CN112100163A - Road network state space-time prediction method based on three-dimensional convolutional neural network - Google Patents
Road network state space-time prediction method based on three-dimensional convolutional neural network Download PDFInfo
- Publication number
- CN112100163A CN112100163A CN202010836658.1A CN202010836658A CN112100163A CN 112100163 A CN112100163 A CN 112100163A CN 202010836658 A CN202010836658 A CN 202010836658A CN 112100163 A CN112100163 A CN 112100163A
- Authority
- CN
- China
- Prior art keywords
- data
- road network
- road
- time
- convolutional neural
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- 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
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Fuzzy Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Quality & Reliability (AREA)
- Traffic Control Systems (AREA)
Abstract
A road network state space-time prediction method based on a three-dimensional convolutional neural network is characterized in that a computer is used as a prediction tool, and the prediction is divided into six steps, specifically, the steps are as follows, step A is data processing; and B: b, calculating the road section state according to the data obtained in the step A; and C: b, converting the data into a road network state space-time matrix according to the data obtained in the step B; step D: meshing road network data; e, establishing a three-dimensional convolutional neural network state prediction model according to the gridding data obtained in the step D; step F: and E, optimizing the relevant content of the model obtained in the step E. According to the invention, the calculation result obtained through six steps is wider in the target coverage range, the road network updating time scale can be reduced to two minutes, and the method has stronger practical significance for the management and control of the intelligent traffic system, thereby playing a powerful technical support for the realization of the construction of the Intelligent Traffic System (ITS). In conclusion, the invention has good application prospect.
Description
Technical Field
The invention relates to the technical field of traffic network prediction methods, in particular to a road network state space-time prediction method based on a three-dimensional convolutional neural network.
Background
Along with the continuous and stable development of economy in China, the physical life and mental requirements of people are gradually improved, the travel requirements on rapidness and comfort are higher and higher, and corresponding traffic jam is more prominent due to the continuous rising of the travel volume of cars and the like. Traffic congestion has become a social problem facing developed countries, particularly some developing countries. The method solves the traffic jam and can not be limited to building roads, installing traffic auxiliary equipment and the like. In practical situations, to fundamentally solve the problem of traffic congestion, the behaviors of drivers and passengers should be standardized, more scientific and reasonable path identification is provided, and the overall utilization efficiency of a road network is improved, so that the effect of achieving twice the result with half the effort is achieved. To achieve the above object, then, there is a direct concern that traffic data prediction is a problem. In traffic data prediction, how to predict the degree of road smoothness is an essential task to predict traffic volume in a short period by using the past traffic volume data. At present, an Intelligent Transportation System (ITS) which is developed in a large scale by a transportation department needs accurate and real-time traffic data prediction as a basis, so that the traffic state prediction is an important link in traffic control.
In the conventional prediction method, although a technology for predicting part of roads in short term has been developed to a certain extent, a new intelligent traffic technology (ITS) requires that traffic prediction can predict part of roads, and also needs to predict traffic data of a road network in a large area or even extend the traffic data to the whole urban network.
In summary, the conventional prediction methods have two main disadvantages. On one hand, many scholars do abundant research on prediction of single latitude such as time series and complete a lot of work of road section level traffic prediction, but the work such as dynamic path guidance and prediction of a dead point needs more extensive traffic state prediction, so that the existing road section level-based traffic state prediction lacks of mining of spatial relations among road sections. In fact, traffic states are not only time-related but also spatially related, which often changes dynamically with time. On the other hand, in the existing research, the traditional parameterized prediction method adopts a statistical method, an analysis model (such as a travel time function, a queuing model) or a traffic simulation model, and although the models make reasonable theoretical or physical assumptions on the time evolution law of traffic, the models have a few problems. There are mainly the following problems: statistical methods such as time series and the like have poor prediction effects on nonlinear and large-fluctuation data, various models need synchronous data to support internal model variables, and the data are difficult to obtain, so that the application has great defects and the construction requirements of an Intelligent Transportation System (ITS) can not be effectively met.
