CN113052206B - Road section travel time prediction method and device based on floating car data - Google Patents
Road section travel time prediction method and device based on floating car data Download PDFInfo
- Publication number
- CN113052206B CN113052206B CN202110253250.6A CN202110253250A CN113052206B CN 113052206 B CN113052206 B CN 113052206B CN 202110253250 A CN202110253250 A CN 202110253250A CN 113052206 B CN113052206 B CN 113052206B
- Authority
- CN
- China
- Prior art keywords
- road
- determining
- road section
- data
- travel time
- 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.)
- Active
Links
- 238000007667 floating Methods 0.000 title claims abstract description 74
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000009826 distribution Methods 0.000 claims abstract description 30
- 238000003062 neural network model Methods 0.000 claims abstract description 19
- 230000004927 fusion Effects 0.000 claims abstract description 16
- 238000007418 data mining Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 10
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 238000012937 correction Methods 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 210000004205 output neuron Anatomy 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Mathematical Physics (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Remote Sensing (AREA)
- Fuzzy Systems (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a road section travel time prediction method and device based on floating car data, wherein the method comprises the following steps: acquiring GPS data of a plurality of floating cars; performing data mining on GPS data of the floating car, and determining a plurality of GPS matching points corresponding to the car; estimating the travel time of the whole road section according to two continuous GPS matching points positioned on the same road section and different road sections through a distance confidence factor and/or a space-time sequence influence factor; and aiming at the same road section, carrying out data fusion on a plurality of whole road section travel times according to the distribution weight to determine estimated road section travel time, wherein the historical travel time in the same road section set is input to a neural network model, and a time-space sequence influence factor is determined. According to the method, the influence of the distance confidence and the time-space correlation is considered, the total road travel time of the vehicles in different road sections is estimated, the data fusion is further utilized, a plurality of vehicle estimates are synthesized, and the accuracy of road section travel time estimation is improved.
Description
Technical Field
The invention relates to the technical field of intelligent transportation, in particular to a road section travel time prediction method and device based on floating car data.
Background
Along with the development of science and technology, the modern intelligent traffic equipment and intelligent processing technology are utilized to monitor and master road traffic, and traffic jam nodes which generate root traffic problems are managed and controlled in a targeted manner, so that the method has important significance for the development of a traffic transportation system. The real-time estimation of urban road traffic condition is an important function of a road traffic monitoring system and is the basis of urban traffic control and congestion management.
The application of GPS (Global Positioning System ) technology to obtain effective traffic information has higher practical value in intelligent traffic system. With the continuous perfection of urban construction, the proportion of the floating car loaded with the GPS in the road network running vehicles is continuously improved. The time and space coverage rate of the GPS of the floating car is high, and the GPS has the characteristics of good real-time performance, high accuracy and the like, and can be regarded as a dynamic detector with good urban traffic state. The full utilization of the data collected by the floating car for traffic analysis is a hot spot of research in recent years, and has great potential in estimating the condition of urban road networks.
In order to know the traffic state condition in the urban road network, the road section travel time of the urban area is monitored and estimated by using the data collected by the floating cars. In the prior art, the road section travel time is estimated by adopting a traditional mathematical statistics method, probability solving and other models, however, the model construction process has certain complexity, a large amount of technical data support is needed, and the parameter solving and calibration have certain difficulty. In addition, when the road network range is large, the travel time on different road sections is different from the factor that is interfered, and a single model is difficult to characterize road conditions with different characteristics, so that the road system is difficult to be widely applied to actual traffic management and control. In summary, how to quickly and accurately estimate the travel time of a road segment is a problem to be solved.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a road segment travel time prediction method based on floating car data, so as to solve the problem of quickly and accurately estimating the road segment travel time.
The invention provides a road section travel time prediction method based on floating car data, which comprises the following steps:
acquiring floating car GPS data corresponding to a plurality of vehicles;
performing data mining on the GPS data of the floating car, and determining a plurality of GPS matching points corresponding to the car;
Estimating the travel time of the whole road section corresponding to the vehicle according to the two continuous GPS matching points positioned on the same road section and different road sections through the distance confidence factor and/or the time-space sequence influence factor;
for the same road section, carrying out data fusion on the whole road section travel time corresponding to a plurality of vehicles according to the distribution weight, and determining the corresponding estimated road section travel time;
Wherein the determining of the spatio-temporal sequence influence factor comprises: clustering different road segments through correlation among the road segments, determining a plurality of road segment sets, inputting historical travel time of the road segments in the same road segment set into a well-trained neural network model, and determining the corresponding time-space sequence influence factors.
Further, the data mining of the floating car GPS data, and determining a plurality of continuous GPS matching points corresponding to the vehicle includes:
Judging whether abnormal data and repeated data exist in the GPS data of the floating car, if so, removing, wherein the abnormal data comprise null data or messy code data;
determining a plurality of continuous GPS matching points corresponding to the GPS data of the floating car according to the GPS data of the floating car and a road network through a map matching algorithm based on weight;
Traversing the road network of the electronic map, searching paths, and determining corresponding real driving paths according to the shortest paths between two continuous GPS matching points.
Further, the determining, by the weight-based map matching algorithm, a plurality of continuous GPS matching points corresponding to the floating car GPS data according to the floating car GPS data and the road network includes:
determining a candidate road section which is closest to the GPS data of the floating car and consistent with the running direction of the car;
determining the matching weight of the candidate road section according to the projection distance and the included angle of the floating car GPS data to the candidate road section;
Selecting the matched candidate road segments according to the matching weight, and determining the best matching road segments;
And determining the corresponding GPS matching point according to the projection of the floating car GPS data on the optimal matching road section.
