CN112884014A - Traffic speed short-time prediction method based on road section topological structure classification - Google Patents
Traffic speed short-time prediction method based on road section topological structure classification Download PDFInfo
- Publication number
- CN112884014A CN112884014A CN202110101858.7A CN202110101858A CN112884014A CN 112884014 A CN112884014 A CN 112884014A CN 202110101858 A CN202110101858 A CN 202110101858A CN 112884014 A CN112884014 A CN 112884014A
- Authority
- CN
- China
- Prior art keywords
- road
- road section
- model
- sections
- speed
- 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.)
- Withdrawn
Links
Images
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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/08—Learning methods
-
- 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
- G08G1/0133—Traffic data processing for classifying traffic situation
Abstract
The invention discloses a short-term traffic speed prediction method for road section classification based on topological similarity, and belongs to the field of intelligent traffic. The specific operation steps are as follows: the method comprises the steps of 1, extracting information of upstream and downstream road sections of a road section, carrying out topological similarity analysis on the road sections, generating road section class identification codes, and classifying the road sections according to the identification codes; matching vehicle GPS track data into an urban road network, preprocessing track data corresponding to each type of road section set, and constructing a road traffic speed space-time matrix; 3. and constructing a long short-term memory network (LSTM) model for each type of road section set, and inputting trajectory data to carry out model training and testing. The invention provides a scheme for carrying out short-term traffic speed prediction by combining an LSTM model on the basis of analyzing road section topological similarity to carry out road section classification, and the characteristics of traffic flow modes of similar road sections are reserved, so that the model prediction precision is improved.
Description
Technical Field
The invention relates to the field of intelligent traffic, in particular to a traffic speed short-term prediction method based on road section topological structure classification, belonging to the field of traffic speed short-term prediction based on a neural network.
Background
In an urban road network traffic system, traffic prediction has an important reference value for relieving urban road traffic congestion, wherein speed is the most intuitive index for reflecting road conditions. Accurate speed prediction can help travelers to sense road conditions and plan routes in time, and therefore time cost is saved.
At present, the prediction of traffic flow is mainly divided into two categories, namely a traditional method and an intelligent algorithm. The traditional methods mainly comprise parameter regression model prediction, time series prediction and the like, but the traditional methods have low prediction precision in actual traffic prediction or cannot deal with emergencies so as not to effectively improve traffic states. The intelligent algorithm mainly comprises a neural network, a nonlinear prediction method and the like. The prediction model research of the neural network combines the advantages of machine learning, and the prediction precision is remarkably improved compared with that of the conventional method. However, these prediction methods are based on traffic flow data collected by sensors on roads for prediction, but the track data generated by moving objects, which is more easily obtained in a road network, is not fully utilized due to the high price and maintenance cost of the sensors, and the existing traffic speed prediction methods are mostly directed at expressways or intersections and rarely directed at traffic speed prediction on large-scale road networks, and in addition, due to the space complexity of large-scale road networks, a single prediction model cannot be applied to road segments with different traffic modes, so that the prediction methods cannot be well applied to large-scale road networks in cities.
The traffic flow data is time sequence data which contains strong space-time correlation, and the LSTM can establish the relation between adjacent time information through the connection between nodes, so that a short-time traffic speed prediction model of the LSTM is adopted to 'memorize' the previous information, thereby being well adapted to the non-linear and random characteristics of the traffic flow.
The invention provides a road network traffic speed short-time prediction method for road section classification based on topological similarity by combining the work of the two aspects, the method analyzes the topological similarity of road sections, researches the traffic flow data of similar road sections, and adopts an LSTM model to improve the prediction precision on the basis of keeping the traffic mode characteristics of the similar road sections.
Disclosure of Invention
Aiming at the problems, the invention provides a traffic speed prediction method, which combines LSTM to perform model training on the basis of classification based on road section topological similarity, thereby effectively improving the speed prediction precision.
The technical scheme of the invention is as follows: a short-term traffic speed prediction method based on road section topological structure classification comprises the following specific steps:
step (1.1), carrying out topological similarity analysis on the preprocessed road sections according to the lengths of the road sections and the upstream and downstream road section sets so as to generate road section class identification codes, and classifying the road sections according to the identification codes;
step (1.2), matching vehicle GPS track data into an urban road network and preprocessing the track data corresponding to each type of road section set so as to construct a road traffic speed space-time matrix;
and (1.3) constructing an LSTM model for each type of road section set, and inputting corresponding data into the model for model training and prediction.