Disclosure of Invention
In order to overcome the defects of the prior traffic prediction method of mining on time-space regular data and failing to effectively meet the construction requirements of an Intelligent Traffic System (ITS), the invention provides a method which takes a complex road network comprising ordinary roads, elevated roads, auxiliary roads and intersections as a research and calculation object, has wider target object coverage range, replaces the input of multi-channel images by a single-channel matrix, effectively utilizes the automatic feature extraction of moving traffic state data, synchronously calculates time-space features, is beneficial to mining of time dimension rules, effectively utilizes historical state information to extract time-space features, and finally obtains a prediction model to reduce the updating time scale of the road network to two minutes, therefore, the road network state space-time prediction method based on the three-dimensional convolutional neural network has powerful technical support for the realization of Intelligent Transportation System (ITS) construction.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a road network state space-time prediction method based on a three-dimensional convolutional neural network is used as a prediction tool through computer application, and is characterized in that the prediction is divided into six steps, specifically, the steps are data processing, namely, identifying the missing and distortion conditions of floating car data in the running of a road network to complete the cleaning and completion of the data; and B: b, calculating the road section state according to the data obtained in the step A, specifically calculating the traffic flow, the density, the speed parameters, the number of vehicles, the vehicle speed and the saturation; and C: b, converting the data into a road network state space-time matrix according to the data obtained in the step B; step D: meshing road network data; step E: d, establishing a three-dimensional convolutional neural network state prediction model according to the gridding data obtained in the step D, and completing the extraction of the dynamic road network state features after the datamation representation; step F: and E, optimizing the relevant content of the model obtained in the step E, and mainly adjusting the activation function and the learning rate of the model to obtain better model prediction accuracy.
Further, in the step a, for the identification of the missing value of the vehicle traffic state, linear interpolation is adopted for completion, and in the distortion identification, when the number of traffic state values 0 in each current and subsequent continuous 15 moments is more than 5, it is considered that the missing value is probably caused by a factor that the traffic of the road section is not busy enough, and the missing value is not processed; otherwise, if only 0 is considered, the data deviation is considered, and the same missing value is assigned with a non-zero value by linear interpolation.
Further, in the step B, the vehicle movement speed per unit time of the single road section is calculated as follows:
further, in the step C, for each data acquisition cycle, the state of each time segment corresponding to each numbered road segment in the cycle is extracted into a matrix, so as to obtain a one-day spatio-temporal data sample, and the data is converted into a road network state spatio-temporal matrix by using the following calculation formula:
further, in the step D, the road network data is converted into a matrix so as to conform to the input form of the neural network.
Furthermore, in the step E, for each sampling time, there is one corresponding road network state picture, and state information of the pictures at successive times can be extracted through a convolutional neural network, so as to finally abstract out a plurality of road network state features determined by the weight and the bias parameters; the input expression formula of the model is as follows:
Xi＝[xi，xi+1，...，xi+P-1]，i∈[1，T-P-K+1]
the invention has the beneficial effects that: in the application of the invention, a complex road network comprising ordinary roads, elevated roads, auxiliary roads, intersections and the like is taken as a research and calculation object. The invention utilizes the fixity of the road network, the consistency of the shape in each road network picture enables the corresponding software of the computer to be easier to understand, only the numerical value is changed along with the difference of the traffic state, and the single-channel matrix replaces the input of the multi-channel image, so that the data conforms to the input form of the neural network, and the invention provides a foundation for the road network state prediction model based on artificial intelligence. The invention effectively utilizes the automatic feature extraction of the moving traffic state data, synchronously excavates the space-time feature, avoids the loss of information between each frame of continuous images by applying more two-dimensional convolution, is beneficial to the excavation of time dimension rules, and effectively utilizes the historical state information to extract the space-time feature; the result obtained by calculation in the six steps is wider in targeted object coverage, the road network updating time scale can be reduced to two minutes, and the method has strong practical significance for management and control of an intelligent traffic system, so that powerful technical support is provided for implementation of construction of the Intelligent Traffic System (ITS). In conclusion, the invention has good application prospect.