Further, the estimating the travel time of the whole road section corresponding to the vehicle according to the two continuous GPS matching points located in the same road section and different road sections by the distance confidence factor and/or the space-time sequence influence factor includes:
If two continuous GPS matching points are located on the same road section, determining the driving distance of the vehicle on the same road section according to the position information of the two GPS matching points;
Determining the distance confidence factor according to the ratio of the driving distance to the road section distance;
Determining a signal lamp waiting time factor according to the distribution situation of the two GPS matching points and the stop line position of the intersection;
and determining the travel time of the whole road section according to the distance confidence factor and the signal lamp waiting time factor.
Further, the estimating the travel time of the whole road section corresponding to the vehicle according to the two continuous GPS matching points located in the same road section and different road sections by the distance confidence factor and/or the space-time sequence influence factor includes:
If two continuous GPS matching points are positioned on different road sections, determining the driving distance of the vehicle on the same road section according to the position information of the two GPS matching points;
Determining the distance confidence factor according to the ratio of the driving distance to the road section distance;
Invoking the neural network model and determining the time-space sequence influence factor;
determining the corresponding allocation time of the different road segments according to the distance confidence factor and the time-space sequence influence factor;
Determining a signal lamp waiting time factor according to the distribution situation of the two GPS matching points and the stop line position of the intersection;
And determining the whole road section travel time according to the distance confidence factor, the signal lamp waiting time factor and the distribution time.
Further, for the same road section, performing data fusion on the all road section travel time corresponding to the plurality of vehicles according to the assigned weight, and determining the corresponding estimated road section travel time includes:
Traversing road network sections, and counting the total section travel time estimated correspondingly by the floating car GPS data of N vehicles for each section, wherein N is an integer;
Determining corresponding coverage degree according to the comparison of the travel distance of different vehicles on the road section and the length of the road section, and determining distribution weight according to the coverage degree;
And according to the distribution weight, fusing the total road section travel time of N vehicles, and determining the estimated road section travel time.
Further, the clustering the different road segments through the correlation among the road segments, and determining the plurality of road segment sets includes:
determining corresponding correlation according to the Pearson correlation coefficient among different road segments;
clustering and merging two road sections with highest correlation to form a plurality of road section sets;
Sequentially calculating the correlation between the rest road segments and the road segment sets, and carrying out cluster merging again until the cluster parameters between the road segment sets exceed a preset value;
And correcting a plurality of road segment sets through geometric connectivity among different road segments.
Further, the step of inputting the historical travel time of a plurality of road segments in the same road segment set to a trained neural network model, and the step of determining the corresponding time-space sequence influence factor includes:
sorting the historical travel times into an m x n two-dimensional space-time matrix having a time-to-space relationship, wherein the travel times of n road segments are input in a transverse space dimension, and m travel times in a historical time interval are input in a longitudinal time dimension;
And inputting the two-dimensional space-time matrix into the neural network model, and determining the corresponding space-time sequence influence factors.
Further, the neural network model sequentially comprises an input layer, an implicit layer, a pooling layer, a full-connection layer and an output layer.
The invention also provides a road section travel time prediction device based on the floating car data, which comprises a processor and a memory, wherein the memory is stored with a computer program, and when the computer program is executed by the processor, the road section travel time prediction method based on the floating car data is realized.
Compared with the prior art, the invention has the beneficial effects that: firstly, effectively acquiring GPS data of a floating car; then, data mining is carried out on the GPS data of the floating vehicle, a GPS matching point corresponding to the same vehicle is determined, and the vehicle track of the vehicle is fed back; then, combining a distance confidence factor and/or a time-space sequence influence factor, so that the influence of the road section correlation and the distance confidence on the travel time formed by the whole road section is fully considered, and the accurate estimation of the travel time of the whole road section by utilizing the vehicle track is achieved aiming at the conditions of the same road section and different road sections; furthermore, aiming at the same road section, data fusion is carried out by combining the corresponding estimated travel time of all road sections of different vehicle tracks on the road section, so that the estimation results of various vehicle tracks are considered, and the accuracy of estimating the travel opportunity of the road section is improved; in addition, the invention fully considers the time-space correlation between road sections, effectively determines the time-space sequence influence factors correspondingly formed in an aggregate clustering mode, and utilizes the time-space sequence influence factors to estimate the travel time of the whole road section, thereby further improving the prediction accuracy. In summary, the invention considers the influence of the distance confidence and the time-space correlation, presumes the travel time of the whole road section of the vehicles in different road sections, further utilizes the data fusion, synthesizes a plurality of vehicle estimates, and improves the accuracy of the travel time estimation of the road sections.
Drawings
FIG. 1 is a schematic flow chart of a road section travel time prediction method based on floating car data;
FIG. 2 is a schematic flow chart of determining a plurality of GPS matching points in succession according to the present invention;
fig. 3 is a schematic flow chart of step S22 in fig. 2 according to the present invention;
FIG. 4 is a schematic diagram of a process for determining travel time of a whole road segment according to the present invention;
FIG. 5 is a second flow chart for determining travel time of a whole road segment according to the present invention;
FIG. 6 is a schematic flow chart of determining estimated road travel time according to the present invention;
Fig. 7 is a schematic flow chart of determining a plurality of road segment sets according to the present invention;
Fig. 8 is a schematic flow chart of determining a timing null sequence influence factor according to the present invention.
