CN114386737A - Method for monitoring safety of passengers on network appointment in mobile internet era - Google Patents

Method for monitoring safety of passengers on network appointment in mobile internet era Download PDF

Info

Publication number
CN114386737A
CN114386737A CN202111174950.2A CN202111174950A CN114386737A CN 114386737 A CN114386737 A CN 114386737A CN 202111174950 A CN202111174950 A CN 202111174950A CN 114386737 A CN114386737 A CN 114386737A
Authority
CN
China
Prior art keywords
value
time
driver
passengers
calculating
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202111174950.2A
Other languages
Chinese (zh)
Other versions
CN114386737B (en
Inventor
严佳杰
倪骏
严佳豪
付凤杰
周妍
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang Police College
Original Assignee
Zhejiang Police College
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang Police College filed Critical Zhejiang Police College
Priority to CN202111174950.2A priority Critical patent/CN114386737B/en
Publication of CN114386737A publication Critical patent/CN114386737A/en
Application granted granted Critical
Publication of CN114386737B publication Critical patent/CN114386737B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0645—Rental transactions; Leasing transactions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G06F18/24155—Bayesian classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G06Q50/265—Personal security, identity or safety
    • G06Q50/40—

Abstract

The invention provides a method for monitoring safety of networked car booking passengers in the mobile internet era. The method has the basic idea that multidimensional space-time data are fused, the probability of the online taxi appointment illegal behavior is calculated, and online taxi appointment risk early warning is carried out. In order to achieve the purpose, the monitoring technical method provided by the invention comprises the steps of clustering the driver characteristics based on a K-means algorithm, calculating the risk of the geographic environment based on a kernel density analysis method, extracting and modeling a time-space condition, calculating the probability of the prior illegal action based on an entropy method, calculating the probability of the posterior illegal action based on a Bayesian theory and establishing an illegal action early warning mechanism. The invention makes up the defects of neglecting dynamic geographic environment, unreasonable behavior psychology and existing case statistical law, not only can effectively restrict the driving behavior of the net car reservation driver, but also can timely prevent, discover and stop the illegal behavior through the early warning system, thereby ensuring the legal rights and interests of passengers.

