CN109285346A - A kind of city road net traffic state prediction technique based on key road segment - Google Patents
A kind of city road net traffic state prediction technique based on key road segment Download PDFInfo
- Publication number
- CN109285346A CN109285346A CN201811043484.2A CN201811043484A CN109285346A CN 109285346 A CN109285346 A CN 109285346A CN 201811043484 A CN201811043484 A CN 201811043484A CN 109285346 A CN109285346 A CN 109285346A
- Authority
- CN
- China
- Prior art keywords
- section
- matrix
- road network
- road
- correlation
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 23
- 239000011159 matrix material Substances 0.000 claims abstract description 107
- 230000002123 temporal effect Effects 0.000 claims abstract description 51
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000013459 approach Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 238000012512 characterization method Methods 0.000 claims description 5
- 230000003203 everyday effect Effects 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 230000002596 correlated effect Effects 0.000 claims description 3
- 238000010219 correlation analysis Methods 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000005192 partition Methods 0.000 claims description 3
- 210000005036 nerve Anatomy 0.000 claims 1
- 238000009412 basement excavation Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 9
- 238000013135 deep learning Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- RTAQQCXQSZGOHL-UHFFFAOYSA-N Titanium Chemical compound [Ti] RTAQQCXQSZGOHL-UHFFFAOYSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Operations Research (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Entrepreneurship & Innovation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Evolutionary Computation (AREA)
- Game Theory and Decision Science (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
City road net traffic state prediction technique this patent discloses a kind based on key road segment, which is characterized in that the described method includes: step 1: data prediction;Step 2: road network Spatial weight matrix is established;Step 3: settling time correlation matrix;Step 4: key road segment is identified using temporal correlation matrix;Step 5: establishing depth convolutional neural networks, predicts the following road network state, and evaluate prediction model;The present invention predicts urban transportation stream mode from a wide range of road network level, be conducive to from the temporal and spatial correlations characteristic for macroscopically induce to traffic flow sufficiently excavation traffic flow, pass through the key road segment in identification road network, compared to using the historic state in all sections as input data, the training time of model can be greatly reduced, improve forecasting efficiency;Using convolutional neural networks as prediction model, prediction result is also more accurate.
Description
Technical field
The present invention relates to field of traffic.It is specifically a kind of that pass is identified from city road network using temporal correlation algorithm
Key section, thus the method predicted road network entirety traffic flow modes.
Background technique
The quickening of Urbanization in China and the sustainable growth of vehicle guaranteeding organic quantity, so that urban traffic network congestion problems
It gets worse, becomes one of an important factor for hindering city healthy and rapid development.City road network short-term traffic flow is carried out real-time
Prediction provides real-time reliable route for traveler, improves out line efficiency, and induce traffic behavior.It meanwhile being management
Traffic-information service, traffic guidance, traffic control and the alleviation of traffic jam issue of department provide strong technology branch
It holds.
With intelligent transport technology (Intelligent Traffic System, ITS) fast development and it is universal, from receipts
The huge traffic data for taking the acquisition of the terminals such as station, bayonet, video detector, mobile phone expedites the emergence of the change of traffic system wisdomization, for not
The development for carrying out wisdom traffic provides data basis.In addition, with intelligent algorithm rise and traffic forecast field at
Function application, the data-driven model based on deep learning, also to see clearly, understand, predict that complicated traffic system provides algorithm
Support.In conjunction with big data and artificial intelligence technology, it is applied to the new fields such as traffic situation real-time perception and prediction, traffic cloud computing
Scape is road network Tendency Prediction and analysis, provides new means.
Short-time traffic flow forecast be always domestic and foreign scholars research hot spot, existing method be broadly divided into parameter model,
Nonparametric model, artificial intelligence approach, built-up pattern etc..Parameter model is fairly simple, is easy to construct, but for complicated and changeable
Traffic flow, there are hysteresis, are unable to solution of emergent event;Nonparametric model simple structure, but it is not suitable for the non-of traffic flow
Linearly, uncertain feature;Artificial intelligence approach can adaptive complicated, nonlinear traffic system, and precision of prediction is higher,
But the training of model needs mass data, and the training time is longer;Built-up pattern combines the advantages of a variety of models, but constructs
It is complex.Wherein, deep learning is a kind of method of learning characteristic automatic from big data, and self-learning capability is very strong,
It is successfully applied to image, speech recognition etc., certain achievement is also achieved in traffic status prediction.
The features such as the scale of traffic big data is big, type is more, value density is low are determined and are contained in massive information largely
Redundancy.In research currently with deep learning prediction road network state, is directly predicted, can be made using mass data
Model is complicated but precision improvement is not significant.It finds under study for action, traffic flow shows apparent correlation over time and space
Property, part road is affected to approach way or local network.The present invention proposes to identify road using temporal correlation algorithm
Key road segment in net, and whole road network state is predicted using the traffic parameter of key road segment.This method fully considers
The temporal correlation of traffic flow, can be under the premise of ensuring precision, and simplified model input improves forecasting efficiency, well
It overcomes existing utilization deep learning and predicts that the training time present in road network state is long, biggish deficiency is relied on to data volume.
Summary of the invention
It is an object of the invention to overcome the problems, such as that the training time present in current deep learning is long, forecasting efficiency is low,
The temporal and spatial correlations characteristic for sufficiently excavating road grid traffic, identifies the critical path being affected to approach way and Regional Road Network state
Section proposes a kind of prediction technique using key road segment space-time characterisation prediction road net traffic state.The present invention utilizes floating driving skills
Art first cleans initial data, obtains the topological relation in road network between each road average-speed and section;Then,
Utilize the key road segment in temporal correlation algorithm identification road network;Finally, deep learning network is built, by the speed of key road segment
Spatio-temporal state matrix is converted into as input, prediction model is trained, to predict the traffic flow shape of the following whole road network
State, and model is evaluated.This method considers the temporal correlation of road network, is sufficiently dug to the space-time characterisation of road network
Pick identifies the key road segment in road network, and predicts based on this whole road network state have innovation in theoretical side
Property, and there is stronger directive significance to practical.
For achieving the above object, the present invention adopts the following technical scheme that:
Step 1: data prediction.Initial data is cleaned, the average speed in each section is calculated, is matched
Onto section, and select road network to be studied.
To obtain accurate floating car data, it is necessary first to initial data is pre-processed, in simple terms, deletion error
Data fill up missing data using linear interpolation method.
Secondly, calculating average speed of the section in each period, calculation method are as follows: within some period, certain
The mean value of all average vehicle speeds in section.Calculation formula is such as shown in (1)
In formula: vX, jThe section x is represented in the road average-speed of j-th of period, x ∈ (1,2 ..., l) is l that is, shared
Section.Vehicle i is represented in the average speed of Δ t, n represents the quantity of section vehicle.It can be acquired by (2) formula.
In formula: the length in the behalf section, Δ t represent the length of period.
The velocity amplitude being calculated is matched on section, so that there is average speed in each section in each period
Value.
Step 2: road network Spatial weight matrix is established.According to complex network network theory, closed using the topology between section
System, establishes k rank adjacency matrix between section, and establish road network Spatial weight matrix accordingly.