Detailed Description
The following detailed description of preferred embodiments of the application is made in connection with the accompanying drawings, which form a part hereof, and together with the description of the embodiments of the application, are used to explain the principles of the application and are not intended to limit the scope of the application.
Example 1
The embodiment of the invention provides a road section travel time prediction method based on floating car data, and referring to fig. 1, fig. 1 is a flow chart of the road section travel time prediction method based on floating car data, and the road section travel time prediction method based on floating car data comprises steps S1 to S4, wherein:
In step S1, floating car GPS data corresponding to a plurality of vehicles are acquired;
In step S2, data mining is carried out on GPS data of the floating car, and a plurality of GPS matching points corresponding to the car are determined;
In step S3, the travel time of the whole road section corresponding to the vehicle is estimated according to two continuous GPS matching points positioned on the same road section and different road sections respectively through the distance confidence factor and/or the time-space sequence influence factor;
in step S4, for the same road section, data fusion is performed on all road section travel times corresponding to a plurality of vehicles according to the assigned weights, and corresponding estimated road section travel times are determined;
Wherein the determining of the spatio-temporal sequence influence factor comprises: clustering different road segments through correlation among the road segments, determining a plurality of road segment sets, inputting historical travel time of the road segments in the same road segment set into a well-trained neural network model, and determining corresponding time-space sequence influence factors.
In the embodiment of the invention, firstly, GPS data of a floating car is effectively acquired; then, data mining is carried out on the GPS data of the floating vehicle, a GPS matching point corresponding to the same vehicle is determined, and the vehicle track of the vehicle is fed back; then, combining a distance confidence factor and/or a time-space sequence influence factor, so that the influence of the road section correlation and the distance confidence on the travel time formed by the whole road section is fully considered, and the accurate estimation of the travel time of the whole road section by utilizing the vehicle track is achieved aiming at the conditions of the same road section and different road sections; furthermore, aiming at the same road section, data fusion is carried out by combining the corresponding estimated travel time of all road sections of different vehicle tracks on the road section, so that the estimation results of various vehicle tracks are considered, and the accuracy of estimating the travel opportunity of the road section is improved; in addition, the invention fully considers the time-space correlation between road sections, effectively determines the time-space sequence influence factors correspondingly formed in an aggregate clustering mode, and utilizes the time-space sequence influence factors to estimate the travel time of the whole road section, thereby further improving the prediction accuracy.
It should be noted that, the time and space correlation between road traffic characteristics is an important component in the travel time estimation process, and in the past, the time-space relationship solved by the machine learning algorithm is often applied to the road traffic prediction problem, so that the processing of the road travel time estimation problem is less, and the method has a certain research value. Different estimation sub-models can be adopted for processing according to different distribution conditions of GPS points in a road section, and meanwhile, the solution and calibration of model parameters are solved by using a machine learning algorithm. The machine learning model has the advantages of strong learning ability, wide coverage range and strong adaptability, is more suitable for the travel time estimation problem, and can effectively solve the complexity problem of the traditional mathematical statistics algorithm. In addition, using the same travel time estimation model for the whole road network may lead to model mismatch, but establishing an independent estimation model for each road segment has low application efficiency in actual production. In order to meet the requirements of a large-scale road network, road segments can be clustered based on the correlation of the change trend of the road segment travel time and the geographic space continuity principle, and then a proper model is selected for the data features in each clustering set, so that the travel time estimation efficiency under the large-scale road network is improved on the premise of ensuring the model accuracy.
Preferably, as seen in fig. 2, fig. 2 is a schematic flow chart of determining a plurality of continuous GPS matching points according to the present invention, and step S2 includes steps S21 to S23, where:
In step S21, judging whether abnormal data and repeated data exist in the floating car GPS data, and if so, removing, wherein the abnormal data includes null data or messy code data;
in step S22, determining a plurality of continuous GPS matching points corresponding to the floating car GPS data according to the floating car GPS data and the road network by using a weight-based map matching algorithm;
In step S23, the road network of the electronic map is traversed, a path search is performed, and a corresponding real travel path is determined according to the shortest path between two consecutive GPS matching points.
As a specific embodiment, the embodiment of the invention performs data mining on the GPS data of the floating car, ensures the accuracy of initial data, further performs effective matching on the road section where the GPS data of the floating car is positioned based on a map matching algorithm of weight, determines the GPS matching point corresponding to the same car and feeds back the car track of the car.
Preferably, as seen in fig. 3, fig. 3 is a schematic flow chart of step S22 in fig. 2 provided by the present invention, where step S22 includes steps S221 to S224, and the steps include:
In step S221, a candidate link closest to the floating car GPS data and consistent with the vehicle traveling direction is determined;
in step S222, a matching weight of the candidate road segment is determined according to the projection distance and the included angle between the floating car GPS data and the candidate road segment;
in step S223, according to the matching weight, selecting a matched candidate road segment, and determining a best matching road segment;
in step S224, corresponding GPS matching points are determined according to the projection of the floating car GPS data on the best matching road section.
As a specific embodiment, the embodiment of the invention utilizes the projection distance and the included angle between the GPS data of the floating car and the candidate road section to determine the matching weight, thereby determining the optimal best matching road section and ensuring the accuracy of the GPS matching point.