Description

Method for monitoring safety of passengers on network appointment in mobile internet era
Technical Field
The invention relates to a safety monitoring method for networked car booking passengers in the mobile internet era, which is used for probability prediction of networked car booking illegal behaviors and early warning of the illegal behaviors and belongs to the field of intelligent transportation and early warning of risks of the illegal behaviors.
Background
With the rise of the internet +, the network car appointment gradually becomes the main travel mode of the public. However, a net appointment for car infringement cases occurs at all times. The related monitoring technology is perfected, the early warning of the illegal behaviors is realized, and the illegal behaviors can be prevented, discovered and prevented in time.
At present, the research of supervision paths is still in a searching stage, and a corresponding supervision system is difficult to establish and perfect in a short time. And the network appointment safety monitoring focuses on the judgment of abnormal tracks more, and the probability calculation of passenger risks or driver misbehavior is ignored. In addition, the unfair behavior probability model focuses more on static geographic environment factors, neglects dynamic geographic environment, unfair behavior psychology and existing case statistical rules. Therefore, the method has important significance in realizing the prediction of the probability of the illegal behaviors of the networked car reduction driver and perfecting the real-time monitoring technology of the illegal behaviors by considering the characteristics of the driver, the geographic environment factors, the space-time conditions and the characteristics of passengers and adopting a kernel density analysis method, an entropy method and a Bayesian theorem.
Disclosure of Invention
The invention aims to provide a method for monitoring safety of networked car booking passengers in the mobile internet era.
In order to achieve the above purpose, the monitoring method provided by the invention comprises the following steps: the method comprises the steps of clustering driver features based on a K-means algorithm, calculating geographic environment risks based on a kernel density analysis method, extracting and modeling space-time conditions, calculating the prior illegal action probability based on an entropy method, calculating the posterior illegal action probability based on a Bayesian theory and establishing an illegal action early warning mechanism.
The basic steps of the invention are as follows:
c1, clustering the driver characteristics based on the K-means algorithm;
c2, calculating the risk of the geographic environment based on a nuclear density analysis method;
c3, extracting and modeling a space-time condition;
c4, calculating the probability of the prior illegal action based on the entropy method;
c5, calculating the posterior dishonest behavior probability based on the Bayes theory;
c6, establishing an illegal behavior early warning mechanism.
The driver characteristic clustering process based on the K-means algorithm in the step c1 includes,
c11, obtaining basic data of taxi appointment drivers in a certain urban area network, including sex, age, number of accepted orders, comprehensive rating, poor rating proportion and complaint proportion of each driver, and obtaining 6-dimensional attribute x of any driver ii＝{xi1,xi2,xi3,xi4,xi5,xi6The method concretely comprises the following steps:
c111, setting a virtual variable: sex rule of men xi10, sex rule xi1＝1；
c112, standardization treatment: age, number of orders received, composite score, i.e. xij＝(xij’-μij)/σij. In the formula, xij' is the original value, x, of the jth attribute of driver iijIs xij' normalization result, μijAnd σijRespectively, the mean and variance of the original value of the jth attribute of the driver ith.
c12 sample set R by K-means algorithm6Performing cluster analysis, R6＝{x1,…xM}. The K value is taken as 3, and the category names are defined as constant asset deep type, constant step up type, impulse part-time type. The method comprises the following main steps:
c121, randomly selecting 3 cluster centroids as mu1,μ2,μ3∈R6；
And c122, repeating iterative class division and centroid calculation by using the following formulas until the centroid is unchanged or slightly changed.
Figure RE-GDA0003540133250000031
Figure RE-GDA0003540133250000032
In the formula, ciRepresents the closest class, x, of the ith driver to the 3 clustered centroid pointsi6-dimensional attribute, μ, representing the ith driverkDenotes the k-th cluster centroid point, μkjThe j-th characteristic value of the k-th cluster centroid point is represented, and M represents the sample number of the driver.
The process of calculating the risk of the geographic environment based on the nuclear density analysis in the step c2 includes,
c21, acquiring basic data points representing the geographic environment of the same urban area by using a Baidu map API, assigning the weight p of the data points of the life service, the food and the shopping to be 1, and assigning the weight p of the forest and the construction site to be-1. And performing nuclear density analysis on all data points by utilizing ArcMap and Python to obtain a nuclear density calculation result. The calculation formula of the nuclear density f (x, y) at an arbitrary position (x, y) is as follows,
Figure RE-GDA0003540133250000033
Figure RE-GDA0003540133250000034
Figure RE-GDA0003540133250000041
where r represents search radius/bandwidth, diRepresents the data point i (x)i,yi) And between arbitrary positions (x, y)Sd represents the standard distance, dmRepresenting all data points to the mean center
Figure RE-GDA0003540133250000042
Median value of the distances of, piThe weight of the data point i is shown, and n is the number of data points.
c22, selecting one acquisition point at intervals of 100m, and determining the nuclear density value of any acquisition point on the line according to the analysis and calculation result of the nuclear density of the area. And calculating the overall trip risk according to the nuclear density average value of all the acquisition points, calculating the real-time risk according to the nuclear density value at the real-time positioning position, and normalizing the nuclear density value and the real-time risk by using a Min-Max standardization method.
Figure RE-GDA0003540133250000043
Figure RE-GDA0003540133250000044
In the formula, RfRepresenting the overall trip risk, i.e. the normalized result of the mean value of the kernel densities of J collection points along the route, f (x)j,yj) Nuclear density value, f (x) at the c-th collection pointt,yt) For real-time positioning (x)t,yt) Nuclear density value of (C), RrRepresenting real-time risks of travel, i.e. real-time location (x)t,yt) Normalized result of kernel density value, fminRepresenting the minimum value of the nuclear density of the current region, fmaxRepresenting the current region kernel density maximum.
The process of extracting and modeling spatiotemporal conditions in step c3 includes,
c31, taking 22:00 as a time starting point and 10:00 as a time end point, and carrying out normalization processing on the departure time. Taking 100km as the maximum value of the travel distance, and carrying out normalization processing on the travel distance, wherein the value of more than 100km is 1.