When x and y are connected directly when side, it is adjacent for single order to define the relationship on both sides.Similarly, second order adjoining can be described as one
The adjacent single order of rank is adjacent, and so on, it can establish the k rank adjacency matrix in certain section.Syntople can be described as:
In formula: ωX, yThe value range of x-th of the expression neighbouring relations between y-th of section, x and y are [1, l],
It is iterated between l section.
Urban road network can be abstracted as non-directed graph and digraph, if only considering static network topology, for non-directed graph;
It is digraph if considering traffic flow.In the method, it has fully considered the flow direction of traffic flow, therefore, digraph should have been used
Mode be abstracted to obtain the topological relation of road network, also can more truly react the spatial coherence of traffic flow in this way.With Fig. 2
It is illustrated with for Fig. 3, gives a network being made of 8 sides and 9 nodes, when network is non-directed graph, be based on
The single order and second order adjacency matrix on side are as shown in Figure 1;When network is digraph, adjacency matrix is as shown in Figure 2.
In entire road network, when it is contemplated that 1 to k rank adjacency matrix of spatial object can be established when k rank approach way,
These matrixes are summed up, so that it may obtain the Spatial weight matrix of entire road network.When summing up, it is contemplated that with section
Distance it is remoter, the weaker feature of mutual spatiotemporal influence needs the adjacency matrix to each rank to be assigned to weight in adduction.Power
The distribution of weight can be obtained by formula (4).
In formula: qiIndicate that the weight of the i-th rank adjacency matrix, the value range of i are [1, k], k is the order of adjacency matrix.
The Spatial weight matrix of spatial coherence can be calculated by formula (5) between characterization section after then weighting:
In formula: W is the Spatial weight matrix after weighting, qiIndicate the weight of the i-th rank adjacency matrix, Q(i)For i rank spatial moment
Battle array, the value range of i are [1, k].
Step 3: settling time correlation matrix.Using in Pearson relevance function measurement road network between section when
Between correlation, obtain the temporal correlation matrix of road network.
Specifically, correlation analysis is carried out to the average speed of road network Road Duan Yitian, obtains each link flow in the time
Upper correlativity (related coefficient), then the foundation using related coefficient as evaluation measures the correlation of section in time.Two
The time correlation coefficient in section can be calculated by formula (6).
In formula: γX, yIndicate the time correlation property coefficient of section x and y, wherein X=[x1, x2..., xz] it is the sampling period to be
The one day average speed vector in the section p, x, Y=[y1, y2..., yz] it is the one day average speed vector in the section y, X, Y in section
Element number z it is related with the sampling period,It is the arithmetic average of vector interior element respectively.And in practical road network,
γX, yValue range be [- 1,1], negative value indicates that negatively correlated trend is presented in the section x and the section y, and closer -1 indicates phase
Closing property embodies more significant.For convenience of calculation, need to γX, yValue take absolute value, i.e. γX, yValue closer to 0, show two
The correlation in section is not significant;Closer to 1, indicate that the correlation in two sections is more significant.
Using the traffic flow modes in two sections of Pearson relevance function measurement in intraday temporal correlation, and build
Correlation matrix T indicates the temporal correlation of road network between immediately.
The temporal correlation between two section of element representation in matrix T.For next step convenience of calculation, section is not considered
Autocorrelation, i.e. the diagonal element γ of matrix T1,1、γ2,2、γL, lIt is all zero.Due in the historical data, the traffic of every day
The feature that stream mode is presented has difference, therefore in calculating, for the average speed of every day, settling time correlation matrix
T1, T2..., Td, i.e. d days historic states correspond to the different temporal correlation matrixes of d.
Step 4: key road segment is identified using temporal correlation matrix.
Spatial weight matrix and the temporal correlation matrix of some day are subjected to dot product, obtain the Kongxiang at this day of road network
Closing property matrix, it may be assumed that
Ha=WTa (8)
In formula: HaIndicate that a days spatio-temporal state matrixes, W indicate the space weight of road network after weighting, TaIt indicates a days
Temporal correlation matrix, the value range of a are [1, d].Then HaFor the matrix of a l × l, the value model of each element in matrix
Enclosing indicates that temporal correlation is more significant for [0,1], closer 1.
After obtaining a days road network temporal correlation matrixes, temporal correlation index (i.e. each column in each section are counted
Numerical value) between 0 to 1, with 0.1 for step-length frequency.By taking the section x as an example, the temporal correlation index in the section x, the i.e. section x
The relevance parameter of corresponding column, total l (wherein the auto-correlation coefficient in the section x has been set as 0).Count l parameter [0,
0.1], (0.1,0.2] ..., (0.9,1.0] frequency in each section.Between section under all historic states of comprehensive consideration when
The influence of empty correlation is counting section after the frequency disribution in each day, is being summed up, available in history shape
Certain section is denoted as f in the sum of the frequency in each section under state1、f2、…、f10, respectively represent certain section [0,0.1], (0.1,
0.2] ..., (0.9,1.0] frequency in each section.The evaluation criterion of key road segment is the weighted sum of each section frequency, it may be assumed that
ca=0.95 × f10+0.85×f9+0.75×f8+0.65×f7 (9)
In formula: caIndicate the criticality of section a, f10、f9、f8、f7Respectively represent section a (0.9,1.0], (0.8,
0.9], (0.7,0.8], (0.6,0.7] frequency in section and.
Pass through criticality ca, the criticality in all sections can be ranked up, and according to the percentage of section sum, mention
It takes the preceding section of sequence as key road segment, extracts input value of the average speed as prediction model of key road segment.
Step 5: establishing depth convolutional neural networks (Convolutional Neural Network, CNN), and prediction is not
Incoming road net state, and prediction model is evaluated.
It firstly, establishing sample set, and is training set and test set according to the ratio cut partition of 2:1.
Secondly, extracting the average speed data of key road segment from historical data, and it is translated into spatio-temporal state matrix,
Input as prediction model.Specifically, by the average speed in all sections using abscissa be section number, ordinate as when
Between form be organized into matrix.Later, it is modeled using neural convolutional network, steps are as follows:
(1) input/output variable is determined.Input variable is the spatio-temporal state matrix of the key road segment in road network, input layer mind
It is identical as the number of key road segment through first number;Output variable is the integrality matrix of the following road network, comprising owning in road network
The average speed in section, output layer neuron number are identical as all section numbers of road network.
(2) the deep layer convolutional neural networks model for constructing " end to end ", is arranged the structural parameters of convolutional neural networks.
Under the input of training set, the convolutional neural networks of " input layer-convolutional layer-pond layer-output layer " structure are trained, and
The output valve of output layer and the difference of true value are measured with loss function, are obtained most using BPTT algorithmic minimizing loss function
Excellent model parameter.
(3) following a period of time road network integrality is predicted.Key road segment average speed is extracted from test set,
It is converted into spatio-temporal state matrix, input second step in trained prediction model, obtains output vector, when as next
Between section road network entirety traffic flow modes.
Finally, establishing with root-mean-square error (Mean Square Error, MSE) and root-mean-square error (Root Mean
Square Error, RMAE) be standard prediction model assessment indicator system, the precision of prediction of model is evaluated.MAE and
The calculation method of RMSE such as formula (10) and (11).
In formula: u represents fixed number, and l is section number, npRepresentative is fixed number, np=l × u.