In a specific embodiment of the present invention, the step S2 specifically includes:
Traversing the investigation data, cleaning the data, and eliminating abnormal (null/messy code) data and repeated data;
And (3) accurately matching the GPS data points with the road network by using a weight-based map matching algorithm, and selecting a road section which is closer to the GPS observation point and consistent with the running direction of the vehicle for matching. And comprehensively judging the projection distance and the direction included angle of each candidate road section by using the weight proportion, selecting the road section with the highest comprehensive score as a matched road section, and taking the projection of the GPS on the road section as a final matched position point.
Wherein, the weight ws, j of the floating car GPS data p veh,s matched to the GPS matching point s egj is:
ws,j=kd×ds,j+kα×|αs,j|
Where d s,j represents the projected distance of p veh,s to the candidate segment s egj; alpha s,j represents an included angle between the running direction of the vehicle head and the candidate road section segj; k d represents a weight coefficient of the projection distance d; k α represents a weight coefficient of the direction included angle alpha;
traversing a basic road network of the electronic map, searching a path, and selecting the shortest path between two continuous matching points as a real driving path.
Preferably, as seen in fig. 4, fig. 4 is a schematic flow chart of determining a travel time of a whole road section according to the present invention, and the step S3 further includes steps S31 to S34, wherein:
in step S31, if two consecutive GPS matching points are located on the same road segment, determining a driving distance of the vehicle on the same road segment according to the position information of the two GPS matching points;
in step S32, a distance confidence factor is determined according to the ratio of the travel distance to the link distance;
in step S33, determining a signal lamp waiting time factor according to the distribution situation of the two GPS matching points and the stop line position of the intersection;
In step S34, the whole road travel time is determined according to the distance confidence factor and the traffic light waiting time factor.
As a specific embodiment, in the case that two continuous GPS matching points are located on the same road section, the embodiment of the invention determines the whole road section travel time of the vehicle on the road section according to the distance confidence factor and the signal lamp waiting time factor, and uses the travel track of the vehicle on the road section to effectively estimate.
In a specific embodiment of the present invention, the step S3 specifically includes:
Judging whether the two GPS matching points are positioned on the same road section according to the position information of the two continuous GPS matching points;
According to the position information of the two GPS matching points, solving the running length x of the vehicle on the road section, and solving the distance confidence factor eta through the ratio of the running length to the road section length;
according to the distribution situation of the two GPS matching points and the stop line position of the intersection, searching a signal lamp waiting time factor sigt corresponding to the driving road section through an empirical value;
Solving an estimated travel time value T by using a travel time correction model;
wherein, the correction model is:
Wherein sigt veh,s,j denotes the waiting time of the vehicle outside the stop line at the intersection; η veh,s,j represents a distance confidence factor.
Preferably, as seen in fig. 5, fig. 5 is a second flow chart for determining a travel time of an entire road segment according to the present invention, and the step S3 further includes steps S35 to S310, wherein:
In step S35, if two consecutive GPS matching points are located on different road segments, determining a driving distance of the vehicle on the same road segment according to the position information of the two GPS matching points;
in step S36, a distance confidence factor is determined according to the ratio of the travel distance to the link distance;
in step S37, a neural network model is called, and a time-space sequence influence factor is determined;
In step S38, determining the corresponding allocation time of different road segments according to the distance confidence factor and the time-space sequence influence factor;
in step S39, determining a signal lamp waiting time factor according to the distribution of the two GPS matching points and the stop line position of the intersection;
in step S310, the whole road travel time is determined according to the distance confidence factor, the traffic light waiting time factor, and the distribution time.
As a specific embodiment, in the embodiment of the invention, under the condition that two continuous GPS matching points are positioned on different road sections, the distribution time of the vehicle on different road sections is determined according to the distance confidence coefficient factor and the time-space sequence influence factor, and then the travel time of the vehicle on the whole road section in one road section is determined by using the distance confidence coefficient factor, the signal lamp waiting time factor and the distribution time, so that the travel track of the vehicle on the road section is effectively estimated.
In a specific embodiment of the present invention, the step S3 specifically further includes:
Judging whether the GPS matching points are positioned on two adjacent road sections or a plurality of non-adjacent road sections according to the two continuous GPS matching points;
According to the position information of the two GPS matching points, solving the running length x j of the vehicle on each road section, and solving the distance confidence factor gamma j on the road section j to be estimated according to the ratio of the running length to the road section length;
Invoking the spatiotemporal sequence influence factor beta of the current road segment presumed by the neural network model i,j
Solving a travel time value t allocated to each road section by using a travel time allocation model;
wherein, travel time distribution model is:
In the method, in the process of the invention, The distribution parameters of the space-time sequence influence and the distance scale factor are considered on the road section j to be estimated in the current moment i 0; gamma veh,s,j represents the distance confidence factor on the road section j to be estimated, which is the ratio of the vehicle driving distance x veh,s,j to the total road section length L j; beta est,j,i0 represents the estimated value of the spatio-temporal correlation of the j road segments at the current moment i 0, which takes into account the historical travel times of the previous m moments and the influence of the travel times of n adjacent road segments in the current moment i 0 on the road segment j to be estimated.
Aiming at the head and tail road sections covered by the GPS driving path, searching signal lamp waiting time factors sigt corresponding to the driving road sections through experience values according to the distribution situation of the GPS points and stop line positions of intersections;
correcting the head and tail road sections by using a travel time correction model, and estimating a road section travel time value T;
The specific formula is as follows:
Wherein sigt veh,s,j denotes the waiting time of the vehicle outside the stop line at the intersection; η veh,s,j represents a distance confidence factor.