c32, when the driver is yawing and not switching to the usual course, is considered to be an abnormal yaw. According to the travel OD, a hundredth map API is used for obtaining historical optimal travel routes, including common routes which are shortest in time and distance and are capable of avoiding congestion in different periods (early peak, average peak, late peak and low peak) on different working days and non-working days, and a common route set L (L is { L ═ L { (L is obtained1, L2,…,LN}. Setting an abnormal yaw distance initial value to 0(ds ═ 0), and possibly switching route set L ═ L1,L2,…,LEIf the initial value of the possible switching route number is E (E ═ N-1), the method and the steps for judging the abnormal yaw are as follows:
c321, calculating the current location (x) according to a formulat,yt) And point (x)D,yD) A distance d betweenDtLast positioning (x)t’,yt’) And (x)D,yD) A distance d betweenDt’And the difference delta d therebetweenDt。
Figure RE-GDA0003540133250000051
c322, when Δ dDt>When 0, jump to c 323; otherwise, the result is returned as normal driving, betadJump to c321, 0.
c323, sequentially calculating the current location (x) according to a formulat,yt) Shortest distance to each common route and its variation Δ dft。
Δdft＝d((xt,yt),Lf)-d((xt',yt'),Lf)，f＝1,2,...,E
In the formula,. DELTA.dDtIndicates the current location (x)t,yt) And point (x)D,yD) The distance between and the last positioning (x)t’,yt’) And (x)D,yD) The difference between the distances, f, represents the f-th common route, f 1, 2.
c324, f is 1,2, …, E, and Δ d is sequentially determinedft(ii) a When Δ dftWhen not less than 0, E is equal to E-1, and L' is eliminatedf。
c325, if E >0, return to c 321; otherwise jump to c 326.
c326, let the abnormal yaw distance ds＝ΔdDtAbnormal yaw distance ratio (normalized value) betad＝ΔdDtAnd/s, t equals t +10, and returns to c 321.
c33, when the real-time road condition is smooth and the vehicle speed is less than 5km/h, the vehicle is considered to be abnormally stopped. And if the running time is longer than the predicted travel time, expanding the abnormal parking time by using the danger coefficient. By utilizing the Baidu map API and the GPS terminal, real-time information of the vehicle, including the speed vtTime of travel T, predicted travel time T0Real-time road condition S of road section where vehicle is locatedlt: congestion (0), slowness (1), smoothness (2). The method comprises the following specific steps:
c331, calculating the abnormal parking risk coefficient eta, wherein the formula is as follows:
η＝max(Tt/T0,1)
in the formula, TtAcquired real-time travel time, T, of the vehicle0Indicating the expected travel time.
c332, judging the road condition and the vehicle speed in real time, and when S is reachedlt2 and vt<At 5km/h, ts +10, the abnormal parking time ts'. eta.ts, and the abnormal parking time ratio (normalized value) βt＝ts’/T0。
c333, t equals t +10, return c 331.
The calculation process of the prior unfairness probability based on the entropy method in the step c4 includes:
c41, calculating entropy of 3 types of drivers with steady and serious resources, steady ascending and impulsive facultative driver, and sequentially judging the overall travel risk R of different types of driversfTrip real-time risk RrDeparture time t0Travel distance s and abnormal yaw distance ratio betadAbnormal parking time length ratio betatRandomness and degree of disorder of 6 factors. Analyzing m trip samples, 6 evaluation indexes to form an original index data matrix:
Figure RE-GDA0003540133250000061
in the formula, XmjAnd the j-th item of evaluation index of the m-th trip sample is represented by a numerical value.
c42, standardizing each index, wherein the formula is as follows,
Figure RE-GDA0003540133250000071
in the formula, XmjIs the original value, X 'of the jth attribute of travel m'mjIs XmjNormalized result of (1), μmjAnd σmjRespectively, the mean and variance of the original value of the jth attribute of the trip m.
And c43, calculating the proportion of the m sample mark value in the j index, and establishing a proportion matrix of the data.
Figure RE-GDA0003540133250000072
Figure RE-GDA0003540133250000073
In the formula, hmjAnd (4) the specific gravity of the m-th sample mark value in the j-th index is represented, and H represents a specific gravity matrix of the data.
c44, calculating the entropy value of the j index.
Figure RE-GDA0003540133250000074
Figure RE-GDA0003540133250000075
0≤ej≤1
In the formula, ejEntropy representing j index, k is constant, and m represents number of samples of trip。
c45, defining the degree of difference and the weight of the jth index.
dj＝1-ej
Figure RE-GDA0003540133250000076
In the formula (d)jAnd wjRespectively representing the degree of difference and the weight of the jth index, ejRepresenting the entropy value of the j-th index.
c46, comprehensively evaluating the value of the mth sample, namely the probability prior value p (C) of the illegal behavior of the taxi booking driver.
Figure RE-GDA0003540133250000081
The posterior dishonest behavior probability calculation process based on the bayes theory in the step c5 comprises the following steps:
c51, classifying the passengers into five classes according to age, gender and number of people, and the proportion of each class of passengers in normal travel and the proportion of each class of passengers in the travel with the driver having the abnormal behavior are shown in the following table, which respectively correspond to the conditional probability P (P | N) of each class of passengers when the travel is known to be normal and the conditional probability P (P | C) of each class of passengers when the driver is known to have the abnormal behavior.
Figure RE-GDA0003540133250000082
c52, extracting passenger information mainly comprising gender, age and number of people, and judging the passenger type. And calculating the posterior value of the probability of the illegal behaviors of the network car booking driver by using Bayesian theorem according to the prior value of the probability of the illegal behaviors of the network car booking driver and the conditional probability of the class of passengers.
Figure RE-GDA0003540133250000083
Wherein P represents a passenger category, and P is 1,2, 3, 4, 5; p (C | P) represents the probability value of the illegal behavior of the net appointment driver when the passenger is known to be in the class P, and P (N) represents the probability of the illegal behavior of the driver.
The process of establishing the illegal action early warning mechanism in the step c6 comprises the following steps:
and c61, automatically sending early warning reminding and confirmation messages to the passengers, reminding the passengers of paying attention to the driving behaviors of the drivers, and confirming whether the drivers have suspicious behaviors, wherein the normal feedback is 1, and the feedback is 0 when no feedback or problems exist.
c62, when the feedback is 1, returning to the step c 61; when the feedback is 0, an early warning prompt and a confirmation message are automatically sent to the passenger emergency contact person, the vehicle information is sent to the passenger emergency contact person, whether the passenger is safe or not is confirmed, the normal feedback is 3, and the feedback is 2 when no feedback or a problem exists.