Advantages of the present invention:
(1) present invention predicts urban transportation stream mode from a wide range of road network level, can be on the whole to city
Traffic situation develop and carry out control, be conducive to from macroscopically inducing traffic flow;
(2) the temporal and spatial correlations characteristic for sufficiently excavating traffic flow is improved defeated when prediction by the key road segment in identification road network
Enter the quality of data, it is often more important that, compared to using the historic state in all sections as input data, mould can be greatly reduced
The training time of type improves forecasting efficiency;
(3) using convolutional neural networks as prediction model, compared to traditional parameter model, nonparametric model etc., energy
Enough preferably adaptive traffic flow complexity, nonlinear feature, have better robustness, prediction result is also more accurate.
Detailed description of the invention
Fig. 1 is non-directed graph adjacency matrix schematic diagram
Fig. 2 is digraph adjacency matrix schematic diagram
Fig. 3 is to pass through pretreated road average-speed
Fig. 4 is area to be studied road network
Fig. 5 is road network single order and second order adjacency matrix (part)
Fig. 6 is road network Spatial weight matrix (part)
Fig. 7 temporal correlation matrix (part) between section on June 1
Fig. 8 temporal correlation matrix (part) between August section on the 1st
Fig. 9 temporal correlation matrix (part) between section on June 1
Figure 10 is that section criticality sorts (part)
Figure 11 is key road segment distribution in road network
Figure 12 is spatio-temporal state matrix conversion schematic diagram
Figure 13 is CNN prediction model structure
Figure 14 is MSE with frequency of training change curve
Figure 15 is this method flow chart.
Specific embodiment
Specific introduction is done to the present invention below with reference to real data.It should be noted that the data used are certain companies
The some region of floating car data in Beijing of offer includes section number 278, and the sample frequency of data is 2 minutes, includes
6,7,8 three months totally 92 days data.For convenience of calculation, the data of daily 6:00 to 23:00 are screened, that is, have eliminated traffic flow
Measure lesser night run the period.
Realization route of the invention includes several steps:
Step 1: data prediction.Initial data is cleaned, the average speed in each section is calculated, is matched
Onto section, and select road network to be studied.
To obtain accurate floating car data, it is necessary first to initial data is pre-processed, in simple terms, deletion error
Data fill up missing data using linear interpolation method.
Secondly, calculating average speed of the section in each period, calculation method are as follows: within some period, certain
The mean value of all average vehicle speeds in section.Calculation formula is such as shown in (12)
In formula: vX, jThe section x is represented in the road average-speed of j-th of period, x ∈ (1,2 ..., 278).Represent vehicle
In the average speed of Δ t, n represents the quantity of section vehicle by i.It can be acquired by (13) formula.
In formula: the length in the behalf section, Δ t represent the length of period.
The velocity amplitude being calculated is matched on section, so that there is average speed in each section in each period
Value.By pretreated input as shown in figure 3, including three time dt, section number linkid, speed speed (km/h) words
Section.
In the present invention, by the road network of 278 sections composition as research object, the schematic diagram of road network is as shown in Figure 4.
Step 2: road network Spatial weight matrix is established.According to complex network network theory, closed using the topology between section
System, establishes 5 rank adjacency matrix between section, and establish road network Spatial weight matrix accordingly.
When x and y are connected directly when side, it is adjacent for single order to define the relationship on both sides.Similarly, second order adjoining can be described as one
The adjacent single order of rank is adjacent, and so on, it can establish the 5 rank adjacency matrix in certain section.Syntople can be described as:
In formula: ωX, yIndicate x-th of neighbouring relations between y-th of section, the value range of x and y be [1,
278], i.e., it is iterated between 278 sections.
Urban road network can be abstracted as non-directed graph and digraph, if only considering static network topology, for non-directed graph;
It is digraph if considering traffic flow.In the method, it has fully considered the flow direction of traffic flow, therefore, digraph should have been used
Mode be abstracted to obtain the topological relation of road network, also can more truly react the spatial coherence of traffic flow in this way.According to
The topological relation of road network, the single order of foundation and second order adjacency matrix part are as shown in Figure 5.(section number is reduced to 1,2 ... ...)
In this example, it is contemplated that when 5 rank approach way, 1 to 5 rank adjacency matrix of spatial object is established, to these squares
Battle array sums up, so that it may obtain the Spatial weight matrix of entire road network.When summing up, it is contemplated that got over at a distance from section
Far, the weaker feature of mutual spatiotemporal influence needs the adjacency matrix to each rank to be assigned to weight in adduction.The distribution of weight
It can be obtained by formula (15).
In formula: qiIndicate that the weight of the i-th rank adjacency matrix, the value range of i are [1,5].
The Spatial weight matrix of spatial coherence can be calculated by formula (16) between characterization section after then weighting:
In formula: W is the Spatial weight matrix after weighting, qiIndicate the weight of the i-th rank adjacency matrix, Q(i)For i rank spatial moment
Battle array, the value range of i are [1,5].Fig. 6 is the segment space weight matrix after the network power.
Step 3: settling time correlation matrix.Using in Pearson relevance function measurement road network between section when
Between correlation, obtain the temporal correlation matrix of road network.
Specifically, correlation analysis is carried out to the average speed of road network Road Duan Yitian, obtains each link flow in the time
Upper correlativity (related coefficient), then the foundation using related coefficient as evaluation measures the correlation of section in time.Two
The time correlation coefficient in section can be calculated by formula (17).
In formula: γX, yIndicate the time correlation property coefficient of section x and y, wherein X=[x1, x2..., x481] it is the sampling period
It is 2 minutes, the average speed vector of the section x 6:00-23:00, Y=[y1, y2..., y481] it is the section y 6:00-23:00 in section
Average speed vector,It is the arithmetic average of vector interior element respectively, the value range of x and y are [1,278].And
In practical road network, γX, yValue range be [- 1,1], negative value indicates that negatively correlated trend is presented in the section x and the section y, and more
It is embodied close to -1 expression correlation more significant.For convenience of calculation, need to γX, yValue take absolute value, i.e. γX, yValue get over
Close to 0, show that the correlation in two sections is not significant;Closer to 1, indicate that the correlation in two sections is more significant.
Using the traffic flow modes in two sections of Pearson relevance function measurement in intraday temporal correlation, and build
Correlation matrix T indicates the temporal correlation of road network between immediately.
The temporal correlation between two section of element representation in matrix T.For next step convenience of calculation, section is not considered
Autocorrelation, i.e. the diagonal element γ of matrix T1,1、γ2,2、γ278,278It is all zero.Due in the historical data, the friendship of every day
The feature that open position is presented has difference, therefore in calculating, for the average speed of every day, settling time correlation square
Battle array T1, T2..., T92, i.e., the corresponding 92 different temporal correlation matrixes of 92 days historic states.Fig. 7 and Fig. 8 be June 1 and
August two days on the 1st part way temporal correlation matrixes.
Step 4: key road segment is identified using temporal correlation matrix.
Spatial weight matrix and the temporal correlation matrix of some day are subjected to dot product, obtain the Kongxiang at this day of road network
Closing property matrix, it may be assumed that
Ha=WTa (19)
In formula: HaIndicate that a days spatio-temporal state matrixes, W indicate the space weight of road network after weighting, TaIt indicates a days
Temporal correlation matrix, the value range of a are [1,92].Then HaFor one 278 × 278 matrix, each element in matrix
The value range of value is [0,1], indicates that temporal correlation is more significant closer to 1.Kongxiang when Fig. 9 is part way on the 1st in June
Closing property matrix.