Preferably, as seen in fig. 6, fig. 6 is a schematic flow chart of determining estimated travel time of a road segment according to the present invention, and step S4 includes steps S41 to S43, wherein:
In step S41, traversing road network segments, and counting the total road segment travel time estimated correspondingly by the floating car GPS data of N vehicles for each segment, wherein N is an integer;
In step S42, corresponding coverage degrees are determined according to the comparison between the travel distance of different vehicles on the road section and the length of the road section, and allocation weights are determined according to the coverage degrees;
in step S43, the estimated link travel time is determined by fusing all the link travel times of the N vehicles according to the assigned weights.
As a specific embodiment, the embodiment of the invention performs data fusion by using the distribution weights aiming at the same road section and combining the estimated total road section travel time corresponding to different vehicle tracks on the road section, thereby considering the estimation results of various vehicle tracks and improving the accuracy of the travel opportunity estimation of the road section.
In a specific embodiment of the present invention, the step S4 specifically includes:
traversing road network sections, and counting travel time result values obtained by estimating different travel tracks according to each section j;
on each road section, solving the distribution weight omega according to the coverage degree of the driving distance and the road section length of different tracks on the road section;
According to the road section travel time fusion model, a final road section estimation result T j,estimation is solved;
Wherein, the fusion model is:
Where ω k,s,j represents the estimated travel time value of the kth vehicle track, and T k,s,j represents the assigned weight in the total estimate T j,estimation.
Preferably, as seen in fig. 7, fig. 7 is a schematic flow chart of determining a plurality of road segment sets according to the present invention, including steps S01 to S04, wherein:
in step S01, determining a corresponding correlation according to pearson correlation coefficients between different road segments;
in step S02, two road segments with highest relativity are clustered and combined to form a plurality of road segment sets;
In step S03, calculating the correlation between the remaining road segments and the road segment sets in sequence, and performing cluster merging again until the cluster parameters between the road segment sets exceed a preset value;
in step S04, the plurality of road segment sets are modified by geometric connectivity between different road segments.
As a specific embodiment, the embodiment of the invention utilizes the Pearson correlation coefficient among different road segments to determine the space-time correlation among different road segments, so that the correlation is utilized to perform set clustering, the geometric connectivity among the road segments is utilized to correct the clustering result, the discrete non-clustered road segments are classified into a proper clustering set, so that as many adjacent road segments with high correlation can be in the same set as much as possible, and the high correlation of aggregation is ensured.
In a specific embodiment of the present invention, the process of determining the plurality of road segment sets specifically includes:
Regarding each individual road section as a class, and calculating the relevance or distance of travel time characteristic values between the classes by using the Pearson correlation coefficient;
wherein, the pearson correlation coefficient is calculated as:
wherein cov denotes the covariance of variable X and variable Y; sigma represents standard deviation; mu represents the mean; e represents a mathematical expectation;
Combining the two classes with the highest correlation or the smallest distance into a new class, and calculating the correlation or the distance between the new combined class and other residual classes again;
Repeating the above two steps until the clustering criterion (correlation or distance) between classes exceeds the set threshold 80%;
And correcting the clustering result by utilizing the geometric connectivity among the road sections, classifying the discrete non-clustered road sections into a proper clustering set, and enabling as many adjacent road sections with high correlation as possible to be in the same set.
Preferably, as seen in fig. 8, fig. 8 is a flow chart of determining a time-space sequence influence factor according to the present invention, including steps S05 to S06, wherein:
In step S05, the historical travel times are organized into an m×n two-dimensional spatio-temporal matrix having a time-to-space relationship, wherein the travel times for n road segments are input in the lateral spatial dimension and m travel times within the historical time interval are input in the longitudinal temporal dimension;
In step S06, a two-dimensional spatiotemporal matrix is input to the neural network model, and a corresponding spatiotemporal sequence influence factor is determined.
As a specific embodiment, in the embodiment of the invention, in the same clustering set, complex time-space correlation in the road network is captured by using the convolutional neural network, the mutual factors between the road networks at the current moment are deduced, the result is applied to a road section travel time distribution model, the time-space correlation between road sections is fully considered, and the time-space sequence influence factors correspondingly formed are effectively determined in a clustering mode.
Preferably, the neural network model comprises an input layer, an implicit layer, a pooling layer, a fully connected layer and an output layer in sequence. As a specific embodiment, the embodiment of the invention uses a neural network model, takes the historical travel time of different road sections as input, and efficiently and quickly obtains the corresponding time-space sequence influence factors so as to estimate the travel time.
In a specific embodiment of the present invention, in the same cluster set, complex space-time correlation in the road network is captured by using a convolutional neural network, the mutual factors between the road networks at the current moment are deduced, and the result is applied to a road section travel time distribution model, wherein the structure of the neural network model is specifically as follows:
Input layer: similar to the processing method of picture identification, the historical travel time of a plurality of road segments in the same cluster set is used as an input neuron of CNN, and the road segment travel time is organized into a two-dimensional space-time matrix with a time-space relation: the lateral space dimension inputs travel times for n road segments and the longitudinal time dimension inputs m travel times within the time interval of the history i 0-m, similar to an n x m x 1 picture.