c63, when the feedback is 3, returning to the step c 61; when the feedback is 2, early warning reminding and confirmation messages are automatically sent to the online car booking platform, the current states of a driver and passengers are confirmed to the online car booking platform, the normal feedback is 5, and the feedback is 4 when no feedback or problems exist.
c64, when the feedback is 5, returning to the step c 61; and when the feedback is 4, automatically sending the information of the vehicle and the driver to a management department, and the management department confirms the current states of the driver and the passenger by using the data of the card port until the passenger safely arrives at the destination.
The invention has the beneficial effects that: the method is based on a clustering algorithm, a kernel density analysis and a discrimination algorithm, and converts space-time multidimensional data such as driver characteristics, geographic environment, space-time conditions and the like into main factors influencing the probability of the illegal action; then, analyzing the influence of the driver characteristics, the geographic environment and the space-time conditions on the probability of the improper behavior by using an entropy method, and modeling; and finally, according to a Bayesian theory, mining and utilizing passenger information to calculate the posterior illicit behavior probability. Various factors are considered comprehensively, different analysis methods are adopted according to different influences of different factors and different characteristics of different time-space data, scientific reasonability of a calculation result of the probability of the illegal action can be improved, and therefore invalid early warning is reduced on the basis of guaranteeing timely early warning.
Drawings
FIG. 1 flow chart of a risk early warning system
Detailed Description
The invention is explained in detail below with reference to the attached drawing 1, and the specific steps of the invention are as follows:
the method comprises the following steps of firstly, extracting various features of a driver, and carrying out cluster analysis on the driver by using a K-means algorithm:
(1) and acquiring basic data of taxi appointment drivers in a certain urban area network, wherein the basic data comprises the sex, the age, the number of accepted orders, comprehensive scores, poor score ratios, complaint ratios and the like of each driver. Because the features of different dimensions are in different numerical magnitudes, in order to reduce the influence of the features with large variance and accelerate the convergence speed, the data needs to be processed to obtain the 6-dimensional attribute x of any driver ii＝{xi1,xi2,xi3,xi4,xi5, xi6The specific method is as follows:
virtual variables: sex rule of men xi10, sex rule xi1＝1。
And (6) standardization treatment: age, number of orders received, composite score, i.e. xij＝(xij’-μij)/σij. In the formula, xij' is the original value, x, of the jth attribute of driver iijIs xij' normalization result, μijAnd σijRespectively, the mean and variance of the original value of the jth attribute of the driver ith.
(2) Using K-means algorithm to sample set R6Performing cluster analysis, R6＝{x1,…xM}. Since excessive classification causes difficulty in scoring by experts and decreases accuracy, the K value is set to 3 according to actual experience, and the category name is defined as a steady capital type, a steady ascending type, and an impulsive facultative type. The algorithm has two main steps:
selecting 3 cluster centroids as mu at random1,μ2,μ3∈R6；
And secondly, repeating iterative class division and centroid calculation by using the following formulas until the centroid is unchanged or slightly changed.
Figure RE-GDA0003540133250000111
Figure RE-GDA0003540133250000112
In the formula, ciRepresents the closest class, x, of the ith driver to the 3 clustered centroid pointsi6-dimensional attribute, u, representing the ith driverkDenotes the k-th cluster centroid point, μkjThe j-th characteristic value of the k-th cluster centroid point is represented, and M represents the sample number of the driver.
Step two, calculating the whole risk and the real-time risk by using a nuclear density analysis method according to different weights of places such as life services, food and drink, shopping places, mountain forests/parks and the like:
(1) basic data points representing geographic environments of the same urban area are obtained by using a Baidu map API, the weight p of the data points of life service, food and shopping is assigned to be 1, and the weight p of a forest and a construction site is assigned to be-1. And performing nuclear density analysis on all data points by utilizing ArcMap and Python to obtain a nuclear density calculation result. The calculation formula of the nuclear density f (x, y) at an arbitrary position (x, y) is as follows:
Figure RE-GDA0003540133250000113
Figure RE-GDA0003540133250000114
Figure RE-GDA0003540133250000115
where r represents search radius/bandwidth, diRepresents the data point i (x)i,yi) Distance from an arbitrary position (x, y), sd represents a standard distance, dmRepresents all data pointsTo the mean center
Figure RE-GDA0003540133250000116
Median value of the distances of, piThe weight of the data point i is shown, and n is the number of data points.
(2) One acquisition point is selected at intervals of 100m, and the nuclear density value of any acquisition point on the line can be determined according to the analysis and calculation result of the nuclear density of the area. The overall trip risk is calculated according to the average nuclear density value of all the collection points, the real-time risk is calculated according to the nuclear density value of the real-time positioning position, and the Min-Max standardization method is utilized to carry out normalization processing on the overall trip risk and the real-time risk.
Figure RE-GDA0003540133250000121
Figure RE-GDA0003540133250000122
In the formula, RfRepresenting the overall trip risk, i.e. the normalized result of the mean value of the kernel densities of J collection points along the route, f (x)j,yj) Nuclear density value, f (x) at the c-th collection pointt,yt) For real-time positioning (x)t,yt) Nuclear density value of (C), RrRepresenting real-time risks of travel, i.e. real-time location (x)t,yt) Normalized result of kernel density value, fminRepresenting the minimum value of the nuclear density of the current region, fmaxRepresenting the current region kernel density maximum.
Thirdly, modeling according to data such as travel characteristics, yaw, parking and the like by using a discrimination algorithm:
(1) since the time when the net taxi driver takes the unfair act is generally more than night, the departure time is normalized with 22:00 as the maximum value and 10:00 as the minimum value. Most of the taxi appointment trips are trips inside the city, therefore, 100km is taken as the maximum value of trip distance, the trip distance is normalized, and the value greater than 100km is 1.
(2) When the driver is yawing and not switching to a usual course, it is considered to be an abnormal yaw. According to the travel OD, a hundredth map API is used for obtaining historical optimal travel routes, including common routes which are shortest in time and distance and are capable of avoiding congestion in different periods (early peak, average peak, late peak and low peak) on different working days and non-working days, and a common route set L (L is { L ═ L { (L is obtained1, L2,…,LN}. Setting an abnormal yaw distance initial value to 0(ds ═ 0), and possibly switching route set L ═ L1,L2,…,LEIf the initial value of the possible switching route number is E (E ═ N-1), the method and the steps for judging the abnormal yaw are as follows:
calculating the current location (x) according to a formulat,yt) And D point (x)D,yD) A distance d betweenDtLast positioning (x)t’,yt’) And (x)D,yD) A distance d betweenDt’And the difference delta d therebetweenDt。
Figure RE-GDA0003540133250000131
When delta dDt>When 0, jumping to the third step; otherwise, the result is returned as normal driving, betadJump to (0).
Calculating the current location (x) according to the formulat,yt) Shortest distance to each common route and its variation Δ dft。
Δdft＝d((xt,yt),Lf)-d((xt',yt'),Lf)
f＝1,2,...,E
(iv) f 1,2, …, E, and sequentially determining Δ dft(ii) a When Δ dftWhen not less than 0, E is equal to E-1, and L' is eliminatedp。
If E is greater than 0, returning to the step I; otherwise, the process goes to the step of sixthly.
Sixthly, abnormal yaw distance ds＝ΔdDtAbnormal yaw distance ratio (normalized value) betad＝ΔdDtAnd/s, t is t +10, and the process returns to the step of (i).
(3) When the real-time road condition is smooth and the vehicle speed is too low (less than 5km/h), the vehicle is considered to be abnormally stopped. And if the running time is longer than the predicted travel time, expanding the abnormal parking time by using the danger coefficient. By utilizing the Baidu map API and a GPS terminal (such as a mobile phone), real-time information of the vehicle can be acquired, including the speed vtTime of travel T, predicted travel time T0Real-time road condition S of road section where vehicle is locatedlt: congestion (0), slowness (1), smoothness (2). The method comprises the following specific steps:
calculating an abnormal parking risk coefficient eta, wherein the formula is as follows:
η＝max(Tt/T0,1)
secondly, judging road conditions and vehicle speed in real time, and when S is used, judging the speed of the vehiclelt2 and vt<At 5km/h, ts +10, the abnormal parking time ts'. eta.ts, and the abnormal parking time ratio (normalized value) βt＝ts’/T0。
And t is t +10, and the process returns to the step (r).
Step four, calculating the prior illegal behavior probability based on the entropy method:
(1) sequentially judging the integral travel risk R of different classes of drivers by calculating the entropy values of 3 classes of drivers with stable and heavy resources, stably ascending and impulsive concurrent functionsfTrip real-time risk RrDeparture time t0Travel distance s and abnormal yaw distance ratio betadAbnormal parking time length ratio betatThese 6 factors are randomness and degree of disorder. Analyzing m trip samples, 6 evaluation indexes to form an original index data matrix:
Figure RE-GDA0003540133250000141
in the formula, XmjAnd the j-th item of evaluation index of the m-th trip sample is represented by a numerical value.
(2) In order to eliminate the influence on the evaluation result due to different dimensions, each index needs to be standardized, and the formula is as follows:
Figure RE-GDA0003540133250000142
in the formula, XmjIs the original value, X 'of the jth attribute of travel m'mjIs XmjNormalized result of (1), μmjAnd σmjRespectively, the mean and variance of the original value of the jth attribute of the trip m.
(3) And calculating the proportion of the mark value of the mth sample in the jth index, and establishing a proportion matrix of the data.
Figure RE-GDA0003540133250000151
Figure RE-GDA0003540133250000152
In the formula, hmjAnd (4) the specific gravity of the m-th sample mark value in the j-th index is represented, and H represents a specific gravity matrix of the data.
(4) And calculating the entropy value of the j index.
Figure RE-GDA0003540133250000153
Figure RE-GDA0003540133250000154
0≤ej≤1
In the formula, ejRepresents the entropy of the j index, k is a constant, and m represents the number of samples in a trip.
(5) The degree of difference and the weight of the jth index are defined.
dj＝1-ej
Figure RE-GDA0003540133250000155
In the formula (d)jAnd wjRespectively representing the degree of difference and the weight of the jth index, ejRepresenting the entropy value of the j-th index.
(6) And comprehensively evaluating the value of the mth sample, namely obtaining the probability prior value p (C) of the illegal behaviors of the taxi booking driver.
Figure RE-GDA0003540133250000156
Fifthly, calculating the posterior dishonest behavior probability based on the Bayes theory:
(1) the passengers are classified into five classes according to age, gender and number of people, and the proportion of each class of passengers in normal travel and in the travel with the driver having the wrong behavior (50 cases) is shown in the following table, which respectively corresponds to the conditional probability P (P | N) of each class of passengers when the known travel is normal and the conditional probability P (P | C) of each class of passengers when the known driver has the wrong behavior.
Figure RE-GDA0003540133250000161
(2) Passenger information, mainly including gender, age and number of people, is extracted, and passenger categories are judged. And calculating the posterior value of the probability of the illegal behaviors of the network car booking driver by using Bayesian theorem according to the prior value of the probability of the illegal behaviors of the network car booking driver and the conditional probability of the class of passengers.
Figure RE-GDA0003540133250000162
Wherein P represents a passenger category, and P is 1,2, 3, 4, 5; p (C | P) represents the probability value of the illegal behavior of the net appointment driver when the passenger is known to be in the class P, and P (N) represents the probability of the illegal behavior of the driver.
Step six, establishing an illegal behavior early warning mechanism:
the method comprises the steps of automatically sending early warning reminding and confirmation messages to passengers, reminding the passengers of paying attention to driving behaviors of drivers, and confirming whether the drivers have suspicious behaviors, wherein normal feedback is 1, and the feedback is 0 when no feedback or problems exist.
When the feedback is 1, returning to the step I; when the feedback is 0, an early warning prompt and a confirmation message are automatically sent to the passenger emergency contact person, the vehicle information is sent to the passenger emergency contact person, whether the passenger is safe or not is confirmed, the normal feedback is 3, and the feedback is 2 when no feedback or a problem exists.
When the feedback is 3, returning to the step I; when the feedback is 2, early warning reminding and confirmation messages are automatically sent to the online car booking platform, the current states of a driver and passengers are confirmed to the online car booking platform, the normal feedback is 5, and the feedback is 4 when no feedback or problems exist.
Fourthly, when the feedback is 5, returning to the first step; and when the feedback is 4, automatically sending the information of the vehicle and the driver to a management department, and the management department confirms the current states of the driver and the passenger by using the data of the card port until the passenger safely arrives at the destination.