After obtaining a days road network temporal correlation matrixes, the temporal correlation index for counting each section is (i.e. each
The numerical value of column) between 0 to 1, with 0.1 for step-length frequency.By taking the section x as an example, the temporal correlation index in the section x, the i.e. road x
The corresponding relevance parameter arranged of section, totally 278 (wherein the auto-correlation coefficient in the section x has been set as 0).Count 278 parameters
[0,0.1], (0.1,0.2] ..., (0.9,1.0] frequency in each section.Between section under all historic states of comprehensive consideration
Temporal correlation influence, counting section after the frequency disribution in each day, summed up, it is available to go through
Certain section is denoted as f in the sum of the frequency in each section under history state1、f2、…、f10, respectively represent certain section [0,0.1],
(0.1,0.2] ..., (0.9,1.0] frequency in each section.The evaluation criterion of key road segment is the weighted sum of each section frequency, it may be assumed that
ca=0.95 × f10+0.85×f9+0.75×f8+0.65×f7 (20)
In formula: caIndicate the criticality of section a, f10、f9、f8、f7Respectively represent section a (0.9,1.0], (0.8,
0.9], (0.7,0.8], (0.6,0.7] frequency in section and.
Pass through criticality ca, the criticality in all sections can be ranked up.Figure 10 illustrates criticality in road network and arranges
9 section, is numbered according to section before sequence, can extract key road segment and its velocity information.In this example, preceding 60% is extracted,
As shown in red road segments in Figure 11, the average speed of key road segment is extracted as key road segment in the forward section of i.e. 166 sequences
Input value as prediction model.
Step 5: establishing depth convolutional neural networks (Convolutional Neural Network, CNN), and prediction is not
Incoming road net state, and prediction model is evaluated.
It firstly, establishing sample set, and is training set and test set, i.e. 6, totally 61 days July of extraction according to the ratio cut partition of 2:1
Data as training set, August totally 31 days data as test set.
Secondly, extracting the average speed data of key road segment from historical data, and it is translated into spatio-temporal state matrix,
Input as prediction model.Specifically, by the average speed in all sections using abscissa be section number, ordinate as when
Between form be organized into matrix.It is spatio-temporal state matrix that Figure 12, which is illustrated the rate conversion of key road segment, and carrying out to it can
It is shown depending on changing.As can be seen that the traffic flow speed in section shows very strong temporal correlation.
Later, it is modeled using neural convolutional network, steps are as follows:
(1) input/output variable is determined.Input variable is the spatio-temporal state matrix of the key road segment in road network, input layer mind
It is identical as the number of key road segment through first number, i.e., 166;Output variable is the integrality matrix of the following road network, includes road network
In all sections average speed, output layer neuron number is identical as all section numbers of road network, i.e., 278.The structure of model
Schematic diagram is as shown in figure 13, and left side input is the traffic flow speed information of key road segment, and right side output is the traffic of road network entirety
Stream mode.Intermediate training and prediction is based on deep layer convolutional neural networks.
(2) the deep layer convolutional neural networks model for constructing " end to end ", is arranged the structural parameters of convolutional neural networks,
Specific structure is as shown in table 1.Under the input of training set, to the convolution of " input layer-convolutional layer-pond layer-output layer " structure
Neural network is trained, and the output valve of output layer and the difference of true value are measured with loss function, utilizes BPTT algorithm
It minimizes loss function and obtains optimal model parameter.
The setting of 1 CNN model parameter of table
(3) following a period of time road network integrality is predicted.Key road segment average speed is extracted from test set,
It is converted into spatio-temporal state matrix, input second step in trained prediction model, obtains output vector, when as next
Between section road network entirety traffic flow modes.
Finally, establishing with root-mean-square error (Mean Square Error, MSE) and root-mean-square error (Root Mean
Square Error, RMAE) be standard prediction model assessment indicator system, the precision of prediction of model is evaluated.MAE and
The calculation method of RMSE such as formula (21) and (22).
In formula: u=79846, l=278, npRepresentative is fixed number, np=l × u=22197188.
This example is using the traffic flow velocity after road network 2 minutes of 166 key road segment prediction of speed, 278 sections composition
Degree.In the training process of model, MSE is gradually becoming smaller with the increase of train epochs, as shown in figure 14, i.e. the prediction of model
Precision is also being continuously improved.
It is predicted using integrality of 166 key road segments to 278 sections of road network, last precision of prediction is
RMSE=6.64, the training time of model are 370 seconds (being based on four road TITAN Xp video card concurrent operations).In identical model layer
Number is hidden under the setting of layer parameter (being shown in Table 1) and frequency of training (100 times), and does not use key road segment, that is, uses 278 roads
The average speed of section carries out prediction to the state behind 278 sections 2 minutes and compares, and precision of prediction improves 1.04%, training institute
The time needed reduces 10.62%.
Table 2 is compared using key road segment prediction effect
As it can be seen that this method is not much different with former methodical precision of prediction, or even under the premise of increasing, Neng Gouti
The training effectiveness of high model.It is more larger when city road network, the advantage of this method will embody it is more obvious, thus more
Add and rapidly the state of the following road network is predicted, be very suitable to the on-line training of traffic behavior, predicts in real time.
Claims (2)
1. a kind of city road net traffic state prediction technique based on key road segment, which is characterized in that the described method includes:
Step 1: data prediction
Initial data is cleaned, the average speed in each section is calculated, matches it on section, and selection target road
Net;The data prediction includes deletion error data, is filled up to missing data using linear interpolation method;And it calculates
Average speed of the section in each period;
Step 2: road network Spatial weight matrix is established
The topological relation of road network is obtained come abstract by the way of digraph, and using the topological relation between section, establishes road
K rank adjacency matrix between section, and road network Spatial weight matrix is established accordingly;When x and y are connected directly when side, the relationship on both sides is defined
It is adjacent for single order;The adjacent adjacent single order of single order that can be described as of second order is adjacent, and so on, the k rank for establishing certain section is adjacent
Matrix;Syntople description are as follows:In formula: ωx,yX-th of expression adjacent between y-th of section
The value range of relationship, x and y are [1, l], i.e., are iterated between l section;
In entire road network, when considering k rank approach way, 1 to k rank adjacency matrix of spatial object can be established, to these matrixes
It sums up, obtains the Spatial weight matrix of entire road network;When summing up, based on it is more remote mutual at a distance from section when
Sky influences weaker feature, is assigned to weight in adduction to the adjacency matrix of each rank;The distribution of weight by?