Drawings
FIG. 1 is a flow chart of a road network state space-time prediction method based on a three-dimensional convolutional neural network;
FIG. 2 is a schematic diagram of the extraction of spatial features of a three-dimensional convolutional neural network.
Detailed Description
Data processing, namely identifying the conditions of missing, distortion and the like of data of a floating vehicle (the floating vehicle generally refers to a vehicle which is provided with a vehicle-mounted GPS positioning device and runs on a road) running on a common road, an overhead road, a subsidiary road, a complex road network at an intersection and the like, and finishing the cleaning and completion of the data; and B: b, calculating the road section state according to the data obtained in the step A, specifically calculating parameters such as traffic flow, density and speed, the number of vehicles, the vehicle speed and the saturation; and C: b, converting the data into a road network state space-time matrix according to the data obtained in the step B; step D: meshing road network data; e, establishing a three-dimensional convolutional neural network state prediction model according to the gridding data obtained in the step D, and completing the extraction of the dynamic road network state features after the datamation representation; step F: and E, optimizing the relevant content of the model obtained in the step E, and mainly adjusting an activation function, a learning rate and the like of the model to obtain better model prediction accuracy.
As shown in fig. 1 and 2, the following description of the present embodiment will take the method of the present invention as an example for predicting the traffic speed of a road network in a certain area, and illustrate the application principle and effectiveness of the present invention.
As shown in the figures 1 and 2, the method comprises the steps of A, processing data, namely identifying the conditions of missing, distortion and the like of the floating car data driven by the common roads, the overhead roads, the auxiliary roads, the complex road networks at intersections and the like, and finishing the cleaning and completion of the data; for the missing value of the vehicle speed, linear interpolation is adopted for completion, for example, the state values v1, v2, v3 and v4 at four adjacent moments are obtained, wherein v2 and v3 lack data, interpolation is carried out by using v1 and v4, and the formula is that v2 is v1+ (v4-v1)/3, and v3 is v1+2(v4-v 1)/3. The distortion mainly refers to that a certain traffic state value is 0, if the vehicle speed of a certain road section at a certain moment is 0, considering that the signal cycle length is possible to occur, and when the number of the traffic speed values of 0 in each current and subsequent continuous 15 moments is more than 5, the distortion is considered to be possibly caused by the factors of insufficient traffic of the road section and the like and is not processed; otherwise, if only 0 is considered, the data deviation is considered, and the same missing value is assigned with a non-zero value by linear interpolation.
Shown in fig. 1 and 2, step B: according to step ACalculating the road section state by the obtained data, specifically calculating parameters such as traffic flow, density and speed, and the number of vehicles, vehicle speed and saturation; in the calculation, the average speed is calculated from the average value of the average speeds of all vehicles passing through the corresponding road segment within a certain time period, as shown in formula (1), wherein l represents the l-th road segment in the road network, n represents the number of vehicles passing through the road segment in the time period, j is the time period number, s represents the length of the road segment, Δ t represents the length of the time period,
representing the average speed of the vehicle over the corresponding Δ t.
Shown in fig. 1 and 2, step C: b, converting the data into a road network state space-time matrix according to the data obtained in the step B; for each data acquisition cycle of 16 hours, each sampling interval is 2 minutes, and the state of each time corresponding to each numbered road section in the cycle is extracted into a matrix, so that a one-day spatio-temporal data sample is obtained, as shown in the following formula (2), wherein t represents the total time period number, and k represents the road section number.
Each row of the matrix comprises the time characteristics of the traffic state, the statistical period and the sampling frequency of each day are determined by floating car data and actual needs, the maximum analysis precision of the data on the time characteristics of the traffic flow is determined by the size of a time period, and when the interval is too large, the data change times in time is few, loss information is large, and the evolution rule of the traffic state is not beneficial to extraction; the interval is too small, and the fluctuation is too frequent although the amount of information is greatly increased. Each row of the matrix is a characteristic of the traffic flow at different spatial positions and represents the speed values of all road sections in the road network at each moment, and each row represents a vector of the state of the road network.
Shown in fig. 1 and 2, step D: meshing road network data; for the convolutional neural network, the data input is actually in a matrix form, so the road network needs to be converted into a matrix so as to conform to the input form of the neural network. In practical situations, the data gridding is to fill the average value of the traffic states of the corresponding passing road sections in each small grid area on the basis of dividing the road network into a plurality of grids, and the blank area is assigned to be zero. If two sections of curves pass through one intersection, all grid values in the range of the intersection are the mean value of the average speeds of the two road sections. Each grid is filled with data, so that the original smooth road network is directly substituted for curves in a small range, each pixel value represents a minimum space basic unit of a block of speed for the pixel value understood by a computer, and the grid number of the whole road network can be determined according to needs, so that the input information quantity can be effectively controlled. In this embodiment, each grid has a size of 10m × 10m, corresponding to a latitude and longitude of about (0.0001 ° ).
As shown in fig. 1 and 2, step E: and D, establishing a three-dimensional convolutional neural network state prediction model according to the gridding data obtained in the step D, and completing the extraction of the dynamic road network state features after the data representation. In practical application, each sampling moment is provided with a corresponding road network state picture, the state information of the pictures at the continuous moments can be extracted through a convolutional neural network, and finally a plurality of road network state features determined by parameters such as weight, bias and the like are abstracted. The three-dimensional convolutional neural network state prediction model is established by firstly determining the input and the output of the model. The gridding data is used as an input of a model for effectively expressing the road network state. Specifically, assuming that the input picture sequence is represented by X, the total time step is T, the input time length (i.e. the time step number) is K, and the predicted step size is P, the ith input of the model can be expressed as:
Xi＝[xi，xi+1，...，xi+P-1]，i∈[1，T-P-K+1]
considering that short-term traffic prediction and single-trip commuting time of most cities in China is about 30 minutes, it is appropriate to take time steps of 30 minutes, namely 15 moments, as the size of an input window, and information contained in 30 minutes is relatively large. And D, outputting the model to a certain row in the road network space-time matrix, using the space-time state matrix in the step D as input in training, mapping the features extracted by the convolutional neural network to an output vector space through full connection, and enabling the output to be in one-to-one correspondence with each road section in the road network.
Shown in fig. 1 and 2, step F: and E, optimizing the relevant content of the model obtained in the step E, and mainly adjusting an activation function, a learning rate and the like of the model to obtain better model prediction accuracy. The Adam optimizer and the PReLU activation function are selected after multiple attempts, a BN batch normalization layer and Dropout neuron inactivation are applied, and effects superior to those of other deep learning methods are obtained, and specific effects are shown in table 1 (in the table, MAE is an average absolute error value, MAPE is an average absolute percentage error, RMSE is a root mean square error; in the table, various data of a road network state space-time prediction method based on a three-dimensional convolutional neural network are smaller than those of other methods, and the fact that the method is obviously superior to other methods is shown).
Table 1 shows the comparison of the predicted effects of the methods used in the present invention.
In summary, in the application of the invention, a complex road network comprising ordinary roads, elevated roads, auxiliary roads and intersections is taken as a research and calculation object. The invention utilizes the fixity of the road network, the consistency of the shape in each road network picture enables the corresponding software of the computer to be easier to understand, only the numerical value is changed along with the difference of the traffic state, and the single-channel matrix replaces the input of the multi-channel image, so that the data conforms to the input form of the neural network, and the invention provides a foundation for the road network state prediction model based on artificial intelligence. The invention effectively utilizes the automatic feature extraction of the moving traffic state data, synchronously excavates the space-time feature, avoids the loss of information between each frame of continuous images by applying more two-dimensional convolution, is beneficial to the excavation of time dimension rules, and effectively utilizes the historical state information to extract the space-time feature. The result obtained by calculation in the six steps is wider in targeted object coverage, the road network updating time scale can be reduced to two minutes, and the method has strong practical significance for management and control of an intelligent traffic system, so that powerful technical support is provided for implementation of construction of the Intelligent Traffic System (ITS).
While there have been shown and described what are at present considered the fundamental principles and essential features of the invention and its advantages, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Furthermore, it should be understood that although the present description refers to embodiments, the embodiments do not include only one independent technical solution, and such description is only for clarity, and those skilled in the art should take the description as a whole, and the technical solutions in the embodiments may be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims (6)
1. A road network state space-time prediction method based on a three-dimensional convolutional neural network is used as a prediction tool through computer application, and is characterized in that the prediction is divided into six steps, specifically, the steps are data processing, namely, identifying the missing and distortion conditions of floating car data in the running of a road network to complete the cleaning and completion of the data; and B: b, calculating the road section state according to the data obtained in the step A, specifically calculating the traffic flow, the density, the speed parameters, the number of vehicles, the vehicle speed and the saturation; and C: b, converting the data into a road network state space-time matrix according to the data obtained in the step B; step D: meshing road network data; e, establishing a three-dimensional convolutional neural network state prediction model according to the gridding data obtained in the step D, and completing the extraction of the dynamic road network state features after the datamation representation; step F: and E, optimizing the relevant content of the model obtained in the step E, and mainly adjusting the activation function and the learning rate of the model to obtain better model prediction accuracy.
2. The road network state space-time prediction method based on the three-dimensional convolutional neural network as claimed in claim 1, wherein in the step a, for the identification of the missing value of the vehicle traffic state, linear interpolation is adopted for completion, and in the distortion identification, when the number of traffic state values 0 in each of the current and subsequent continuous 15 moments is more than 5, it is considered that the traffic state values are possibly caused by the fact that the road traffic is not busy enough, and the processing is not performed; otherwise, if only 0 is considered, the data deviation is considered, and the same missing value is assigned with a non-zero value by linear interpolation.
3. The road network state space-time prediction method based on the three-dimensional convolutional neural network as claimed in claim 1, wherein in step B, the vehicle motion speed per unit time of a single road section is calculated as follows:
4. the road network state space-time prediction method based on the three-dimensional convolutional neural network as claimed in claim 1, wherein in step C, for each data acquisition cycle, the state of each time segment in the cycle corresponding to each numbered road segment is extracted into a matrix, so as to obtain a one-day space-time data sample, and the calculation formula adopted by the data conversion into the road network state space-time matrix is as follows:
5. the road network state space-time prediction method based on the three-dimensional convolutional neural network as claimed in claim 1, wherein in step D, the road network data is converted into a matrix to make it conform to the input form of the neural network.
6. The road network state space-time prediction method based on the three-dimensional convolutional neural network as claimed in claim 1, wherein in step E, for each sampling time, there is one corresponding road network state picture, and the convolutional neural network can extract the state information of the pictures at successive times, and finally abstract out a plurality of road network state features determined by weight and bias parameters; the input expression formula of the model is as follows:
Xi＝[xi，xi+1，...，xi+P-1]，i∈[1，T-P-K+1]。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010836658.1A CN112100163A (en) | 2020-08-19 | 2020-08-19 | Road network state space-time prediction method based on three-dimensional convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010836658.1A CN112100163A (en) | 2020-08-19 | 2020-08-19 | Road network state space-time prediction method based on three-dimensional convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112100163A true CN112100163A (en) | 2020-12-18 |
Family
ID=73754081
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010836658.1A Pending CN112100163A (en) | 2020-08-19 | 2020-08-19 | Road network state space-time prediction method based on three-dimensional convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112100163A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114049764A (en) * | 2021-10-27 | 2022-02-15 | 东南大学 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
| CN116153069A (en) * | 2023-02-09 | 2023-05-23 | 东南大学 | Traffic flow model and data fusion driven traffic state estimation method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108629976A (en) * | 2018-05-17 | 2018-10-09 | 同济大学 | Urban traffic blocking predetermined depth learning method based on GPS |
| CN108647834A (en) * | 2018-05-24 | 2018-10-12 | 浙江工业大学 | A kind of traffic flow forecasting method based on convolutional neural networks structure |
| US20200135018A1 (en) * | 2018-10-24 | 2020-04-30 | Bluesignal Corporation | Method of predicting traffic congestion and controlling traffic signals based on deep learning and server for performing the same |
| CN111242292A (en) * | 2020-04-26 | 2020-06-05 | 北京全路通信信号研究设计院集团有限公司 | OD data prediction method and system based on deep space-time network |
-
2020
- 2020-08-19 CN CN202010836658.1A patent/CN112100163A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108629976A (en) * | 2018-05-17 | 2018-10-09 | 同济大学 | Urban traffic blocking predetermined depth learning method based on GPS |
| CN108647834A (en) * | 2018-05-24 | 2018-10-12 | 浙江工业大学 | A kind of traffic flow forecasting method based on convolutional neural networks structure |
| US20200135018A1 (en) * | 2018-10-24 | 2020-04-30 | Bluesignal Corporation | Method of predicting traffic congestion and controlling traffic signals based on deep learning and server for performing the same |
| CN111242292A (en) * | 2020-04-26 | 2020-06-05 | 北京全路通信信号研究设计院集团有限公司 | OD data prediction method and system based on deep space-time network |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114049764A (en) * | 2021-10-27 | 2022-02-15 | 东南大学 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
| CN116153069A (en) * | 2023-02-09 | 2023-05-23 | 东南大学 | Traffic flow model and data fusion driven traffic state estimation method and device |
| CN116153069B (en) * | 2023-02-09 | 2024-01-30 | 东南大学 | Traffic flow model and data fusion driven traffic state estimation method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102021992B1 (en) | Apparatus for controling a trafic signal, method for controling a trafic signal, and recoding medium for controling a tarfic signal | |
| CN109544908B (en) | Method for predicting bus arrival time in real time, electronic equipment and storage medium | |
| CN108805348B (en) | Method and device for controlling and optimizing intersection signal timing | |
| CN113487066B (en) | Long-time-sequence freight volume prediction method based on multi-attribute enhanced graph convolution-Informer model | |
| CN110570651A (en) | Road network traffic situation prediction method and system based on deep learning | |
| CN109272157A (en) | A kind of freeway traffic flow parameter prediction method and system based on gate neural network | |
| CN112330215A (en) | Urban vehicle demand prediction method, equipment and storage medium | |
| CN112633602B (en) | Traffic congestion index prediction method and device based on GIS map information | |
| CN115063978B (en) | Bus arrival time prediction method based on digital twins | |
| CN110599236A (en) | Short-time parking demand prediction method based on GRU model | |
| Li et al. | Deep learning based parking prediction on cloud platform | |
| CN111275965A (en) | Real-time traffic simulation analysis system and method based on internet big data | |
| CN112100163A (en) | Road network state space-time prediction method based on three-dimensional convolutional neural network | |
| WO2021115320A1 (en) | Traffic evaluation method and system | |
| CN113112791A (en) | Traffic flow prediction method based on sliding window long-and-short term memory network | |
| CN115204478A (en) | Public traffic flow prediction method combining urban interest points and space-time causal relationship | |
| Li et al. | Model controlled prediction: A reciprocal alternative of model predictive control | |
| CN111860621B (en) | Data-driven distributed traffic flow prediction method and system | |
| CN118297775B (en) | Urban planning management and control system based on digital twin technology | |
| CN113674524A (en) | LSTM-GASVR-based multi-scale short-time traffic flow prediction modeling and prediction method and system | |
| CN113821547B (en) | Rapid and efficient short-time prediction method, system and storage medium for occupancy of parking lot | |
| CN109064750B (en) | Urban road network traffic estimation method and system | |
| CN112613630B (en) | Short-term traffic demand prediction method integrating multi-scale space-time statistical information | |
| CN114091772B (en) | Multi-angle fusion road traffic flow prediction method based on encoder and decoder | |
| Liu et al. | Traffic flow prediction based on spatio-temporal attention block |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20201218 |