Further, in step (1.1), the specific operation steps of performing topology similarity analysis on the preprocessed road segments according to the lengths of the road segments and the upstream and downstream road segment sets are as follows:
(1.1.1) preprocessing a road network to obtain road section length and upstream and downstream road section information, and defining that similar road sections are road sections with similar lengths and equal upstream and downstream road section set numbers respectively;
(1.1.2) generating a road section category identification code for each road section according to similar road section definitions, wherein the road sections with the same identification codes are mapped into the same hash bucket and are represented as a type of road sections with similar traffic modes.
Further, in the step (1.2), the specific operation steps of matching the vehicle GPS track data to the urban road network and preprocessing the track data corresponding to each type of road segment set are as follows:
(1.2.1) matching a road network to each road section by vehicle GPS track data according to the longitude and latitude, dividing the track data into space-time cells, and preprocessing to obtain a speed time sequence;
(1.2.2) constructing a space-time speed matrix data set, wherein the space-time speed matrix data set is expressed as a matrix formed by speed time sequences corresponding to the upstream and downstream road section sets of the target road section.
Further, in the step (1.3),
for each classified road section set, constructing an initial LSTM model, inputting the track data corresponding to each road section set into the model for training and predicting;
in the evaluation link of the model, a root mean square error and a defined model precision evaluation index ACC are used as model evaluation standards, and the specific expressions are respectively expressed as follows:
wherein RMSE represents the root mean square error; m represents the number of test samples;
indicates the predicted value of speed at time i, yiRepresenting the true value; ACC represents the prediction accuracy of the model.
The invention has the beneficial effects that: the traffic speed short-term prediction method based on road section topological structure classification carries out topological similarity analysis on road sections and classifies the road sections to keep the characteristic of traffic flow modes of similar road sections in a preprocessing stage, and adopts an LSTM model to train in a model training stage of similar road sections, thereby effectively improving the prediction precision.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic structural diagram of a road network structure according to the present invention;
FIG. 3 is a schematic illustration of a similar road segment in the present invention;
FIG. 4 is a schematic configuration diagram of a link classification in the present invention;
FIG. 5 is a schematic block diagram of a spatiotemporal cell in accordance with the present invention;
fig. 6 is a schematic diagram of an LSTM node in the present invention.
Detailed Description
In order to more clearly illustrate the technical solution of the present invention, the following detailed description is made with reference to the accompanying drawings:
a traffic speed short-term prediction method based on road section topological structure classification is disclosed, wherein a flow chart of the whole method is shown in figure 1, and the method combines LSTM training to obtain a prediction model for keeping the characteristics of traffic flow modes of similar road sections on the basis of road section classification based on topological similarity, and finally obtains the future speed of a target road section; firstly, classifying road sections according to the lengths of the road sections and information of upstream and downstream road sections, classifying the road sections with similar traffic modes into one class, preprocessing the track data of the same type of road sections, constructing an LSTM model, defining model input, dividing a data set, training to obtain a prediction model, and finally substituting real-time speed data to obtain the speed prediction of a target road section; the method comprises the following specific steps:
step (1.1), carrying out topological similarity analysis on the preprocessed road sections according to the lengths of the road sections and the upstream and downstream road section sets so as to generate road section class identification codes, and classifying the road sections according to the identification codes;
in the step (1.1), the upstream and downstream information of the road sections in the urban road network is extracted, the topological similarity analysis is carried out on the road sections to generate road section class identification codes, the road sections are classified according to the identification codes, and the operation steps are as follows:
(1.1.1) preprocessing a road network to obtain road section length and upstream and downstream road section information, analyzing road section topological similarity, and defining that similar road sections are road sections with similar lengths and equal upstream and downstream road section set numbers respectively, namely that two road sections with similar lengths and equal upstream and downstream road section set numbers are similar road sections respectively;