It arrives, in which: qiIndicate that the weight of the i-th rank adjacency matrix, the value range of i are [1, k], k is the order of adjacency matrix;Then weight
The Spatial weight matrix of spatial coherence can be by formula between characterization section afterwards It is calculated, in formula: W is
Spatial weight matrix after weighting, qiIndicate the weight of the i-th rank adjacency matrix, Q(i)For i rank space matrix, the value range of i is
[1,k];
Step 3: settling time correlation matrix
Using the temporal correlation between section in Pearson relevance function measurement road network, the temporal correlation of road network is obtained
Matrix;
The one day average speed in section carries out correlation analysis in road network, obtains each link flow related coefficient in time, then
The correlation of foundation measurement section in time using related coefficient as evaluation;The time correlation coefficient in two sections can be by formulaIt calculates, in formula: γx,yIndicate the time correlation property coefficient of section x and y, wherein X=[x1,
x2,…,xz] be the sampling period be p, the one day average speed vector in the section x, Y=[y1,y2,…,yz] it is the section y one in section
It average speed vector, the element number z of X, Y and sampling period are related,It is the arithmetic mean of vector interior element respectively
Value;γx,yValue range be [- 1,1], negative value indicates that negatively correlated trend is presented in the section x and the section y, and closer -1 indicates
Correlation embodies more significant;To γx,yValue take absolute value, i.e. γx,yValue closer to 0, show the correlation in two sections not
Significantly;Closer to 1, indicate that the correlation in two sections is more significant;The traffic flow in two sections is measured using Pearson relevance function
State is in intraday temporal correlation, and settling time correlation matrix T indicates the temporal correlation of road networkThe diagonal element γ of matrix T1,1、γ2,2、γl,lIt is all zero, in calculating, for being averaged for every day
Speed, settling time correlation matrix T1,T2,…,Td, i.e. d days historic states correspond to the different temporal correlation matrixes of d;
Step 4: key road segment is identified using temporal correlation matrix.
Spatial weight matrix and the temporal correlation matrix of some day are subjected to dot product, obtain road network in this day temporal correlation
Matrix Ha=WTaIn formula: HaIndicate that a days spatio-temporal state matrixes, W indicate the space weight of road network after weighting, TaIndicate the
A days temporal correlation matrixes, the value range of a are [1, d];HaFor the matrix of a l × l, each element is taken in matrix
Being worth range is [0,1], indicates that temporal correlation is more significant closer to 1;
After obtaining a days road network temporal correlation matrixes, the temporal correlation index in each section, i.e., the number of each column are counted
Value, between 0 to 1, with 0.1 for step-length frequency;Section is being counted after the frequency disribution in each day, is being summed up
It obtains certain section under historic state and is denoted as f in the sum of the frequency in each section1、f2、…、f10, respectively represent the section and exist
[0,0.1], (0.1,0.2] ..., (0.9,1.0] each section frequency;The evaluation criterion of key road segment is that each section frequency adds
Quan He, ca=0.95 × f10+0.85×f9+0.75×f8+0.65×f7In formula: caIndicate the criticality of section a, f10、f9、f8、
f7Respectively represent section a (0.9,1.0], (0.8,0.9], (0.7,0.8], (0.6,0.7] frequency in section and;Pass through pass
Key degree ca, the criticality in all sections is ranked up, and according to the percentage of section sum, extracts preceding section of sorting and make
For key road segment, input value of the average speed as prediction model of key road segment is extracted;
Step 5: establishing depth convolutional neural networks, predicts the following road network state, and evaluate prediction model;Firstly, building
Vertical sample set, and be training set and test set according to the ratio cut partition of 2:1;Secondly, extracting key road segment from historical data
Average speed data, and it is translated into spatio-temporal state matrix, the input as prediction model;Input variable is the pass in road network
The spatio-temporal state matrix in key section, input layer number are identical as the number of key road segment;Output variable is the following road network
Integrality matrix, the average speed comprising sections all in road network, output layer neuron number and all section numbers of road network
Mesh is identical.
2. a kind of city road net traffic state prediction technique based on key road segment according to claim 1, feature exist
In being specifically included in the step 5:
S501 by the average speed in all sections using abscissa be section number, ordinate as the time in the form of be organized into matrix;
S502 is using neural convolutional network modeling comprising:
(1) input/output variable is determined
Input variable is the spatio-temporal state matrix of the key road segment in road network, the number of input layer number and key road segment
It is identical;Output variable is the integrality matrix of the following road network, the average speed comprising sections all in road network, output layer nerve
First number is identical as all section numbers of road network;
(2) deep layer convolutional neural networks model is constructed, the structural parameters of convolutional neural networks are set
Under the input of training set, the convolutional neural networks of " input layer-convolutional layer-pond layer-output layer " structure are instructed
Practice, and measure the output valve of output layer and the difference of true value with loss function, utilizes BPTT algorithmic minimizing loss function
Obtain optimal model parameter;
(3) following a period of time road network integrality is predicted
Key road segment average speed is extracted from test set, is converted into spatio-temporal state matrix, and input second step is trained
In prediction model, output vector, the traffic flow modes of as next period road network entirety are obtained.
(4) prediction using root-mean-square error and root-mean-square error as the prediction model assessment indicator system of standard, to model is established
Precision is evaluated.Square root error calculation formula isRoot-mean-square error calculation formula isIn formula: u represents fixed number, and l is section number, npRepresentative is fixed number, np=l
×u。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811043484.2A CN109285346B (en) | 2018-09-07 | 2018-09-07 | Urban road network traffic state prediction method based on key road sections |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811043484.2A CN109285346B (en) | 2018-09-07 | 2018-09-07 | Urban road network traffic state prediction method based on key road sections |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109285346A true CN109285346A (en) | 2019-01-29 |
CN109285346B CN109285346B (en) | 2020-05-05 |
Family
ID=65184296
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811043484.2A Expired - Fee Related CN109285346B (en) | 2018-09-07 | 2018-09-07 | Urban road network traffic state prediction method based on key road sections |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109285346B (en) |