Wherein the input unit Xi0 at the current time i 0 is represented as:
Wherein T jn,i0-m: representing travel time of the jn-th road segment in a historical i0-m time period;
Hidden layer: and the convolution layer is used for completing feature extraction, calculation and analysis of input data and outputting a feature map. The convolution layer is composed of one or more convolution kernels, which are a set of weights of fixed size, and an a×b convolution kernel W a×b can be expressed as:
wherein w a,b represents the weight value at the corresponding position in the convolution kernel;
In the convolution cross calculation process of each convolution kernel, firstly determining whether to fill the picture according to the requirement, and using the convolution kernel to move and scan the whole two-dimensional graph according to a preset step length; secondly, carrying out convolution calculation operation by using the weight in the convolution kernel and the input value of the corresponding coverage area in the picture; and finally, when one convolution kernel sweeps all position points of the whole picture and carries out convolution calculation, outputting a feature map with a new size. Assuming that the convolution layer contains K convolution kernels, the feature map h k output through the kth convolution kernel processing is:
Wherein: h k represents a feature map of the kth convolution kernel output; values of the feature map at the m-th row and n-th column positions representing the kth convolution kernel output; /(I) Representing local input values, typically a three-dimensional matrix, intercepted when the convolution kernel moves to an input matrix position (m, n); w k represents a weight matrix corresponding to the kth convolution kernel, and is generally a three-dimensional matrix; b k denotes the offset value corresponding to the kth convolution kernel. f (-) represents the activation function, which is employed in this study as the ReLU function.
Pooling layer: and (3) carrying out data feature selection and information filtering on the feature map output by convolution by using a pooling layer.
Full tie layer: and flattening the feature map, connecting the feature map with a plurality of full-connection layers, and outputting a result through activating functions such as ReLU and the like.
Output layer: and realizing data characteristic extraction, finishing dimension transformation and outputting a result processed by the system in the form of a specific value. The output neurons are travel time estimation parameter values representing space-time correlation among different road segments at the current moment in the same cluster set, the quantity Numoutput is the number n of the road segments in the cluster set, and the output layer neurons Yi0 at the current moment i0 are expressed as follows:
Wherein: y i0,jn denotes the value of the jn-th link output neuron at the current time i 0; omega i0,h',jn represents the connection weight between the h' th hidden layer neuron and the jn th output layer neuron at the current time instant i 0; b i0, jn represents the bias value of the jn-th output layer neuron at the current time instant i 0; ζ represents the ReLU activation function.
Example 2
The embodiment of the invention provides a road section travel time prediction device based on floating car data, which comprises a processor and a memory, wherein the memory is stored with a computer program, and when the computer program is executed by the processor, the road section travel time prediction method based on the floating car data is realized.
The invention discloses a road section travel time prediction method and device based on floating car data, firstly, effectively acquiring the GPS data of a floating car; then, data mining is carried out on the GPS data of the floating vehicle, a GPS matching point corresponding to the same vehicle is determined, and the vehicle track of the vehicle is fed back; then, combining a distance confidence factor and/or a time-space sequence influence factor, so that the influence of the road section correlation and the distance confidence on the travel time formed by the whole road section is fully considered, and the accurate estimation of the travel time of the whole road section by utilizing the vehicle track is achieved aiming at the conditions of the same road section and different road sections; furthermore, aiming at the same road section, data fusion is carried out by combining the corresponding estimated travel time of all road sections of different vehicle tracks on the road section, so that the estimation results of various vehicle tracks are considered, and the accuracy of estimating the travel opportunity of the road section is improved; in addition, the invention fully considers the time-space correlation between road sections, effectively determines the time-space sequence influence factors correspondingly formed in an aggregate clustering mode, and utilizes the time-space sequence influence factors to estimate the travel time of the whole road section, thereby further improving the prediction accuracy.
According to the technical scheme, a travel time correction model, a distribution model and a fusion model are designed for different distribution conditions of GPS points on road sections to realize estimation of the travel time of the road sections, in the travel time distribution model, a convolution network model is adopted to model complex space-time correlations of adjacent road sections in a road network, and under the premise of meeting basic accuracy, in order to meet travel time estimation requirements of a large number of road sections of an urban road network, model accuracy and calculation efficiency are improved. In summary, the invention considers the influence of the distance confidence and the time-space correlation, presumes the travel time of the whole road section of the vehicles in different road sections, further utilizes the data fusion, synthesizes a plurality of vehicle estimates, and improves the accuracy of the travel time estimation of the road sections.
The present invention is not limited to the above-mentioned embodiments, and any changes or substitutions that can be easily understood by those skilled in the art within the technical scope of the present invention are intended to be included in the scope of the present invention.