specifically, 1, the urban road network is represented as a directed road network and consists of an intersection set J and a directed road section set E, each directed road section es E between two intersections is represented as a 3-tuple (len, up, down), wherein len is length, up is an upstream road section set of the road section, and down is a downstream road section set of the road section; acquiring id and length len of each road section es according to the definition, and processing to obtain an upstream road section set up and a downstream road section set down of each road section; as shown in the schematic diagram of the road network structure in fig. 2, an upstream link set of a link d is represented as up ═ e, f, g, and a downstream link set is represented as down ═ a, b, c;
since the road segments with similar lengths and equal numbers of the upstream and downstream road segments have similar traffic flow patterns in the actual road network, the topological similarity of the road segments is defined as follows: assume that there are two road segments es1 and es2 if they satisfy nupes1＝nupes2And ndownes1＝ndownes2And | lenes1-lenes2|<1 i.e. road segments es1 and es2 are considered similar road segments; where len is length in kilometers (km), nup refers to the number of upstream road segments and ndown refers to the number of downstream road segments; as shown in fig. 3, for example, the section es1 is 1.3km in length, the number of upstream sets of road segments is 3, the number of downstream sets of road segments is 2, the section es2 is 1.5km in length, the number of upstream sets of road segments is 3, and the number of downstream sets of road segments is 2, and since they satisfy topological similarity, the section es1 and the section es2 are similar sections;
(1.1.2) generating a road section category identification code for each road section according to similar road section definitions, wherein the road sections with the same identification code are mapped into the same hash bucket and are represented as a type of road sections with similar traffic modes; the road section length and the number of the upstream and downstream road section sets can determine a type of road sections with similar traffic flow characteristics, so that a category identification code of each road section is generated and is mapped into different road section sets according to the identification code;
specifically, 2, generating a category identification code for each road section according to the topological similarity, specifically representing a character string skey formed by sequentially splicing the upper road section number nup, the lower road section number ndown and the lower integral value of the road section length; the method comprises the steps of dividing road sections with the same category identification codes into the same category, mapping the road sections to the same road section set by adopting a Hash method, and finally obtaining a plurality of road section sets; fig. 4 is a schematic diagram of road segment classification.
Step (1.2), matching vehicle GPS track data into an urban road network and preprocessing the track data corresponding to each type of road section set so as to construct a road traffic speed space-time matrix;
in the step (1.2), the vehicle GPS track data is matched into an urban road network, the track data corresponding to each type of road section set is preprocessed, and a road traffic speed space-time matrix is constructed, wherein the specific operation steps are as follows:
(1.2.1) road network matching is carried out on vehicle GPS track data to each road section according to longitude and latitude, then further track data is divided into space-time cells from time, track data in the cells are preprocessed to obtain track point speeds, and further speed time sequences are obtained;
specifically, 1, map matching is performed according to the longitude and latitude of each GPS track point to obtain a nearest road section, and finally, track data corresponding to each road section is obtained, wherein the track data includes track data of different dates and different times, so that the map data is further divided in time, the track data is divided into a plurality of equal time periods for each day, and is finally divided into space-time cells TSG shown in fig. 5, each space-time cell includes a plurality of track data of a certain time period on a certain road, and each track point data includes information such as a moving object id, the longitude and latitude of the moving object, and track collection time;
calculating the displacement of two track points according to the longitude and latitude information of each track point and the next track point, and calculating the average speed of the track point according to the time difference collected by the two track points, wherein each track point corresponds to a moving object; processing the speed of each track point in the time-space cell, taking the average speed of all moving objects in the time period as the speed of the road section in the time period, and finally generating a speed time sequence corresponding to each road section;
(1.2.2) constructing a space-time speed matrix data set for each type of road section set, and specifically inputting a matrix formed by speed time sequences corresponding to the upstream and downstream road section sets of the target road section;
specifically, 2, for each road segment set, the training set and the test set are divided according to the number of the road segments, the division ratio is 7:3, and the input of the LSTM prediction model is defined as follows:
where V represents a traffic velocity space-time matrix, Vi(t) is expressed as the speed of the corresponding road segment at time t for road segment i, where m represents the total number of upstream and downstream road segments for the considered target road segment and n represents the time window size.