Cited By (50)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109754126A (en) * | 2019-01-30 | 2019-05-14 | 银江股份有限公司 | Short-time Traffic Flow Forecasting Methods based on temporal correlation and convolutional neural networks |
CN109887282A (en) * | 2019-03-05 | 2019-06-14 | 中南大学 | A kind of road network traffic flow prediction technique based on level timing diagram convolutional network |
CN109961180A (en) * | 2019-03-15 | 2019-07-02 | 浙江工业大学 | A kind of short-term traffic flow forecast method based on temporal correlation |
CN110322688A (en) * | 2019-05-20 | 2019-10-11 | 华为技术有限公司 | A kind of method of data processing, the method for data query and relevant device |
CN110415516A (en) * | 2019-07-15 | 2019-11-05 | 厦门大学 | Urban traffic flow prediction technique and medium based on figure convolutional neural networks |
CN110675621A (en) * | 2019-05-23 | 2020-01-10 | 北京嘀嘀无限科技发展有限公司 | System and method for predicting traffic information |
CN110738853A (en) * | 2019-10-25 | 2020-01-31 | 北京航空航天大学 | key node identification method based on complex network correlation |
CN110827544A (en) * | 2019-11-11 | 2020-02-21 | 重庆邮电大学 | Short-term traffic flow control method based on graph convolution recurrent neural network |
CN110826788A (en) * | 2019-10-30 | 2020-02-21 | 南京智慧航空研究院有限公司 | Airport scene variable slide-out time prediction method based on big data deep learning |
CN110853349A (en) * | 2019-10-24 | 2020-02-28 | 杭州飞步科技有限公司 | Vehicle scheduling method, device and equipment |
CN111210088A (en) * | 2020-01-16 | 2020-05-29 | 上海理工大学 | Traffic state index prediction method based on space-time factors |
CN111275964A (en) * | 2020-01-14 | 2020-06-12 | 浙江浙大中控信息技术有限公司 | Road section correlation matrix calculation method based on checkpoint data |
CN111291924A (en) * | 2020-01-17 | 2020-06-16 | 同济大学 | Geometric algebraic deep neural network model method for long-term traffic speed prediction |
CN111383452A (en) * | 2019-12-03 | 2020-07-07 | 东南大学 | Method for estimating and predicting short-term traffic running state of urban road network |
CN111540198A (en) * | 2020-04-17 | 2020-08-14 | 浙江工业大学 | Urban traffic situation recognition method based on directed graph convolution neural network |
CN111639787A (en) * | 2020-04-28 | 2020-09-08 | 北京工商大学 | Spatio-temporal data prediction method based on graph convolution network |
CN111680745A (en) * | 2020-06-08 | 2020-09-18 | 青岛大学 | Burst congestion judging method and system based on multi-source traffic big data fusion |
CN111696355A (en) * | 2020-06-29 | 2020-09-22 | 中南大学 | Dynamic graph convolution traffic speed prediction method |
CN111710162A (en) * | 2020-07-07 | 2020-09-25 | 深圳市数字城市工程研究中心 | Urban road network traffic operation condition monitoring method and system |
CN111724601A (en) * | 2020-06-30 | 2020-09-29 | 西安电子科技大学 | Motor vehicle traffic flow prediction method suitable for data loss |
CN111833605A (en) * | 2020-07-10 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Road condition prediction method, road condition prediction model training device and storage medium |
CN111860763A (en) * | 2020-06-05 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Model training method, parameter prediction method, model training device, parameter prediction device, electronic equipment and storage medium |
CN111862595A (en) * | 2020-06-08 | 2020-10-30 | 同济大学 | Speed prediction method, system, medium and device based on road network topological relation |
CN111914052A (en) * | 2020-08-04 | 2020-11-10 | 中国人民解放军战略支援部队信息工程大学 | Method and device for determining regional spatial incidence relation |
CN112201036A (en) * | 2020-09-25 | 2021-01-08 | 同济大学 | Urban expressway travel speed short-time prediction method based on increment-CNN |
CN112241814A (en) * | 2020-10-20 | 2021-01-19 | 河南大学 | Traffic prediction method based on reinforced space-time diagram neural network |
CN112598906A (en) * | 2020-12-14 | 2021-04-02 | 成都易书桥科技有限公司 | Prediction correction algorithm based on space-time dependence |
CN112652167A (en) * | 2020-12-15 | 2021-04-13 | 东北大学 | Traffic speed prediction method based on multi-graph convolution neural network |
CN112863180A (en) * | 2021-01-11 | 2021-05-28 | 腾讯大地通途(北京)科技有限公司 | Traffic speed prediction method, device, electronic equipment and computer readable medium |
CN112927501A (en) * | 2021-01-21 | 2021-06-08 | 南京理工大学 | Urban road network space-time traffic state collaborative prediction method based on dynamic factor model |
CN112927510A (en) * | 2021-02-08 | 2021-06-08 | 湖南大学 | Traffic flow prediction method |
CN112967498A (en) * | 2021-02-02 | 2021-06-15 | 重庆首讯科技股份有限公司 | Short-term traffic flow prediction method based on dynamic space-time correlation characteristic optimization |
CN112991721A (en) * | 2021-02-04 | 2021-06-18 | 南通大学 | Urban road network traffic speed prediction method based on graph convolution network node association degree |
CN113096404A (en) * | 2021-04-23 | 2021-07-09 | 中南大学 | Road blockade oriented quantitative calculation method for change of traffic flow of road network |
CN113160570A (en) * | 2021-05-27 | 2021-07-23 | 长春理工大学 | Traffic jam prediction method and system |
CN113160556A (en) * | 2021-03-12 | 2021-07-23 | 北京邮电大学 | Urban road network dynamic division method and device, computer equipment and storage medium |
CN113313947A (en) * | 2021-05-31 | 2021-08-27 | 湖南大学 | Road condition evaluation method of short-term traffic prediction graph convolution network |
CN113409576A (en) * | 2021-06-24 | 2021-09-17 | 北京航空航天大学 | Bayesian network-based traffic network dynamic prediction method and system |
WO2021212866A1 (en) * | 2020-04-21 | 2021-10-28 | 长安大学 | Vehicle travel volume prediction model construction method, and prediction method and system |
CN113971885A (en) * | 2020-07-06 | 2022-01-25 | 华为技术有限公司 | Vehicle speed prediction method, device and system |
CN114120628A (en) * | 2021-10-12 | 2022-03-01 | 武汉理工大学 | Sparse sensor arrangement-based road state ripple prediction verification method and system |
US20220108603A1 (en) * | 2020-04-10 | 2022-04-07 | Central South University | Detection method of key road-sections based on ricci flow |
CN114333323A (en) * | 2022-01-05 | 2022-04-12 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Highway travel speed prediction method based on pressure characteristics |
CN114926981A (en) * | 2022-05-07 | 2022-08-19 | 同济大学 | Road network traffic jam active management and control method and system |
CN115422694A (en) * | 2022-11-03 | 2022-12-02 | 深圳市城市交通规划设计研究中心股份有限公司 | Road section missing flow calculation method, electronic equipment and storage medium |
CN115526382A (en) * | 2022-09-09 | 2022-12-27 | 扬州大学 | Interpretability analysis method of road network traffic flow prediction model |
CN115759291A (en) * | 2022-11-21 | 2023-03-07 | 武汉大学 | Space nonlinear regression method and system based on ensemble learning |
CN115878861A (en) * | 2023-02-07 | 2023-03-31 | 东南大学 | Selection method of integrated key node group for graph data completion |
CN117437786A (en) * | 2023-12-21 | 2024-01-23 | 速度科技股份有限公司 | Real-time traffic flow prediction method based on artificial intelligence for traffic network |
CN117649769A (en) * | 2024-01-29 | 2024-03-05 | 南京邮电大学 | Future traffic flow intelligent prediction method based on space-time correlation feature matrix |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102034350A (en) * | 2009-09-30 | 2011-04-27 | 北京四通智能交通系统集成有限公司 | Short-time prediction method and system of traffic flow data |
