavior evaluation model that characterizes the mapping relationship between freight driver behavior and different levels. ; The driver behavior level determination module 500 is used to obtain the real-time behavior data corresponding to each transport order of the driver to be evaluated and perform preprocessing and principal component analysis, and project the real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis onto a two-dimensional visualization map identifying the level using the driver behavior evaluation model to determine the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated; The driver behavior guidance module 600 is used to calculate the distance index between the real-time behavior data point corresponding to each transport order of the driver to be evaluated and the highest-level behavior pattern point in the motion trajectory map, and guide the behavior of the driver to be evaluated based on the relationship between the distance index between the real-time behavior data point corresponding to each transport order of the driver to be evaluated and the highest-level behavior pattern point and the preset distance threshold; The driver behavior stability judgment module 700 is used to calculate the level change rate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory map, and judge the stability of the behavior of the driver to be evaluated based on the relationship between the level change rate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated and the stability threshold.

[0120] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A driver behavior evaluation method based on an online freight platform, characterized in that: include: Obtain a historical behavior dataset of freight drivers on the online freight platform, preprocess the historical behavior dataset, and divide the preprocessed historical behavior dataset into different levels: calculate the corresponding freight driver's transportation efficiency, transportation quality, and transportation cost based on each set of historical behavior data samples in the preprocessed historical behavior dataset; perform a weighted sum of the freight driver's transportation efficiency, transportation quality, and transportation cost based on the preset weights corresponding to the freight driver's transportation efficiency, transportation quality, and transportation cost, determine the total score of each set of historical behavior data samples in the preprocessed historical behavior dataset, and sort them; and divide the preprocessed historical behavior dataset into different levels based on the sorting results; The principal component analysis method is used to reduce the dimension of the historical behavior dataset after classification and project it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph: According to the preprocessed historical behavior dataset Calculate the covariance matrix ;Use eigenvalue decomposition method to find The eigenvalues ​​and eigenvectors of ; Arrange the obtained eigenvalues ​​in descending order and select the two largest eigenvalues, which are and , and use the corresponding two eigenvectors as row vectors to form an eigenvector matrix ; Calculate the coordinates of the historical behavior dataset in the two-dimensional coordinate system ,construct a two-dimensional visualization graph; According to the credibility region equation Generate a two-dimensional visualization graph with a credibility region and delete historical behavior data points outside the credibility region; and are the first and second eigenvalues ​​selected after principal component analysis, is the credibility level, is the horizontal coordinate of the historical behavior data point in the two-dimensional visualization diagram, The vertical coordinate of the historical behavior data point in the two-dimensional visualization graph; Identifying different levels in the two-dimensional visualization map with the credibility region, obtaining a two-dimensional visualization map with the identified levels, and establishing a driver behavior evaluation model that characterizes the mapping relationship between freight driver behavior and different levels; The real-time behavior data corresponding to each transport order of the driver to be evaluated is obtained and preprocessed and principal component analysis is performed. The real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis is projected onto a two-dimensional visualization diagram of the identification level using a driver behavior evaluation model to determine the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated.

2. The driver behavior evaluation method based on the online freight platform according to claim 1 is characterized in that: The historical behavior data set includes multiple groups of historical behavior data samples, each group of historical behavior data samples includes but is not limited to one or more of the freight driver's capital flow data, transportation location data, transportation cargo data, transportation time data, and traffic violation data; The preprocessing includes but is not limited to one or more of repeated observation processing, missing value deletion, missing value replacement, missing value interpolation, outlier detection, outlier deletion, outlier replacement, and data feature combination.

3. The driver behavior evaluation method based on the online freight platform according to claim 2 is characterized in that: Performing data feature combination processing on the historical data set includes: The historical behavior data set is characterized and filtered using statistical indicators, wherein the statistical indicators include but are not limited to one or more of the number of transports in a single month, average transport time, average transport distance, cargo types, and average capital turnover.

4. The driver behavior evaluation method based on the online freight platform according to claim 1 is characterized in that: The method further includes obtaining the real-time behavior data corresponding to each transport order of the driver to be evaluated, performing preprocessing and principal component analysis, projecting the real-time behavior data corresponding to each transport order of the driver to be evaluated onto a two-dimensional visualization graph using a driver behavior evaluation model, and determining the level of the real-time behavior data corresponding to each transport order of the driver to be evaluated. According to the time spent on each transport order of the driver to be evaluated, a motion trajectory diagram of the real-time behavior data points corresponding to each transport order of the driver to be evaluated changing over time is obtained; Calculate the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior point in the motion trajectory diagram, and guide the behavior of the driver to be evaluated based on the relationship between the distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior pattern point and the preset distance threshold; Calculate the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated, and judge the behavioral stability of the driver to be evaluated based on the relationship between the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the preset stability threshold.