Step (1.3), constructing an LSTM model for each type of road section set, and inputting corresponding data into the model to carry out model training and prediction;
in the step (1.3), the step (c),
for each classified road section set, constructing an initial LSTM model, inputting the track data corresponding to each road section set into the model for training and predicting;
in the evaluation link of the model, a root mean square error and a defined model precision evaluation index ACC are used as model evaluation standards, and the specific expressions are respectively expressed as follows:
wherein RMSE represents the root mean square error; m represents the number of test samples;
indicates the predicted value of speed at time i, yiRepresenting the true value; ACC represents the prediction accuracy of the model.
Specifically, an LSTM model is constructed for each type of road segment set, a specific structure of the LSTM model is as shown in fig. 6, and a calculation formula at each time t is as follows:
it＝σ(Wixxt+Wihht-1+bi)
ft＝σ(Wfxxt+Wfhht-1+bf)
ot＝σ(Woxxt+Wohht+bo)
st＝gt⊙it+ft⊙st-1
in the above formula, it，ft，otRespectively representing an input gate, a forgetting gate and an output gate, gtRepresenting a cell input unit containing input samples x at the current timetAnd the output h of the hidden layer node at the previous momentt-1；stA memory cell representing a unit cell,
and σ denote activation functions tanh and sigmoid, respectively, W denotes a weight matrix, b denotes a corresponding bias vector, and |, denotes a dot-by-dot product.
And carrying out model training on the corresponding data training set, inputting the test set into the prediction model to realize speed prediction of the target road section, and defining evaluation indexes to carry out accuracy analysis on the prediction result.
The specific embodiment is as follows: the research data selected in the example is track data of taxies which are counted for 7 days from No. 2/2008 to No. 2/8 in Beijing City, and the method is explained, wherein 10357 taxies are contained in total, and 157955 road sections are contained in total; taking 5 minutes as a time period, dividing one day into 288 time periods;
step 1: carrying out topological similarity analysis on the preprocessed road sections according to the lengths of the road sections and the upstream and downstream road section sets to generate road section category identification codes, and classifying the road sections according to the identification codes; the specific implementation steps are as follows:
s1.1: the method comprises the steps that an urban road network in an analysis embodiment obtains a road section es set, the road section es set is processed to obtain the length len of each road section, an upstream road section set up of the road section and a downstream road section set down of the road section, and information of the length of a part of the road section and the number of the upstream road section set and the downstream road section set is shown in a table 1;
table 1: road section corresponding information table
S1.2: for each road section, character strings obtained by sequentially splicing lower integral values of the number nup of the upstream road sections, the number ndown of the downstream road sections and the length len of the downstream road sections are used as the unique identification code skey of the road section category; mapping similar road sections into the same set by adopting a Hash method according to the identification code, wherein one set represents a class with the same traffic mode; for example, for a road section with the length of 1.3km, the number of upstream road sections of 3 and the number of downstream road sections of 4, the character string obtained by splicing the character strings according to the definition is represented as 341, and a category identification code is generated for each road section according to the method; road network of the example is classified based on road sections with topological similarity, and identification codes generated by part of road sections are shown in table 1; the category identification codes skey of the partial road sections and the number of the corresponding road section sets which are finally obtained after classification according to the category identification codes are shown in the table 2;
table 2: road section class identification code and corresponding number table thereof
Step S2: matching vehicle GPS track data into an urban road network, preprocessing the track data corresponding to each type of road section set, and constructing a road traffic speed space-time matrix, wherein the specific implementation steps are as follows:
s2.1: matching all GPS points to corresponding road sections through a map matching algorithm, wherein detailed information of a partial track data set and the matched road sections in the experiment are shown in a table 3;
table 3: experimental data sheet
According to the classified road section set obtained in the step S1, sequentially and respectively selecting 1000 road sections with dense track points from the road section sets with road section category identification codes of 321, 330 and 442, dividing the corresponding track data into space-time cells, calculating the speed of each track point in the space-time cells, wherein each track point corresponds to a unique moving object, averaging the speeds of the obtained multiple moving objects to obtain the speeds of the road sections in different time periods, and further obtaining the traffic speed time sequence of each road section;
s2.2: defining the input of the LSTM model as a matrix formed by speed time sequences of an upstream road section set and a downstream road section set of a target road section; and for each road section set, dividing the road section set into a training set and a test set according to the number of the road sections, wherein the dividing ratio is 7: 3.