Claims (8)
1. A road segment travel time prediction method based on floating car data, comprising:
acquiring floating car GPS data corresponding to a plurality of vehicles;
performing data mining on the GPS data of the floating car, and determining a plurality of GPS matching points corresponding to the car;
Estimating the travel time of the whole road section corresponding to the vehicle according to the two continuous GPS matching points positioned on the same road section and different road sections through the distance confidence factor and/or the time-space sequence influence factor;
for the same road section, carrying out data fusion on the whole road section travel time corresponding to a plurality of vehicles according to the distribution weight, and determining the corresponding estimated road section travel time;
Wherein the determining of the spatio-temporal sequence influence factor comprises: clustering different road segments through correlation among the road segments, determining a plurality of road segment sets, inputting historical travel time of the road segments in the same road segment set into a well-trained neural network model, and determining the corresponding time-space sequence influence factors;
the neural network model includes an hidden layer; the hidden layer performs feature extraction, calculation and analysis on input data by using a convolution layer and outputs a feature map;
The step of carrying out data mining on the floating car GPS data, and the step of determining a plurality of GPS matching points corresponding to the car comprises the following steps:
Judging whether abnormal data and repeated data exist in the GPS data of the floating car, if so, removing, wherein the abnormal data comprise null data or messy code data;
determining a plurality of continuous GPS matching points corresponding to the GPS data of the floating car according to the GPS data of the floating car and a road network through a map matching algorithm based on weight;
Traversing a road network of the electronic map, searching paths, and determining corresponding real driving paths according to the shortest paths between two continuous GPS matching points;
The determining, by the weight-based map matching algorithm, a plurality of continuous GPS matching points corresponding to the floating car GPS data according to the floating car GPS data and the road network includes:
determining a candidate road section which is closest to the GPS data of the floating car and consistent with the running direction of the car;
determining the matching weight of the candidate road section according to the projection distance and the included angle of the floating car GPS data to the candidate road section;
Selecting the matched candidate road segments according to the matching weight, and determining the best matching road segments;
determining the corresponding GPS matching point according to the projection of the floating car GPS data on the best matching road section;
The expression of the matching weight is as follows:
Wherein w s,j represents a matching weight, d s,j represents a projection distance from the floating car GPS data to the candidate road section s egj, alpha s,j represents an included angle between the running direction of the car head and the candidate road section s egj, k d represents a weight coefficient of the projection distance d, and k α represents a weight coefficient of the direction included angle alpha.
2. The floating car data-based road segment travel time prediction method according to claim 1, wherein the estimating the total road segment travel time corresponding to the vehicle according to two consecutive GPS matching points located in the same road segment and different road segments by the distance confidence factor and/or the time-space sequence influence factor comprises:
If two continuous GPS matching points are located on the same road section, determining the driving distance of the vehicle on the same road section according to the position information of the two GPS matching points;
Determining the distance confidence factor according to the ratio of the driving distance to the road section distance;
Determining a signal lamp waiting time factor according to the distribution situation of the two GPS matching points and the stop line position of the intersection;
and determining the travel time of the whole road section according to the distance confidence factor and the signal lamp waiting time factor.
3. The floating car data-based road segment travel time prediction method according to claim 1, wherein the estimating the total road segment travel time corresponding to the vehicle according to two consecutive GPS matching points located in the same road segment and different road segments by the distance confidence factor and/or the time-space sequence influence factor comprises:
If two continuous GPS matching points are positioned on different road sections, determining the driving distance of the vehicle on the same road section according to the position information of the two GPS matching points;
Determining the distance confidence factor according to the ratio of the driving distance to the road section distance;
Invoking the neural network model and determining the time-space sequence influence factor;
determining the corresponding allocation time of the different road segments according to the distance confidence factor and the time-space sequence influence factor;
Determining a signal lamp waiting time factor according to the distribution situation of the two GPS matching points and the stop line position of the intersection;
And determining the whole road section travel time according to the distance confidence factor, the signal lamp waiting time factor and the distribution time.
4. The floating car data-based road segment travel time prediction method according to claim 1, wherein the determining the corresponding estimated road segment travel time by data fusion of the full road segment travel time corresponding to a plurality of vehicles according to the assigned weights for the same road segment comprises:
Traversing road network sections, and counting the total section travel time estimated correspondingly by the floating car GPS data of N vehicles for each section, wherein N is an integer;
Determining corresponding coverage degree according to the comparison of the travel distance of different vehicles on the road section and the length of the road section, and determining distribution weight according to the coverage degree;
And according to the distribution weight, fusing the total road section travel time of N vehicles, and determining the estimated road section travel time.
5. The floating car data based link travel time prediction method of claim 4 wherein said determining a plurality of sets of links comprises:
determining corresponding correlation according to the Pearson correlation coefficient among different road segments;
clustering and merging two road sections with highest correlation to form a plurality of road section sets;
Sequentially calculating the correlation between the rest road segments and the road segment sets, and carrying out cluster merging again until the cluster parameters between the road segment sets exceed a preset value;
And correcting a plurality of road segment sets through geometric connectivity among different road segments.
6. The floating car data-based road segment travel time prediction method according to claim 1, wherein the step of inputting the historical travel times of a plurality of road segments in the same road segment set into a trained neural network model, and the step of determining the corresponding spatio-temporal sequence influence factor comprises:
sorting the historical travel times into an m x n two-dimensional space-time matrix having a time-to-space relationship, wherein the travel times of n road segments are input in a transverse space dimension, and m travel times in a historical time interval are input in a longitudinal time dimension;
And inputting the two-dimensional space-time matrix into the neural network model, and determining the corresponding space-time sequence influence factors.
7. The floating car data based road segment travel time prediction method of claim 6, wherein the neural network model comprises an input layer, a pooling layer, a full connection layer and an output layer in this order.
8. A floating car data based road segment travel time prediction apparatus comprising a processor and a memory, the memory having stored thereon a computer program which, when executed by the processor, implements the floating car data based road segment travel time prediction method according to any one of claims 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110253250.6A CN113052206B (en) | 2021-03-04 | 2021-03-04 | Road section travel time prediction method and device based on floating car data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110253250.6A CN113052206B (en) | 2021-03-04 | 2021-03-04 | Road section travel time prediction method and device based on floating car data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113052206A CN113052206A (en) | 2021-06-29 |
CN113052206B true CN113052206B (en) | 2024-04-19 |
Family
ID=76510487
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110253250.6A Active CN113052206B (en) | 2021-03-04 | 2021-03-04 | Road section travel time prediction method and device based on floating car data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113052206B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113837442A (en) * | 2021-08-26 | 2021-12-24 | 深圳市跨越新科技有限公司 | Method and system for generating correction track oscillation route, terminal device and storage medium |
US12017680B2 (en) * | 2021-12-27 | 2024-06-25 | Here Global B.V. | Method, apparatus, and computer program product for real-time map-matching |
CN115662142B (en) * | 2022-11-21 | 2023-04-07 | 南京大学 | Deep learning traffic state estimation method and system based on sparse information of Internet of vehicles |
CN118052347B (en) * | 2024-04-16 | 2024-07-19 | 北京航空航天大学 | Travel time estimation method and system based on travel track sequence of floating car |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101270997A (en) * | 2007-03-21 | 2008-09-24 | 北京交通发展研究中心 | Floating car dynamic real-time traffic information processing method based on GPS data |
CN104575085A (en) * | 2015-01-14 | 2015-04-29 | 合肥革绿信息科技有限公司 | Public bus arrival dynamic inducing method and device based on floating buses |
CN104778274A (en) * | 2015-04-23 | 2015-07-15 | 山东大学 | Wide-range urban road network travel time estimation method based on sparse taxi GPS (Global Positioning System) data |
CN106228808A (en) * | 2016-08-05 | 2016-12-14 | 北京航空航天大学 | City expressway travel time prediction method based on Floating Car space-time grid data |
CN110070711A (en) * | 2019-04-10 | 2019-07-30 | 华砺智行(武汉)科技有限公司 | A kind of section travelling speed interval estimation system and method based on intelligent network connection car data |
CN111368881A (en) * | 2020-02-19 | 2020-07-03 | 浙江工业大学 | Low-frequency GPS track road network matching method based on multidimensional data fusion analysis |
-
2021
- 2021-03-04 CN CN202110253250.6A patent/CN113052206B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101270997A (en) * | 2007-03-21 | 2008-09-24 | 北京交通发展研究中心 | Floating car dynamic real-time traffic information processing method based on GPS data |
CN104575085A (en) * | 2015-01-14 | 2015-04-29 | 合肥革绿信息科技有限公司 | Public bus arrival dynamic inducing method and device based on floating buses |
CN104778274A (en) * | 2015-04-23 | 2015-07-15 | 山东大学 | Wide-range urban road network travel time estimation method based on sparse taxi GPS (Global Positioning System) data |
CN106228808A (en) * | 2016-08-05 | 2016-12-14 | 北京航空航天大学 | City expressway travel time prediction method based on Floating Car space-time grid data |
CN110070711A (en) * | 2019-04-10 | 2019-07-30 | 华砺智行(武汉)科技有限公司 | A kind of section travelling speed interval estimation system and method based on intelligent network connection car data |
CN111368881A (en) * | 2020-02-19 | 2020-07-03 | 浙江工业大学 | Low-frequency GPS track road network matching method based on multidimensional data fusion analysis |
Non-Patent Citations (1)
Title |
---|
Utilizing Artificial Neural Network in GPS-Equipped Probe Vehicles Data- Based Travel Time Estimation;MENGYUN XU 等;IEEE;第89412-89426页 * |
Also Published As
Publication number | Publication date |
---|---|
CN113052206A (en) | 2021-06-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113052206B (en) | Road section travel time prediction method and device based on floating car data | |
US11131993B2 (en) | Methods and systems for trajectory forecasting with recurrent neural networks using inertial behavioral rollout | |
CN112356830B (en) | Intelligent parking method based on model reinforcement learning | |
CN106022541B (en) | Arrival time prediction method | |
JP7263216B2 (en) | Object Shape Regression Using Wasserstein Distance | |
CN110488842B (en) | Vehicle track prediction method based on bidirectional kernel ridge regression | |
CN113591380B (en) | Traffic flow prediction method, medium and equipment based on graph Gaussian process | |
Piccoli et al. | Fussi-net: Fusion of spatio-temporal skeletons for intention prediction network | |
CN108564326A (en) | Prediction technique and device, computer-readable medium, the logistics system of order | |
CN112078592B (en) | Method and device for predicting vehicle behavior and/or vehicle track | |
WO2019232335A1 (en) | Methodologies, systems, and components for incremental and continual learning for scalable improvement of autonomous systems | |
CN112071062A (en) | Driving time estimation method based on graph convolution network and graph attention network | |
CN112598165A (en) | Private car data-based urban functional area transfer flow prediction method and device | |
CN116503446A (en) | Multi-mode vehicle track prediction method for target driving and distribution thermodynamic diagram output | |
CN113903173B (en) | Vehicle track feature extraction method based on directed graph structure and LSTM | |
CN114495036A (en) | Vehicle track prediction method based on three-stage attention mechanism | |
CN114299607A (en) | Human-vehicle collision risk degree analysis method based on automatic driving of vehicle | |
CN111310919B (en) | Driving control strategy training method based on scene segmentation and local path planning | |
Cardoso et al. | A large-scale mapping method based on deep neural networks applied to self-driving car localization | |
CN117232545A (en) | Path planning method based on deep learning road environment perception | |
CN117037479A (en) | Signal transmission system for measuring traffic state by using road network sensor | |
CN115905434B (en) | Road network track completion method based on learning interpolation prediction | |
WO2023242223A1 (en) | Motion prediction for mobile agents | |
CN114916013B (en) | Edge task unloading delay optimization method, system and medium based on vehicle track prediction | |
CN112907953B (en) | Bus travel time prediction method based on sparse GPS data |
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 |