5. The driver behavior evaluation method based on the online freight platform according to claim 4 is characterized in that: The distance index between the real-time behavior data points corresponding to each transport order of the driver to be evaluated and the highest-level behavior point , and its calculation formula is: ； in, is the horizontal coordinate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory diagram, is the ordinate of the real-time behavior data point corresponding to each transport order of the driver to be evaluated in the motion trajectory diagram, is the horizontal coordinate of the highest level behavior point in the motion trajectory diagram, It is the vertical coordinate of the highest level behavior point in the motion trajectory diagram.

6. The driver behavior evaluation method based on the online freight platform according to claim 4 is characterized in that: The level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated , and its calculation formula is: ； in, For drivers to be evaluated The total time used for each transport order, Drivers awaiting evaluation The real-time behavior data points corresponding to each transport order are expressed as follows: ； Drivers to be evaluated The distance between the real-time behavior data points corresponding to a transport order is , and its calculation formula is: ； Drivers awaiting evaluation The total distance of the real-time behavior data points corresponding to each transport order is: 。 7. A driver behavior evaluation system based on an online freight platform, characterized in that: include: Data preprocessing module: used to obtain the historical behavior data set of freight drivers on the online freight platform, preprocess the historical behavior data set and divide the preprocessed historical behavior data set into different levels: according to each group of historical behavior data samples in the preprocessed historical behavior data set, calculate the corresponding freight driver's transportation efficiency, transportation quality, and transportation cost; according to the preset weights corresponding to the freight driver's transportation efficiency, transportation quality, and transportation cost, perform weighted summation of the freight driver's transportation efficiency, transportation quality, and transportation cost, determine the total score of each group of historical behavior data samples in the preprocessed historical behavior data set, and sort them; and divide the preprocessed historical behavior data set into different levels according to the sorting results; Principal Component Analysis Module: It is used to reduce the dimension of the historical behavior data set after classification using principal component analysis and project it into a two-dimensional coordinate system to obtain a two-dimensional visualization graph: Based on the pre-processed historical behavior data set Calculate the covariance matrix ;Use eigenvalue decomposition method to find The eigenvalues ​​and eigenvectors of ; Arrange the obtained eigenvalues ​​in descending order and select the two largest eigenvalues, which are and , and use the corresponding two eigenvectors as row vectors to form an eigenvector matrix ; Calculate the coordinates of the historical behavior dataset in the two-dimensional coordinate system ,construct a two-dimensional visualization graph; 2D visualization module: used to generate credible region equations Generate a two-dimensional visualization graph with a credibility region and delete historical behavior data points outside the credibility region; and are the first and second eigenvalues ​​selected after principal component analysis, is the credibility level, is the horizontal coordinate of the historical behavior data point in the two-dimensional visualization diagram, The vertical coordinate of the historical behavior data point in the two-dimensional visualization graph; Driver Behavior Assessment Model Building Module: This module is used to identify different levels in a two-dimensional visualization with credibility regions, generate a two-dimensional visualization of the identified levels, and establish a driver behavior assessment model that characterizes the mapping relationship between freight driver behavior and different levels; Driver behavior grade determination module: used to obtain real-time behavior data corresponding to each transport order of the driver to be evaluated, preprocess and perform principal component analysis on the real-time behavior data corresponding to each transport order of the driver to be evaluated, and project the real-time behavior data corresponding to each transport order of the driver to be evaluated after preprocessing and principal component analysis onto a two-dimensional visualization map indicating the grade using the driver behavior evaluation model to determine the grade of the real-time behavior data corresponding to each transport order of the driver to be evaluated; Driver Behavior Guidance Module: Calculates the distance between the real-time behavior data points and the highest-level behavior pattern points for each transport order of the driver to be evaluated in the motion trajectory diagram, and guides the behavior of the driver to be evaluated based on the relationship between the distance between the real-time behavior data points and the highest-level behavior pattern points for each transport order of the driver to be evaluated and a preset distance threshold; Driver behavior stability judgment module: used to calculate the level change rate of the real-time behavior data points corresponding to each transport order of the driver to be evaluated in the motion trajectory diagram, and judge the stability of the driver's behavior to be evaluated based on the relationship between the level change rate corresponding to each transport order of the driver to be evaluated and the preset stability threshold.

Images