CN102346964A (en) * | 2010-08-05 | 2012-02-08 | 王学鹰 | Real-time jam prediction and intelligent management system for road traffic network area |
CN106981198A (en) * | 2017-05-24 | 2017-07-25 | 北京航空航天大学 | Deep learning network model and its method for building up for predicting travel time |
CN107180530A (en) * | 2017-05-22 | 2017-09-19 | 北京航空航天大学 | A kind of road network trend prediction method based on depth space-time convolution loop network |
CN108399745A (en) * | 2018-03-01 | 2018-08-14 | 北京航空航天大学合肥创新研究院 | A kind of city road network trend prediction method at times based on unmanned plane |
-
2018
- 2018-09-07 CN CN201811043484.2A patent/CN109285346B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102034350A (en) * | 2009-09-30 | 2011-04-27 | 北京四通智能交通系统集成有限公司 | Short-time prediction method and system of traffic flow data |
CN102346964A (en) * | 2010-08-05 | 2012-02-08 | 王学鹰 | Real-time jam prediction and intelligent management system for road traffic network area |
CN107180530A (en) * | 2017-05-22 | 2017-09-19 | 北京航空航天大学 | A kind of road network trend prediction method based on depth space-time convolution loop network |
CN106981198A (en) * | 2017-05-24 | 2017-07-25 | 北京航空航天大学 | Deep learning network model and its method for building up for predicting travel time |
CN108399745A (en) * | 2018-03-01 | 2018-08-14 | 北京航空航天大学合肥创新研究院 | A kind of city road network trend prediction method at times based on unmanned plane |
Cited By (74)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109754126A (en) * | 2019-01-30 | 2019-05-14 | 银江股份有限公司 | Short-time Traffic Flow Forecasting Methods based on temporal correlation and convolutional neural networks |
CN109887282A (en) * | 2019-03-05 | 2019-06-14 | 中南大学 | A kind of road network traffic flow prediction technique based on level timing diagram convolutional network |
CN109887282B (en) * | 2019-03-05 | 2022-01-21 | 中南大学 | Road network traffic flow prediction method based on hierarchical timing diagram convolutional network |
CN109961180A (en) * | 2019-03-15 | 2019-07-02 | 浙江工业大学 | A kind of short-term traffic flow forecast method based on temporal correlation |
CN110322688A (en) * | 2019-05-20 | 2019-10-11 | 华为技术有限公司 | A kind of method of data processing, the method for data query and relevant device |
CN110675621B (en) * | 2019-05-23 | 2021-01-15 | 北京嘀嘀无限科技发展有限公司 | System and method for predicting traffic information |
CN110675621A (en) * | 2019-05-23 | 2020-01-10 | 北京嘀嘀无限科技发展有限公司 | System and method for predicting traffic information |
CN110415516A (en) * | 2019-07-15 | 2019-11-05 | 厦门大学 | Urban traffic flow prediction technique and medium based on figure convolutional neural networks |
CN110853349A (en) * | 2019-10-24 | 2020-02-28 | 杭州飞步科技有限公司 | Vehicle scheduling method, device and equipment |
CN110738853A (en) * | 2019-10-25 | 2020-01-31 | 北京航空航天大学 | key node identification method based on complex network correlation |
CN110826788A (en) * | 2019-10-30 | 2020-02-21 | 南京智慧航空研究院有限公司 | Airport scene variable slide-out time prediction method based on big data deep learning |
CN110827544A (en) * | 2019-11-11 | 2020-02-21 | 重庆邮电大学 | Short-term traffic flow control method based on graph convolution recurrent neural network |
WO2021109318A1 (en) * | 2019-12-03 | 2021-06-10 | 东南大学 | Method for estimating and predicting short-term traffic circulation state of urban road network |
CN111383452A (en) * | 2019-12-03 | 2020-07-07 | 东南大学 | Method for estimating and predicting short-term traffic running state of urban road network |
CN111275964A (en) * | 2020-01-14 | 2020-06-12 | 浙江浙大中控信息技术有限公司 | Road section correlation matrix calculation method based on checkpoint data |
CN111210088B (en) * | 2020-01-16 | 2023-06-02 | 上海理工大学 | Traffic state index prediction method based on space-time factors |
CN111210088A (en) * | 2020-01-16 | 2020-05-29 | 上海理工大学 | Traffic state index prediction method based on space-time factors |
CN111291924B (en) * | 2020-01-17 | 2023-06-06 | 同济大学 | Geometric algebra depth neural network model method for long-term traffic speed prediction |
CN111291924A (en) * | 2020-01-17 | 2020-06-16 | 同济大学 | Geometric algebraic deep neural network model method for long-term traffic speed prediction |
US20220108603A1 (en) * | 2020-04-10 | 2022-04-07 | Central South University | Detection method of key road-sections based on ricci flow |
US12046133B2 (en) * | 2020-04-10 | 2024-07-23 | Central South University | Detection method of key road-sections based on ricci flow |
CN111540198A (en) * | 2020-04-17 | 2020-08-14 | 浙江工业大学 | Urban traffic situation recognition method based on directed graph convolution neural network |
WO2021212866A1 (en) * | 2020-04-21 | 2021-10-28 | 长安大学 | Vehicle travel volume prediction model construction method, and prediction method and system |
CN111639787A (en) * | 2020-04-28 | 2020-09-08 | 北京工商大学 | Spatio-temporal data prediction method based on graph convolution network |
CN111639787B (en) * | 2020-04-28 | 2024-03-15 | 北京工商大学 | Spatio-temporal data prediction method based on graph convolution network |
CN111860763A (en) * | 2020-06-05 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Model training method, parameter prediction method, model training device, parameter prediction device, electronic equipment and storage medium |
CN111862595A (en) * | 2020-06-08 | 2020-10-30 | 同济大学 | Speed prediction method, system, medium and device based on road network topological relation |
CN111680745B (en) * | 2020-06-08 | 2021-03-16 | 青岛大学 | Burst congestion judging method and system based on multi-source traffic big data fusion |
CN111680745A (en) * | 2020-06-08 | 2020-09-18 | 青岛大学 | Burst congestion judging method and system based on multi-source traffic big data fusion |
CN111862595B (en) * | 2020-06-08 | 2021-12-31 | 同济大学 | Speed prediction method, system, medium and device based on road network topological relation |
CN111696355A (en) * | 2020-06-29 | 2020-09-22 | 中南大学 | Dynamic graph convolution traffic speed prediction method |
CN111724601A (en) * | 2020-06-30 | 2020-09-29 | 西安电子科技大学 | Motor vehicle traffic flow prediction method suitable for data loss |
CN111724601B (en) * | 2020-06-30 | 2022-03-18 | 西安电子科技大学 | Motor vehicle traffic flow prediction method suitable for data loss |
CN113971885A (en) * | 2020-07-06 | 2022-01-25 | 华为技术有限公司 | Vehicle speed prediction method, device and system |
CN111710162A (en) * | 2020-07-07 | 2020-09-25 | 深圳市数字城市工程研究中心 | Urban road network traffic operation condition monitoring method and system |
CN111833605A (en) * | 2020-07-10 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Road condition prediction method, road condition prediction model training device and storage medium |
CN111914052A (en) * | 2020-08-04 | 2020-11-10 | 中国人民解放军战略支援部队信息工程大学 | Method and device for determining regional spatial incidence relation |
CN111914052B (en) * | 2020-08-04 | 2022-11-29 | 中国人民解放军战略支援部队信息工程大学 | Method and device for determining regional spatial incidence relation |
CN112201036A (en) * | 2020-09-25 | 2021-01-08 | 同济大学 | Urban expressway travel speed short-time prediction method based on increment-CNN |
CN112201036B (en) * | 2020-09-25 | 2022-06-24 | 同济大学 | Urban expressway travel speed short-time prediction method based on inclusion-CNN |
CN112241814B (en) * | 2020-10-20 | 2022-12-02 | 河南大学 | Traffic prediction method based on reinforced space-time diagram neural network |
CN112241814A (en) * | 2020-10-20 | 2021-01-19 | 河南大学 | Traffic prediction method based on reinforced space-time diagram neural network |
CN112598906B (en) * | 2020-12-14 | 2024-03-19 | 成都易书桥科技有限公司 | Prediction correction algorithm based on space-time dependence |