Step S3: constructing an LSTM model for each type of road section set, inputting corresponding data into the model for model training and prediction, and specifically implementing the following steps:
s3.1: for the selected 3 types of road section sets, taking the speed data of the first 12 time periods of the predicted time as training data to carry out LSTM network training; setting the number of LSTM network hidden layer nodes to be 32, the time prediction sequence value to be 1 and the initial learning rate lr to be 0.01;
s3.2: the test set of the models corresponding to the three classes of road segments trained in the previous step is checked, and the obtained prediction model effects are shown in table 4:
table 4: model prediction result table corresponding to different road section sets
The invention provides a short-time prediction scheme of road network traffic speed for classifying road sections based on topological similarity aiming at the problem that the existing speed prediction model constructed for a single road section cannot be well adapted to the speed prediction of other road sections with topological structures, makes full use of the accessibility of intelligent device GPS track data in a road network, well reserves the characteristic of a traffic flow mode of a similar road section for road section classification based on the topological similarity, can realize one-time multi-path section prediction of the road network level traffic speed, and has good spatial applicability.
Finally, it should be understood that the embodiments described herein are merely illustrative of the principles of embodiments of the present invention; other variations are possible within the scope of the invention; thus, by way of example, and not limitation, alternative configurations of embodiments of the invention may be considered consistent with the teachings of the present invention; accordingly, the embodiments of the invention are not limited to the embodiments explicitly described and depicted.
Claims (4)
1. A traffic speed short-term prediction method based on road section topological structure classification is characterized by comprising the following specific steps:
step (1.1), carrying out topological similarity analysis on the preprocessed road sections according to the lengths of the road sections and the upstream and downstream road section sets so as to generate road section class identification codes, and classifying the road sections according to the identification codes;
step (1.2), matching vehicle GPS track data into an urban road network and preprocessing the track data corresponding to each type of road section set so as to construct a road traffic speed space-time matrix;
and (1.3) constructing an LSTM model for each type of road section set, and inputting corresponding data into the model for model training and prediction.
2. The short-term traffic speed prediction method based on road section topological structure classification as claimed in claim 1, wherein in step (1.1), the specific operation steps of performing topological similarity analysis on the preprocessed road sections according to the length of the preprocessed road sections and the upstream and downstream road section sets are as follows:
(1.1.1) preprocessing a road network to obtain road section length and upstream and downstream road section information, and defining that similar road sections are road sections with similar lengths and equal upstream and downstream road section set numbers respectively;
(1.1.2) generating a road section category identification code for each road section according to similar road section definitions, wherein the road sections with the same identification codes are mapped into the same hash bucket and are represented as a type of road sections with similar traffic modes.
3. The traffic speed short-term prediction method based on road section topological structure classification as claimed in claim 1, wherein in step (1.2), the specific operation steps of matching vehicle GPS track data into city road network and preprocessing the track data corresponding to each road section set are as follows:
(1.2.1) matching a road network to each road section by vehicle GPS track data according to the longitude and latitude, dividing the track data into space-time cells, and preprocessing to obtain a speed time sequence;
(1.2.2) constructing a space-time speed matrix data set, wherein the space-time speed matrix data set is expressed as a matrix formed by speed time sequences corresponding to the upstream and downstream road section sets of the target road section.