Claims (7)

1. A method for monitoring safety of networked car booking passengers in the era of mobile internet is characterized by comprising the following steps:
c1, clustering the driver characteristics based on the K-means algorithm;
c2, calculating the risk of the geographic environment based on a nuclear density analysis method;
c3, extracting and modeling a space-time condition;
c4, calculating the probability of the prior illegal action based on the entropy method;
c5, calculating the posterior dishonest behavior probability based on the Bayes theory;
c6, establishing an illegal behavior early warning mechanism.
2. The method for monitoring the safety of the passengers online for taxi appointment in the era of mobile internet according to claim 1, wherein the method comprises the following steps:
c11, obtaining basic data of taxi appointment drivers in urban area including sex, age, number of accepted orders, comprehensive rating, poor rating and complaint rating of each driver to obtain any6-dimensional attribute x of driver ii＝{xi1,xi2,xi3,xi4,xi5,xi6The method concretely comprises the following steps:
c111, setting a virtual variable: sex rule of men xi10, sex rule xi1＝1；
c112, standardization treatment: age, number of orders received, composite score, i.e. xij＝(xij’-μij)/σij(ii) a In the formula, xij' is the original value, x, of the jth attribute of driver iijIs xij' normalization result, μijAnd σijRespectively, the average value and the variance of the original value of the jth attribute of the driver ith;
c12 sample set R by K-means algorithm6Performing cluster analysis, R6＝{x1,…xM}; the K value is taken as 3, and the category names are defined as a steady capital deep type, a steady ascending type and an impulse concurrent type, and the method comprises the following two main steps:
c121, randomly selecting 3 cluster centroids as mu1,μ2,μ3∈R6；
c122, repeating iterative class division and centroid calculation by using the following formulas until the centroid is unchanged or slightly changed;
Figure RE-FDA0003540133240000021
Figure RE-FDA0003540133240000022
in the formula, ciRepresents the closest class, x, of the ith driver to the 3 clustered centroid pointsi6-dimensional attribute, μ, representing the ith driverkDenotes the k-th cluster centroid point, μkjThe j-th characteristic value of the k-th cluster centroid point is represented, and M represents the sample number of the driver.
3. The method for monitoring the safety of the passengers online under the appointment in the era of mobile internet according to claim 2, wherein the method comprises the following steps: the process of calculating the risk of the geographic environment based on the nuclear density analysis method in the step c2 comprises the following steps:
c21, acquiring basic data points representing the geographic environment of the same urban area by using a Baidu map API, assigning the weight p of the data points of living services, food and shopping to be 1, and assigning the weight p of the forest and the construction site to be-1;
performing nuclear density analysis on all data points by utilizing ArcMap and Python to obtain a nuclear density calculation result; the calculation formula of the nuclear density f (x, y) at an arbitrary position (x, y) is as follows,
Figure RE-FDA0003540133240000023
Figure RE-FDA0003540133240000024
Figure RE-FDA0003540133240000025
where r represents search radius/bandwidth, diRepresents the data point i (x)i,yi) Distance from an arbitrary position (x, y), sd represents a standard distance, dmRepresenting all data points to the mean center
Figure RE-FDA0003540133240000033
Median value of the distances of, piThe weight of the data point i is taken as n is the number of the data points;
c22, selecting an acquisition point at an interval of 100m, and determining the nuclear density value of any acquisition point on the line according to the analysis and calculation result of the regional nuclear density;
calculating the overall trip risk according to the nuclear density average value of all the acquisition points, calculating the real-time risk according to the nuclear density value of the real-time positioning position, and performing normalization processing on the overall trip risk and the real-time risk by using a Min-Max standardization method;
Figure RE-FDA0003540133240000031
Figure RE-FDA0003540133240000032
in the formula, RfRepresenting the overall trip risk, i.e. the normalized result of the mean value of the kernel densities of J collection points along the route, f (x)j,yj) Nuclear density value, f (x) at the c-th collection pointt,yt) For real-time positioning (x)t,yt) Nuclear density value of (C), RrRepresenting real-time risks of travel, i.e. real-time location (x)t,yt) Normalized result of kernel density value, fminRepresenting the minimum value of the nuclear density of the current region, fmaxRepresenting the current region kernel density maximum.
4. The method for monitoring the safety of passengers online for taxi appointment in the era of mobile internet according to claim 3, wherein the method comprises the following steps: the process of extracting and modeling the space-time conditions in step c3 includes
c31, taking 22:00 as a time starting point and 10:00 as a time end point, and carrying out normalization processing on the starting time;
taking 100km as the maximum value of the travel distance, and carrying out normalization processing on the travel distance, wherein the value of more than 100km is 1;
c32, when the driver is yawing and not switching to a common course, is considered abnormal yaw; according to the travel OD, a hundredth degree map API is used for obtaining historical optimal travel routes, including common routes which are shortest in time and distance and are capable of avoiding congestion in different time periods on different working days and non-working days, and a common route set L ═ { L ═ L { (L })1,L2,…,LN}；
Setting the initial value of the abnormal yaw distance ds to be 0, and possibly switching the route set L' ═ { L }1,L2,…,LEIf the initial value of the possible switching route number is E, the method and the steps for judging the abnormal yaw are as follows:
c321, calculating the current location (x) according to a formulat,yt) And point (x)D,yD) A distance d betweenDtLast positioning (x)t’,yt’) And (x)D,yD) A distance d betweenDt’And the difference delta d therebetweenDt；
Figure RE-FDA0003540133240000041
c322, when Δ dDt>When 0, jump to c 323; otherwise, the result is returned as normal driving, betadJump to c321, 0;
c323, sequentially calculating the current location (x) according to a formulat,yt) Shortest distance to each common route and its variation Δ dft；
Δdft＝d((xt,yt),Lf)-d((xt',yt'),Lf)，f＝1,2,...,E
In the formula,. DELTA.dDtIndicates the current location (x)t,yt) And point (x)D,yD) The distance between and the last positioning (x)t’,yt’) And (x)D,yD) The difference in distance between, f denotes the f-th common route, f 1, 2.., E;
c324, f is 1,2, …, E, and Δ d is sequentially determinedft(ii) a When Δ dftWhen not less than 0, E is equal to E-1, and L' is eliminatedf；