CN112598906A (en) * | 2020-12-14 | 2021-04-02 | 成都易书桥科技有限公司 | Prediction correction algorithm based on space-time dependence |
CN112652167B (en) * | 2020-12-15 | 2022-03-25 | 东北大学 | Traffic speed prediction method based on multi-graph convolution neural network |
CN112652167A (en) * | 2020-12-15 | 2021-04-13 | 东北大学 | Traffic speed prediction method based on multi-graph convolution neural network |
CN112863180B (en) * | 2021-01-11 | 2022-05-06 | 腾讯大地通途(北京)科技有限公司 | Traffic speed prediction method, device, electronic equipment and computer readable medium |
CN112863180A (en) * | 2021-01-11 | 2021-05-28 | 腾讯大地通途(北京)科技有限公司 | Traffic speed prediction method, device, electronic equipment and computer readable medium |
CN112927501A (en) * | 2021-01-21 | 2021-06-08 | 南京理工大学 | Urban road network space-time traffic state collaborative prediction method based on dynamic factor model |
CN112927501B (en) * | 2021-01-21 | 2022-05-27 | 南京理工大学 | Urban road network space-time traffic state collaborative prediction method based on dynamic factor model |
CN112967498A (en) * | 2021-02-02 | 2021-06-15 | 重庆首讯科技股份有限公司 | Short-term traffic flow prediction method based on dynamic space-time correlation characteristic optimization |
CN112991721A (en) * | 2021-02-04 | 2021-06-18 | 南通大学 | Urban road network traffic speed prediction method based on graph convolution network node association degree |
CN112927510B (en) * | 2021-02-08 | 2021-12-17 | 湖南大学 | Traffic flow prediction method |
CN112927510A (en) * | 2021-02-08 | 2021-06-08 | 湖南大学 | Traffic flow prediction method |
CN113160556A (en) * | 2021-03-12 | 2021-07-23 | 北京邮电大学 | Urban road network dynamic division method and device, computer equipment and storage medium |
CN113096404A (en) * | 2021-04-23 | 2021-07-09 | 中南大学 | Road blockade oriented quantitative calculation method for change of traffic flow of road network |
CN113096404B (en) * | 2021-04-23 | 2022-02-15 | 中南大学 | Road blockade oriented quantitative calculation method for change of traffic flow of road network |
CN113160570A (en) * | 2021-05-27 | 2021-07-23 | 长春理工大学 | Traffic jam prediction method and system |
CN113313947A (en) * | 2021-05-31 | 2021-08-27 | 湖南大学 | Road condition evaluation method of short-term traffic prediction graph convolution network |
CN113313947B (en) * | 2021-05-31 | 2022-04-19 | 湖南大学 | Road condition evaluation method of short-term traffic prediction graph convolution network |
CN113409576A (en) * | 2021-06-24 | 2021-09-17 | 北京航空航天大学 | Bayesian network-based traffic network dynamic prediction method and system |
CN114120628A (en) * | 2021-10-12 | 2022-03-01 | 武汉理工大学 | Sparse sensor arrangement-based road state ripple prediction verification method and system |
CN114333323A (en) * | 2022-01-05 | 2022-04-12 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Highway travel speed prediction method based on pressure characteristics |
CN114926981B (en) * | 2022-05-07 | 2023-04-11 | 同济大学 | Road network traffic jam active management and control method and system |
CN114926981A (en) * | 2022-05-07 | 2022-08-19 | 同济大学 | Road network traffic jam active management and control method and system |
CN115526382A (en) * | 2022-09-09 | 2022-12-27 | 扬州大学 | Interpretability analysis method of road network traffic flow prediction model |
CN115526382B (en) * | 2022-09-09 | 2023-05-23 | 扬州大学 | Road network level traffic flow prediction model interpretability analysis method |
CN115422694A (en) * | 2022-11-03 | 2022-12-02 | 深圳市城市交通规划设计研究中心股份有限公司 | Road section missing flow calculation method, electronic equipment and storage medium |
CN115759291A (en) * | 2022-11-21 | 2023-03-07 | 武汉大学 | Space nonlinear regression method and system based on ensemble learning |
CN115878861A (en) * | 2023-02-07 | 2023-03-31 | 东南大学 | Selection method of integrated key node group for graph data completion |
CN117437786B (en) * | 2023-12-21 | 2024-02-27 | 速度科技股份有限公司 | Real-time traffic flow prediction method based on artificial intelligence for traffic network |
CN117437786A (en) * | 2023-12-21 | 2024-01-23 | 速度科技股份有限公司 | Real-time traffic flow prediction method based on artificial intelligence for traffic network |
CN117649769A (en) * | 2024-01-29 | 2024-03-05 | 南京邮电大学 | Future traffic flow intelligent prediction method based on space-time correlation feature matrix |
CN117649769B (en) * | 2024-01-29 | 2024-04-16 | 南京邮电大学 | Future traffic flow intelligent prediction method based on space-time correlation feature matrix |
Also Published As
Publication number | Publication date |
---|---|
CN109285346B (en) | 2020-05-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109285346A (en) | A kind of city road net traffic state prediction technique based on key road segment | |
CN109410575B (en) | Road network state prediction method based on capsule network and nested long-time memory neural network | |
Reddy et al. | Solar resource estimation using artificial neural networks and comparison with other correlation models | |
CN113538910B (en) | Self-adaptive full-chain urban area network signal control optimization method | |
CN109887282A (en) | A kind of road network traffic flow prediction technique based on level timing diagram convolutional network | |
CN108399745A (en) | A kind of city road network trend prediction method at times based on unmanned plane | |
CN112767682A (en) | Multi-scale traffic flow prediction method based on graph convolution neural network | |
CN111242292B (en) | OD data prediction method and system based on deep space-time network | |
CN109359166B (en) | Space growth dynamic simulation and driving force factor contribution degree synchronous calculation method | |
CN111242395B (en) | Method and device for constructing prediction model for OD (origin-destination) data | |
CN114925836B (en) | Urban traffic flow reasoning method based on dynamic multi-view graph neural network | |
CN114495500B (en) | Traffic prediction method based on dual dynamic space-time diagram convolution | |
CN108280998A (en) | Short-time Traffic Flow Forecasting Methods based on historical data dynamic select | |
CN116052427B (en) | Inter-city inter-regional mobility prediction method and device based on private car travel track data | |
CN115204478A (en) | Public traffic flow prediction method combining urban interest points and space-time causal relationship | |
CN112182063A (en) | Method for constructing hydrological forecasting model based on space-time characteristics | |
CN114021836A (en) | Multivariable reservoir water inflow amount prediction system based on different-angle fusion, training method and application | |
CN116781139A (en) | Flow prediction satellite path selection method and system based on reinforcement learning | |
Liu et al. | A method for short-term traffic flow forecasting based on GCN-LSTM | |
CN108615360A (en) | Transport need based on neural network Evolution Forecast method day by day | |
CN112561132A (en) | Water flow prediction model based on neural network | |
CN117034780B (en) | Multi-scale sub-season precipitation prediction method based on deep learning | |
CN111163430B (en) | Water quantity prediction method based on mobile phone base station user positioning data | |
CN112561133A (en) | Water flow prediction system and method based on neural network | |
Lyu et al. | Water level prediction model based on GCN and LSTM |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200505 Termination date: 20210907 |
|
CF01 | Termination of patent right due to non-payment of annual fee |