4. A traffic speed short-term prediction method based on road section topological structure classification according to claim 1, characterized by, in step (1.3),
for each classified road section set, constructing an initial LSTM model, inputting the track data corresponding to each road section set into the model for training and predicting;
in the evaluation link of the model, a root mean square error and a defined model precision evaluation index ACC are used as model evaluation standards, and the specific expressions are respectively expressed as follows:
wherein RMSE represents the root mean square error; m represents the number of test samples;
indicates the predicted value of speed at time i, yiRepresenting the true value; ACC represents the prediction accuracy of the model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110101858.7A CN112884014A (en) | 2021-01-26 | 2021-01-26 | Traffic speed short-time prediction method based on road section topological structure classification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110101858.7A CN112884014A (en) | 2021-01-26 | 2021-01-26 | Traffic speed short-time prediction method based on road section topological structure classification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112884014A true CN112884014A (en) | 2021-06-01 |
Family
ID=76051866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110101858.7A Withdrawn CN112884014A (en) | 2021-01-26 | 2021-01-26 | Traffic speed short-time prediction method based on road section topological structure classification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112884014A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114510997A (en) * | 2021-12-30 | 2022-05-17 | 北京掌行通信息技术有限公司 | Toll station passing mode identification method and system |
| CN115424439A (en) * | 2022-08-25 | 2022-12-02 | 广西北投信创科技投资集团有限公司 | Traffic flow prediction method and device based on feature processing |
| CN115565376A (en) * | 2022-09-30 | 2023-01-03 | 福州大学 | Vehicle travel time prediction method and system fusing graph2vec and double-layer LSTM |
| WO2023123456A1 (en) * | 2021-12-31 | 2023-07-06 | 深圳市大疆创新科技有限公司 | Vehicle location prediction method and apparatus, and vehicle and storage medium |
| CN115565376B (en) * | 2022-09-30 | 2024-05-03 | 福州大学 | Vehicle journey time prediction method and system integrating graph2vec and double-layer LSTM |
-
2021
- 2021-01-26 CN CN202110101858.7A patent/CN112884014A/en not_active Withdrawn
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114510997A (en) * | 2021-12-30 | 2022-05-17 | 北京掌行通信息技术有限公司 | Toll station passing mode identification method and system |
| WO2023123456A1 (en) * | 2021-12-31 | 2023-07-06 | 深圳市大疆创新科技有限公司 | Vehicle location prediction method and apparatus, and vehicle and storage medium |
| CN115424439A (en) * | 2022-08-25 | 2022-12-02 | 广西北投信创科技投资集团有限公司 | Traffic flow prediction method and device based on feature processing |
| CN115565376A (en) * | 2022-09-30 | 2023-01-03 | 福州大学 | Vehicle travel time prediction method and system fusing graph2vec and double-layer LSTM |
| CN115565376B (en) * | 2022-09-30 | 2024-05-03 | 福州大学 | Vehicle journey time prediction method and system integrating graph2vec and double-layer LSTM |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108629978B (en) | Traffic track prediction method based on high-dimensional road network and recurrent neural network | |
| CN109670277B (en) | Travel time prediction method based on multi-mode data fusion and multi-model integration | |
| CN106845371B (en) | A kind of city road network automotive emission remote sensing monitoring system | |
| CN111080029B (en) | Urban traffic road speed prediction method and system based on multi-path segment space-time correlation | |
| CN110570651A (en) | Road network traffic situation prediction method and system based on deep learning | |
| CN112884014A (en) | Traffic speed short-time prediction method based on road section topological structure classification | |
| US20160125307A1 (en) | Air quality inference using multiple data sources | |
| Wang et al. | Efficient metropolitan traffic prediction based on graph recurrent neural network | |
| CN110738855B (en) | Road traffic flow condition prediction method in data sparse time period | |
| CN112613225B (en) | Intersection traffic state prediction method based on neural network cell transmission model | |
| WO2021013190A1 (en) | Meteorological parameter-based high-speed train positioning method and system in navigation blind zone | |
| CN111292534A (en) | Traffic state estimation method based on clustering and deep sequence learning | |
| CN113436433B (en) | Efficient urban traffic outlier detection method | |
| Necula | Dynamic traffic flow prediction based on GPS data | |
| CN110889444B (en) | Driving track feature classification method based on convolutional neural network | |
| CN114881356A (en) | Urban traffic carbon emission prediction method based on particle swarm optimization BP neural network optimization | |
| CN111949896A (en) | Track anomaly detection method based on track big data | |
| CN112633602B (en) | Traffic congestion index prediction method and device based on GIS map information | |
| CN110459056A (en) | A kind of public transport arrival time prediction technique based on LSTM neural network | |
| Xiong et al. | A Descriptive Bayesian Approach to Modeling and Calibrating Drivers' En Route Diversion Behavior | |
| CN112863182A (en) | Cross-modal data prediction method based on transfer learning | |
| CN114882069A (en) | Taxi track abnormity detection method based on LSTM network and attention mechanism | |
| CN114202120A (en) | Urban traffic travel time prediction method aiming at multi-source heterogeneous data | |
| CN112101132B (en) | Traffic condition prediction method based on graph embedding model and metric learning | |
| CN112149922A (en) | Method for predicting severity of accident in exit and entrance area of down-link of highway tunnel |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210601 |