c325, if E >0, return to c 321; otherwise jump to c 326;
c326, let the abnormal yaw distance ds＝ΔdDtAbnormal yaw distance ratio betad＝ΔdDtC321 is returned to,/s, t ═ t + 10;
c33, when the real-time road condition is smooth and the vehicle speed is less than 5km/h, the vehicle is considered to be abnormally stopped;
if the driving time is longer than the expected driving timeThe journey time, expand the unusual parking time with danger coefficient; acquiring real-time information of a vehicle, including speed v, by using a Baidu map API and a GPS terminaltTime of travel T, predicted travel time T0Real-time road condition S of road section where vehicle is locatedlt: congestion, slowness and smoothness; the method comprises the following specific steps:
c331, calculating an abnormal parking risk coefficient eta, wherein the formula is as follows,
η＝max(Tt/T0,1)
in the formula, TtAcquired real-time travel time, T, of the vehicle0Representing a predicted travel time;
c332, judging the road condition and the vehicle speed in real time, and when S is reachedlt2 and vt<At 5km/h, ts is ts +10, the abnormal parking time ts' is η · ts, and the abnormal parking time ratio βt＝ts’/T0；
c333, t equals t +10, return c 331.
5. The method for monitoring the safety of passengers online for taxi appointment in the era of mobile internet according to claim 4, wherein the method comprises the following steps: the calculation process of the prior unfairness probability based on the entropy method in the step c4 includes:
c41, calculating entropy of 3 types of drivers with steady and serious resources, steady ascending and impulsive facultative driver, and sequentially judging the overall travel risk R of different types of driversfTrip real-time risk RrDeparture time t0Travel distance s and abnormal yaw distance ratio betadAbnormal parking time length ratio betatRandomness and disorder degree of 6 factors; analyzing m trip samples, 6 evaluation indexes to form an original index data matrix:
Figure RE-FDA0003540133240000051
in the formula, XmjA value representing the j-th evaluation index of the mth trip sample;
c42, standardizing each index, wherein the formula is as follows,
Figure RE-FDA0003540133240000061
in the formula, XmjIs the original value, X 'of the jth attribute of travel m'mjIs XmjNormalized result of (1), μmjAnd σmjRespectively is the average value and the variance of the jth attribute original value of the trip m;
c43, calculating the proportion of the m sample mark value in the j index, and establishing a proportion matrix of the data;
Figure RE-FDA0003540133240000062
Figure RE-FDA0003540133240000063
in the formula, hmjThe specific gravity of the mark value of the mth sample in the jth index is represented, and H represents a specific gravity matrix of data;
c44, calculating the entropy value of the jth index;
Figure RE-FDA0003540133240000064
Figure RE-FDA0003540133240000065
0≤ej≤1
in the formula, ejRepresenting the entropy value of the j index, k is a constant, and m represents the number of samples of a trip;
c45, defining the difference degree and the weight of the jth index;
dj＝1-ej
Figure RE-FDA0003540133240000066
in the formula (d)jAnd wjRespectively representing the degree of difference and the weight of the jth index, ejEntropy representing the jth index;
c46, comprehensively evaluating the value of the mth sample, namely obtaining the probability prior value p (C) of the illegal behavior of the net car booking driver;
Figure RE-FDA0003540133240000071
6. the method for monitoring the safety of passengers online for taxi appointment in the era of mobile internet according to claim 5, wherein the method comprises the following steps: the posterior dishonest behavior probability calculation process based on the bayes theory in the step c5 comprises the following steps:
c51, dividing the passengers into five classes according to age, gender and number of people, wherein the proportion of each class of passengers in normal travel to the proportion of the passengers in the travel with the driver having the abnormal behavior respectively corresponds to the conditional probability P (P | N) of each class of passengers when the travel is known to be normal and the conditional probability P (P | C) of each class of passengers when the driver is known to have the abnormal behavior;
c52, extracting passenger information mainly comprising gender, age and number of people, and judging the passenger category; calculating a posterior value of the probability of the illegal behaviors of the network car booking driver by using Bayesian theorem according to the prior value of the probability of the illegal behaviors of the network car booking driver and the conditional probability of the class of passengers;
Figure RE-FDA0003540133240000072
wherein P represents a passenger category, and P is 1,2, 3, 4, 5; p (C | P) represents the probability value of the illegal behavior of the net appointment driver when the passenger is known to be in the class P, and P (N) represents the probability of the illegal behavior of the driver.
7. The method for monitoring the safety of passengers online for taxi appointment in the era of mobile internet according to claim 6, wherein the method comprises the following steps: the step c6 of establishing an illegal action early warning mechanism comprises the following steps:
c61, automatically sending early warning reminding and confirmation messages to the passengers, reminding the passengers to pay attention to the driving behaviors of the drivers, and confirming whether the drivers have suspicious behaviors, wherein the normal feedback is 1, and the feedback is 0 when no feedback or problems exist;
c62, when the feedback is 1, returning to the step c 61; when the feedback is 0, automatically sending early warning reminding and confirmation messages to the passenger emergency contact, sending vehicle information to the passenger emergency contact, and confirming whether the passenger is safe or not, wherein the normal feedback is 3, and the feedback is 2 when no feedback or a problem exists;
c63, when the feedback is 3, returning to the step c 61; when the feedback is 2, automatically sending early warning reminding and confirmation messages to the online car booking platform, confirming the current states of a driver and passengers to the online car booking platform, wherein the normal feedback is 5, and the normal feedback is 4 when no feedback or problems exist;
c64, when the feedback is 5, returning to the step c 61; and when the feedback is 4, automatically sending the information of the vehicle and the driver to a management department, and the management department confirms the current states of the driver and the passenger by using the data of the card port until the passenger safely arrives at the destination.
CN202111174950.2A 2021-10-09 2021-10-09 Method for monitoring safety of passengers on internet about car in mobile interconnection era Active CN114386737B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111174950.2A CN114386737B (en) 2021-10-09 2021-10-09 Method for monitoring safety of passengers on internet about car in mobile interconnection era

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111174950.2A CN114386737B (en) 2021-10-09 2021-10-09 Method for monitoring safety of passengers on internet about car in mobile interconnection era

Publications (2)

Publication Number Publication Date
CN114386737A true CN114386737A (en) 2022-04-22
CN114386737B CN114386737B (en) 2023-10-13

Family

ID=81194850

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111174950.2A Active CN114386737B (en) 2021-10-09 2021-10-09 Method for monitoring safety of passengers on internet about car in mobile interconnection era

Country Status (1)

Country Link
CN (1) CN114386737B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116029736A (en) * 2023-01-05 2023-04-28 浙江警察学院 Real-time detection and safety early warning method and system for abnormal track of network vehicle

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103955804A (en) * 2014-05-20 2014-07-30 中山大学 Crime risk spatial-temporal pattern recognition method serving policing prevention and control district planning
US20180032891A1 (en) * 2016-07-29 2018-02-01 International Business Machines Corporation Adaptive architecture for crash prediction in vehicle collision avoidance systems
CN109145954A (en) * 2018-07-24 2019-01-04 同济大学 A kind of net based on multisource spatio-temporal data about vehicle safety evaluation method and system
CN109299801A (en) * 2018-09-07 2019-02-01 福建兑信科技有限公司 A kind of net about vehicle lock driver automatic recognition system
CN109712387A (en) * 2018-09-26 2019-05-03 姜洪明 Net about vehicle, the passenger of taxi and driver safety system
CN109711920A (en) * 2018-11-05 2019-05-03 界首市菁华科技信息咨询服务有限公司 A kind of net about vehicle safety monitoring system
CN109767298A (en) * 2019-01-10 2019-05-17 南京邮电大学 The matched method and system of passenger's driver safety
CN110210772A (en) * 2019-06-09 2019-09-06 张子洋 A kind of net about vehicle security management and control implementation method
CN110555536A (en) * 2019-09-06 2019-12-10 南京领行科技股份有限公司 security prediction method, device, equipment and storage medium for network taxi appointment orders
CN111461185A (en) * 2020-03-19 2020-07-28 哈尔滨工程大学 Driving behavior analysis method based on improved K-means

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103955804A (en) * 2014-05-20 2014-07-30 中山大学 Crime risk spatial-temporal pattern recognition method serving policing prevention and control district planning
US20180032891A1 (en) * 2016-07-29 2018-02-01 International Business Machines Corporation Adaptive architecture for crash prediction in vehicle collision avoidance systems
CN109145954A (en) * 2018-07-24 2019-01-04 同济大学 A kind of net based on multisource spatio-temporal data about vehicle safety evaluation method and system
CN109299801A (en) * 2018-09-07 2019-02-01 福建兑信科技有限公司 A kind of net about vehicle lock driver automatic recognition system
CN109712387A (en) * 2018-09-26 2019-05-03 姜洪明 Net about vehicle, the passenger of taxi and driver safety system
CN109711920A (en) * 2018-11-05 2019-05-03 界首市菁华科技信息咨询服务有限公司 A kind of net about vehicle safety monitoring system
CN109767298A (en) * 2019-01-10 2019-05-17 南京邮电大学 The matched method and system of passenger's driver safety
CN110210772A (en) * 2019-06-09 2019-09-06 张子洋 A kind of net about vehicle security management and control implementation method
CN110555536A (en) * 2019-09-06 2019-12-10 南京领行科技股份有限公司 security prediction method, device, equipment and storage medium for network taxi appointment orders
CN111461185A (en) * 2020-03-19 2020-07-28 哈尔滨工程大学 Driving behavior analysis method based on improved K-means

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
王骜葳: "基于熵权法的网约车司机安全指数评定", 现代信息科技, vol. 5, no. 17, pages 2 - 3 *
贾步忠;: "基于网约车GPS数据的用户出行挖掘与推荐研究", 微型电脑应用, no. 12 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116029736A (en) * 2023-01-05 2023-04-28 浙江警察学院 Real-time detection and safety early warning method and system for abnormal track of network vehicle
CN116029736B (en) * 2023-01-05 2023-09-29 浙江警察学院 Real-time detection and safety early warning method and system for abnormal track of network vehicle

Also Published As

Publication number Publication date
CN114386737B (en) 2023-10-13

Similar Documents

Publication Publication Date Title
CN107798876B (en) Road traffic abnormal jam judging method based on event
CN109191896A (en) Personalized parking stall recommended method and system
Wang et al. Prediction and factor identification for crash severity: comparison of discrete choice and tree-based models
CN111242484B (en) Vehicle risk comprehensive evaluation method based on transition probability
CN106530717B (en) The dangerous state evaluation method of construction section driving based on cloud model
CN110363985B (en) Traffic data analysis method, device, storage medium and equipment
CN106919953B (en) Abnormal trip group identification method based on rail transit data analysis
US20220383738A1 (en) Method for short-term traffic risk prediction of road sections using roadside observation data
CN112116263B (en) Traffic intersection risk level assessment method and device, electronic equipment and storage medium
CN113870564B (en) Traffic jam classification method and system for closed road section, electronic device and storage medium
CN112507624B (en) Inter-city road trip mode identification model construction and identification method and device
CN110781873A (en) Driver fatigue grade identification method based on bimodal feature fusion
CN111563555A (en) Driver driving behavior analysis method and system
Oh et al. Real-time inductive-signature-based level of service for signalized intersections
CN114386737A (en) Method for monitoring safety of passengers on network appointment in mobile internet era
Padarthy et al. Identification and classification of slippery winter road conditions using commonly available vehicle variables
Wang et al. Modeling aggressive driving behavior based on graph construction
CN110490750A (en) Data know method for distinguishing, system, electronic equipment and computer storage medium
CN112749819B (en) Network appointment vehicle dispatching method, device, server and storage medium
CN111723871B (en) Estimation method for real-time carriage full load rate of bus
CN116092296B (en) Traffic state evaluation method, device, electronic equipment and storage medium
Feraud et al. Are you a good driver? A data-driven approach to estimate driving style
CN113095584A (en) Intersection safety level short-time prediction method, system, terminal and readable storage medium based on traffic flow characteristics
Kim et al. Mining traffic accident data by subgroup discovery using combinatorial targets
He et al. Research on safety evaluation of commercial vehicle driving behavior based on data mining technology

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant