CN111639791A - Traffic flow prediction method, system, storage medium and terminal - Google Patents
Traffic flow prediction method, system, storage medium and terminal Download PDFInfo
- Publication number
- CN111639791A CN111639791A CN202010393242.7A CN202010393242A CN111639791A CN 111639791 A CN111639791 A CN 111639791A CN 202010393242 A CN202010393242 A CN 202010393242A CN 111639791 A CN111639791 A CN 111639791A
- Authority
- CN
- China
- Prior art keywords
- traffic flow
- layer
- tensor
- module
- fuzzy
- 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
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000013528 artificial neural network Methods 0.000 claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 25
- 230000006403 short-term memory Effects 0.000 claims abstract description 9
- 230000007787 long-term memory Effects 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims abstract description 7
- 230000004927 fusion Effects 0.000 claims description 63
- 230000003442 weekly effect Effects 0.000 claims description 27
- 230000006870 function Effects 0.000 claims description 24
- 230000015654 memory Effects 0.000 claims description 14
- 238000013459 approach Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 10
- 239000000126 substance Substances 0.000 claims description 10
- AGVAZMGAQJOSFJ-WZHZPDAFSA-M cobalt(2+);[(2r,3s,4r,5s)-5-(5,6-dimethylbenzimidazol-1-yl)-4-hydroxy-2-(hydroxymethyl)oxolan-3-yl] [(2r)-1-[3-[(1r,2r,3r,4z,7s,9z,12s,13s,14z,17s,18s,19r)-2,13,18-tris(2-amino-2-oxoethyl)-7,12,17-tris(3-amino-3-oxopropyl)-3,5,8,8,13,15,18,19-octamethyl-2 Chemical compound [Co+2].N#[C-].[N-]([C@@H]1[C@H](CC(N)=O)[C@@]2(C)CCC(=O)NC[C@@H](C)OP(O)(=O)O[C@H]3[C@H]([C@H](O[C@@H]3CO)N3C4=CC(C)=C(C)C=C4N=C3)O)\C2=C(C)/C([C@H](C\2(C)C)CCC(N)=O)=N/C/2=C\C([C@H]([C@@]/2(CC(N)=O)C)CCC(N)=O)=N\C\2=C(C)/C2=N[C@]1(C)[C@@](C)(CC(N)=O)[C@@H]2CCC(N)=O AGVAZMGAQJOSFJ-WZHZPDAFSA-M 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 9
- 230000002354 daily effect Effects 0.000 description 24
- 238000012545 processing Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013531 bayesian neural network Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- 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
-
- 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
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
Abstract
The invention provides a traffic flow prediction method, a system, a storage medium and a terminal; the method comprises the following steps: constructing a traffic flow space-time tensor according to the traffic data; building a traffic flow prediction model based on a fuzzy convolution long-term and short-term memory network; training a traffic flow prediction model based on the traffic flow space-time tensor and the traffic data to obtain the trained traffic flow prediction model; predicting the traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data and the trained traffic flow prediction model; the invention considers the uncertainty of the traffic data and processes the data uncertainty by using fuzzy learning, wherein the fuzzy rule can be adaptively learned without depending on human experience; when traffic flow prediction is carried out, the influence of time-space correlation, external factors and data uncertainty of the traffic flow is fully considered, and the deep convolution LSTM network and the fuzzy neural network are fused, so that the traffic flow prediction effect is more accurate and reliable.
Description
Technical Field
The invention belongs to the field of road traffic monitoring, and particularly relates to a traffic flow prediction method, a traffic flow prediction system, a storage medium and a terminal.
Background
With the rapid development of economy in China, the large-scale increase of the production and the use of vehicles, more and more cities face the problem of Traffic jam, the life and the development of the cities are seriously influenced, and Traffic flow prediction as an important component of an Intelligent Traffic System (ITS) is an important part for ensuring the normal work of the ITS, so that the Traffic flow prediction method has important significance for relieving Traffic pressure and ensuring smooth Traffic.
At present, the existing methods for predicting the traffic flow can be roughly divided into three categories based on data and statistics, a shallow neural network and a deep learning neural network; traffic flow prediction models based on data and statistics, such as a historical Average (historical Average) model and an ARIMA (ARIMA) model, cannot effectively process traffic data; shallow neural network models such as Bayesian neural network models and Kalman filtering models are easy to fall into local extreme values, large-scale complex calculation cannot be effectively simulated, and ideal effects cannot be achieved on complex problems such as traffic flow prediction; the deep learning neural network model achieves certain achievements in the aspect of traffic flow prediction; for example, zhengyu et al explores the spatiotemporal characteristics of traffic flow using a deep convolution residual error network; the rest oceans and the like use a deep convolutional network to explore the spatial characteristics of the traffic flow, and use a Long Short-Term Memory network (LSTM) to explore the spatial characteristics of the traffic flow, and although the deep learning models all obtain good prediction effects, the uncertainty of the data is not considered; chenvian macro et al use fuzzy learning to deal with uncertainty in traffic data, but ignore the effect of external factors on traffic flow.
Disclosure of Invention
In view of the above-mentioned drawbacks of the prior art, an object of the present invention is to provide a traffic flow prediction method, a system, a storage medium, and a terminal, which are used to solve the problem of the prior art that the reliability of the traffic flow prediction result is reduced due to insufficient consideration of processing traffic data when traffic flow prediction is performed.
To achieve the above and other related objects, the present invention provides a traffic flow prediction method, including the steps of: constructing a traffic flow space-time tensor according to the traffic data; building a traffic flow prediction model based on a fuzzy convolution long-term and short-term memory network; training the traffic flow prediction model based on the traffic flow space-time tensor and the traffic data to obtain a trained traffic flow prediction model; and predicting the traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data and the trained traffic flow prediction model.
In an embodiment of the present invention, the formula of the traffic flow space-time tensor constructed according to the traffic data is:
wherein the content of the first and second substances,
and
respectively representing inflow and outflow of the regions of the ith row and the jth column in the tth time interval; obtaining a set of all GPS tracks of which P represents the t-th time interval; gkGeographical position, g, of the kth GPS point representing the track Trk∈ (i, j) denotes gkThe position is in the area where the ith row and the jth column are positioned; gk-1The geographic location of the (k-1) th GPS point representing the trajectory Tr;
denotes gk-1Not in the area of the ith row and the jth column; gk+1The geographic position of the (k +1) th GPS point representing the trajectory Tr;
denotes gk+1The traffic flow of all the regions in the t time interval of a city is expressed by a three-dimensional space-time tensor 2 × I × J, namely a traffic flow space-time tensor Xt，Xt∈R2×I×J(ii) a Wherein
0 represents an inflow;
and 1 denotes the outflow.
In an embodiment of the present invention, the traffic flow prediction model includes an external factor module, a near trend module, a daily trend module, a week trend module and a fusion module; wherein the external factor module comprises a first fully connected layer and a second fully connected layer; the second full connection layer is used for mapping the low-dimensional external factor prediction tensor output by the first full connection layer into a high-dimensional external factor prediction tensor; the approach trend module, the daily trend module and the weekly trend module respectively comprise a fuzzy neural network, a deep neural network, a fusion layer and at least one third full-connection layer; the fusion layer is used for fusing the output of the fuzzy neural network and the output of the deep neural network; the third full connection layer is used for continuously learning the fusion result of the fusion layer and outputting a prediction tensor; the adjacent trend module, the daily trend module and the weekly trend module respectively output an adjacent prediction tensor, a daily prediction tensor and a weekly prediction tensor; the fusion module is used for carrying out first fusion on the adjacent prediction tensor, the daily prediction tensor and the weekly prediction tensor to generate a first fusion result; and the external factor prediction tensor is used for fusing the first fusion result with the high-dimensional external factor prediction tensor again.
In an embodiment of the present invention, the fuzzy neural network includes an input layer, a fuzzy layer and a fuzzy rule layer; each node in the fuzzy layer represents a membership function, and the degree of membership of the input node to the fuzzy set is calculated; the formula of the membership function is as follows:
wherein u isiRepresenting a gaussian function; mu.siAnd σiRespectively representing the center and width, μ, of the Gaussian functioniAnd σiAre all learnable parameters;
representing the output of a node i in the l layers of fuzzification layers;
representing the output of a node k in the l-1 layer, wherein the node k is connected with a node i; in the fuzzy rule layer, executing and operation, wherein the formula is as follows:
wherein the content of the first and second substances,
the output of a node j in the l-layer fuzzy layer is shown, and the node j is connected with a node m in the l + 1-layer fuzzy rule layer; n represents the number of nodes in the l-layer fuzzy layer connected with the node m in the l + 1-layer fuzzy rule layer;
representing the output of node m in the l +1 fuzzy rule layer.
In an embodiment of the present invention, the fusion layer is a fully connected layer, and the fusion formula is:
wherein o isfAnd odRespectively representing the output of the fuzzy neural network and the output of the deep neural network; w is afAnd wdWeights representing an output of the fuzzy neural network and an output of the deep neural network, respectively;
represents a bias value; w is af、wdAnd
are all learnable parameters;
represents the output (o) of a l-1 layer fuzzy neural network through the fusion layerf)(l-1)And the output (o) of the l-1 layer deep neural networkd)(l-1)And performing fusion on the fusion layer nodes i of the layer I after fusion.
In an embodiment of the present invention, the fusion module performs fusion operation in a manner based on a parameter matrix; the first fused formula is:
wherein the content of the first and second substances,
representing a Hadamard product; xc、Xd、XwRespectively representing an adjacent prediction tensor, a daily prediction tensor and a weekly prediction tensor; wc、Wd、WwRespectively representing X for corresponding learnable parametersc、Xd、XwThe degree of influence of (c); xFConvRepresenting a first fusion result;
the formula for re-fusion is:
wherein, XExtAn extrinsic factor prediction tensor representing a high dimension;
and a traffic flow prediction tensor representing an output of the traffic flow prediction model.
In an embodiment of the present invention, the method further includes: establishing a loss function, and training the traffic flow prediction model through the loss function until the value of the loss function is not reduced any more; the calculation formula of the loss function is as follows:
wherein theta represents all learnable parameters in the traffic flow prediction model; xtRepresenting a true traffic flow space-time tensor; and randomly deleting or ignoring at least one node and connection thereof in the process of training the traffic flow prediction model.
The present invention provides a traffic flow prediction system, including: the system comprises a construction module, a building module, a training module and a prediction module; the construction module is used for constructing a traffic flow space-time tensor according to the traffic data; the building module is used for building a traffic flow prediction model based on the fuzzy convolution long-term and short-term memory network; the training module is used for training the traffic flow prediction model based on the traffic flow space-time tensor and the traffic data to obtain a trained traffic flow prediction model; the prediction module is used for predicting the traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data and the trained traffic flow prediction model.
The present invention provides a storage medium having stored thereon a computer program which, when executed by a processor, implements the traffic flow prediction method described above.
The present invention provides a terminal, including: a processor and a memory; the memory is used for storing a computer program; the processor is used for executing the computer program stored in the memory so as to enable the terminal to execute the traffic flow prediction method.
As described above, the traffic flow prediction method, system, storage medium, and terminal according to the present invention have the following advantageous effects:
(1) compared with the prior art, an end-to-end traffic flow prediction model is provided, and the model not only explores the spatial correlation of traffic flow, but also explores the correlation of the traffic flow from three different scales of approach trends, daily trends and weekly trends;
(2) the uncertainty of the traffic data is considered, and the data uncertainty is processed by fuzzy learning, wherein fuzzy rules can be adaptively learned without depending on human experience;
(3) when traffic flow prediction is carried out, the influence of time-space correlation, external factors and data uncertainty of the traffic flow is fully considered, and the deep convolution LSTM network and the fuzzy neural network are fused, so that the traffic flow prediction effect is more accurate and reliable.
Drawings
Fig. 1 is a flowchart illustrating a traffic flow prediction method according to an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of the area division in an embodiment of the invention.
FIG. 3 is a schematic diagram of an embodiment of an outflow matrix according to the present invention.
FIG. 4 is a block diagram of a traffic flow prediction model according to an embodiment of the present invention.
FIG. 5 is a schematic diagram illustrating an internal structure of a trend block according to an embodiment of the present invention.
Fig. 6 is a schematic diagram illustrating an internal structure of a deep neural network according to an embodiment of the present invention.
Fig. 7 is a schematic view illustrating a traffic flow prediction system according to an embodiment of the invention.
Fig. 8 is a schematic structural diagram of a terminal according to an embodiment of the invention.
Description of the reference symbols
71 building block
72 building module
73 training module
74 prediction module
81 processor
82 memory
S1-S4
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.
It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.
The traffic flow prediction method, the system, the storage medium and the terminal of the invention provide an end-to-end traffic flow prediction model, which not only explores the spatial correlation of traffic flow, but also explores the correlation of traffic flow from three different scales of approach trend, daily trend and weekly trend; the uncertainty of the traffic data is considered, and the data uncertainty is processed by fuzzy learning, wherein fuzzy rules can be adaptively learned without depending on human experience; when traffic flow prediction is carried out, the influence of time-space correlation, external factors and data uncertainty of the traffic flow is fully considered, and the deep convolution LSTM network and the fuzzy neural network are fused, so that the traffic flow prediction effect is more accurate and reliable.
As shown in fig. 1, in one embodiment, the traffic flow prediction method of the present invention includes the following steps:
and step S1, constructing a traffic flow space-time tensor according to the traffic data.
Note that the traffic data includes the GPS track of the vehicle and external factors (such as weather, temperature, holidays, etc.).
Specifically, traffic data are collected, and a traffic flow space-time tensor is constructed according to the collected traffic data.
As shown in fig. 2, in a practical application, a city is divided into a plurality of areas uniformly; for each area, two indices are recorded: inflow and outflow; inflow refers to the total number of moving objects (including pedestrians, vehicles, etc.) entering the area per unit time, and outflow refers to the flow rate leaving the area per unit time.
In one embodiment, the formula for constructing the traffic flow space-time tensor according to the traffic data is as follows:
the inflow formula is defined as:
the outflow formula is defined as:
wherein the content of the first and second substances,
and
respectively representing inflow and outflow of the regions of the ith row and the jth column in the tth time interval; obtaining a set of all GPS tracks of which P represents the t-th time interval; gkGeographical position, g, of the kth GPS point representing the track Trk∈ (i, j) denotes gkThe position is in the area where the ith row and the jth column are positioned; gk-1The geographic location of the (k-1) th GPS point representing the trajectory Tr;
denotes gk-1Not in the area of the ith row and the jth column; gk+1The geographic position of the (k +1) th GPS point representing the trajectory Tr;
denotes gk+1Not in the area of the ith row and the jth column.
The traffic flow of all areas in the t-th time interval of a city is expressed by a three-dimensional space-time tensor of 2 × I × J as a traffic flow space-time tensor Xt，Xt∈R2×I×J(ii) a Wherein
0 represents an inflow;
1 denotes the outflow; the outflow matrix at time t is shown in FIG. 3.
And step S2, building a traffic flow prediction model based on the fuzzy convolution long-term and short-term memory network.
It should be noted that, the prediction of regional traffic flow in the urban area is influenced not only by the spatial and temporal correlation of the traffic flow but also by external factors (weather, temperature, holidays, etc.); in addition, a large amount of data is required for predicting traffic flow in the city-wide range, and a problem of uncertainty of data inevitably occurs as the amount of data increases; a convolution long-short term memory network (Convolitional LSTM, which can be abbreviated as ConvLSTM) can simultaneously capture the space-time correlation of traffic data, and fuzzy learning has the advantage that other learning methods are difficult to replace in the aspect of processing data uncertainty; based on the Fuzzy Convolutional long and short term memory network (FConvLSTM) model, the traffic flow in the whole city range is predicted, namely the traffic flow prediction model.
It should be noted that the traffic flow prediction model successively explores the characteristics of traffic flow from three different time scales (time) -a closing Trend (closetend), a Daily Trend (Daily Trend) and a Weekly Trend (Weekly Trend); the approach trend refers to that the flow (inflow and outflow) in the current time interval is influenced by the flow (defined as being nearest to the current time on a time scale and represented by a receiver) in the approach time interval, for example, the traffic jam at 8 am influences the traffic flow at 9 am, and the traffic flow in the approach time interval does not change drastically; the trend of days means that the traffic flow in the same time interval is similar every day (defined as being close to the current time on a time scale and represented by near), for example, the traffic conditions of the peak in the morning and at the evening are similar every day on the continuous working day, and are repeated every 24 hours; week trend means that the traffic flow is similar or slowly changing (defined on a time scale as being further away from the current time, indicated by distance) in the same time interval every week, for example, the morning and evening peak times of the weekend are similar to the weekend of the week.
As shown in fig. 4, in one embodiment, the traffic flow prediction model includes an external factor module, a proximity trend module, a trend-by-date module, a trend-by-week module, and a fusion module.
Wherein the external factor module comprises a first fully connected layer and a second fully connected layer; the external factor module simulates the influence of external factors (Exact Features) on the traffic flow at the time t through two fully-connected layers; the second full connection layer is used for mapping the low-dimensional external factor prediction tensor output by the first full connection layer into a high-dimensional external factor prediction tensor; specifically; the first full connection layer can be regarded as an embedded layer, and after the external factors are input into the first full connection layer, the first full connection layer outputs a low-dimensional external factor prediction tensor; the second fully-connected layer is used to map the low-dimensional extrinsic factor prediction tensor into a high-dimensional extrinsic factor prediction tensor.
In order to make the high-dimensional extrinsic factor prediction tensor dimension output by the extrinsic factor module and the real traffic flow space-time tensor XtSimilarly, the number of nodes in the second fully-connected layer is 2 × I × J.
It should be noted that the number of the full connection layers included in the external factor module is not limited to two, and the greater the number of the full connection layers is, the more effective the external factor module simulates the external factor, but the more complicated the module structure is; in one embodiment, two fully connected layers are preferably selected.
The approach trend module, the daily trend module and the weekly trend module respectively comprise a fuzzy neural network (FN), a deep neural network (DN), a fusion layer (fusion layer) and at least one third full-link layer.
The fusion layer is used for fusing the output of the fuzzy neural network and the output of the deep neural network, and the output of the fusion layer can be regarded as a general expression of traffic data.
The third full connection layer is used for continuously learning the fusion result of the fusion layer and outputting a prediction tensor; specifically, the third fully connected layer is a generic expression used to continue learning traffic data; the adjacent trend module, the daily trend module and the weekly trend module respectively output an adjacent prediction tensor, a daily prediction tensor and a weekly prediction tensor.
It should be noted that the number of the third fully-connected layers is not a condition for limiting the present invention; in practical applications, the larger the number of the third fully-connected layers, the better the learning, but also the more complicated the modular structure, preferably two, the output of the last third fully-connected layer is taken as the prediction tensor.
Further, in practical applications, the result of fusing any one or a combination of several of the external factor prediction tensor, the adjacent prediction tensor, the daily prediction tensor and the weekly prediction tensor can be used as the traffic flow prediction tensor output by the traffic flow prediction model.
Taking the adjacent trend module as an example, the detailed discussion of the internal structure thereof is performed.
As shown in fig. 5, in one embodiment, the fuzzy neural network includes an input layer, a fuzzy layer and a fuzzy rule layer.
And each node in the fuzzy layer represents a membership function, and the degree of membership of the input node to the fuzzy set is calculated.
In one embodiment, the most widely used gaussian function is selected as the membership function, which is formulated as:
wherein u isiRepresenting a gaussian function; mu.siAnd σiRespectively representing the center and width, μ, of the Gaussian functioniAnd σiAre all learnable parameters;
representing the output of a node i in the l layers of fuzzification layers;
representing the output of node k in layer l-1, node k being connected to node i.
In particular, the ratio of l to 2,
the output of the node k of layer 1 is shown because the first layer is an input layer and no operation is performed on the input quantity, and therefore
In the fuzzy rule layer, each node represents an "if part" of a fuzzy rule, the number of nodes in the layer represents the number of rules, AND (AND) operation is usually performed in the fuzzy rule layer, AND the formula is:
wherein the content of the first and second substances,
the output of a node j in the l-layer fuzzy layer is shown, and the node j is connected with a node m in the l + 1-layer fuzzy rule layer; n represents the number of nodes in the l-layer fuzzy layer connected with the node m in the l + 1-layer fuzzy rule layer;
and (3) representing the output of the node m in the l +1 fuzzy rule layer, wherein the output can be regarded as the fuzzy degree.
It should be noted that, layer l is an obfuscation layer; the l +1 layer is a fuzzy rule layer.
It should be noted that the essence of the fuzzy rule is a binary fuzzy relation R defined at X × Y; the fuzzy rule is of the form: if x is A the y is B'.
It should be noted that the fuzzy neural network generally further includes a normalization layer and a deblurring layer, which are used to execute the "then" part of the fuzzy rule; however, the first three layers of the fuzzy neural network already have a fuzzy representation of the input data available, so that the normalization layer and the deblurring layer of the fuzzy neural network are not required here.
Fusing the output of the fuzzy neural network with the output of the deep neural network, and adaptively adjusting the fuzzy rule; notably, fuzzy neural networks are different from traditional neural networks; the knowledge of the fuzzy neural network is stored in the nodes, and the knowledge of the traditional neural network is stored in the weights between layers, which indicates that the weights of the traditional neural network are parameters to be learned, and the parameters in the nodes (such as membership function nodes) of the fuzzy neural network are parameters which can be learned; thus, μiAnd σiAre parameters that need to be learned, and the weights between the input layer, the fuzzy layer, and the fuzzy rule layer, which are set to 1, do not need to be learned.
As shown in FIG. 6, in one embodiment, the deep neural network selects a convolutional LSTM network, which can represent the input data as some kind of high-level feature representation, and the convolutional LSTM can simultaneously explore the spatio-temporal correlation of the traffic data; to obtain deeper features, a multi-layer convolution LSTM overlay may be used; in particular, the deep neural network includes an input layer and at least one convolutional LSTM network.
It should be noted that the dashed lines in fig. 6 indicate that the number of convolutional LSTM networks is not limited.
In one embodiment, the fusion layer is a fully connected layer, and the fusion formula is as follows:
wherein o isfAnd odRespectively representing the output of the fuzzy neural network (namely the output of the fuzzy rule layer) and the output of the deep neural network; w is afAnd wdWeights representing an output of the fuzzy neural network and an output of the deep neural network, respectively;
represents a bias value; w is af、wdAnd
are all learnable parameters;
represents the output (o) of a l-1 layer fuzzy neural network through the fusion layerf)(l-1)And the output (o) of the l-1 layer deep neural networkd)(l-1)And performing fusion on the fusion layer nodes i of the layer I after fusion.
Here, the layer "l" represents a fusion layer; the l-1 layer is the previous layer of the fusion layer and comprises the last layer of the fuzzy neural network and the last layer of the deep neural network.
It should be noted that the output dimension of the fuzzy neural network is different from that of the deep neural network, and they are fused together through a full connection layer; the number of nodes of the fusion layer is equal to the sum of the number of nodes of the last layer of the fuzzy neural network and the number of nodes of the last layer of the deep neural network.
It should be noted that the internal structure and the operation principle of the trend module and the trend module are the same as those of the adjacent trend module, and are not described herein again.
As shown in fig. 5, in an embodiment, the approach trend module further includes a task driving layer, which is located at the last layer and is used for driving the proceeding of the subsequent task.
The fusion module is used for carrying out first fusion on the adjacent prediction tensor, the daily prediction tensor and the weekly prediction tensor to generate a first fusion result; and the external factor prediction tensor is used for fusing the first fusion result with the high-dimensional external factor prediction tensor again.
Specifically, the fusion module fuses the proximity prediction tensor, the daily prediction tensor and the weekly prediction tensor which are respectively output by the proximity trend module, the daily trend module and the weekly trend module, and then fuses the proximity prediction tensor, the daily prediction tensor and the weekly prediction tensor with the external factor prediction tensor after the fusion; although all regions are affected by the proximity trend, the daily trend and the weekly trend, the degrees of the influence of the trends are different for different regions, and the fusion can be selected in a mode based on a parameter matrix.
In one embodiment, the fusion module performs fusion operation in a mode based on a parameter matrix; the first fused formula is:
wherein the content of the first and second substances,
representing a Hadamard product; xc、Xd、XwRespectively representing an adjacent prediction tensor, a daily prediction tensor and a weekly prediction tensor; wc、Wd、WwRespectively representing X for corresponding learnable parametersc、Xd、XwThe degree of influence of (c); xFConvRepresenting a first fusion result;
the formula for re-fusion is:
wherein, XExtAn extrinsic factor prediction tensor representing a high dimension;
and a traffic flow prediction tensor representing an output of the traffic flow prediction model.
And S3, training the traffic flow prediction model based on the traffic flow space-time tensor and the traffic data, and acquiring the trained traffic flow prediction model.
Specifically, the traffic flow space-time tensor obtained in step S1 and the external factors in the traffic data are input into the traffic flow prediction model constructed in step S2, and all learnable parameters including μ in step S2 are obtained by continuously training the traffic flow prediction modeli、σi、wf、wd、
Wc、WdAnd WwFinally, the traffic flow prediction model with the determined learnable parameters, namely the trained traffic flow prediction model, is obtained.
The traffic flow space-time tensor includes a current traffic flow space-time tensor of the vehicle at a current time t, and a neighboring tendency tensor, a daily tendency tensor, and a week tendency tensor in a section nearest to the time t, a section near the time t, and a section far away from the time t, respectively.
In one embodiment, the method further comprises the steps of establishing a loss function, training the traffic flow prediction model through the loss function, and stopping training until the value of the loss function is not reduced any more; the calculation formula of the loss function is as follows:
wherein theta represents all learnable parameters in the traffic flow prediction model; xtThe actual traffic flow space-time tensor (i.e., the traffic flow space-time tensor acquired through step S1) is represented.
Specifically, the traffic flow prediction model is trained by the mean square error between the traffic flow prediction tensor output by each training of the traffic flow prediction model and the real traffic flow space-time tensor input into the traffic flow prediction model.
In one embodiment, the method further comprises randomly deleting or ignoring at least one node and its connection in the process of training the traffic flow prediction model.
It should be noted that, a time Back Propagation (BPTT) algorithm and an Adam algorithm are adopted to train a traffic flow prediction model, the traffic flow prediction model is complex and has many trainable parameters, overfitting is easy to occur in the training process, and in order to prevent overfitting, a dropout technology is selected and used, and the core idea is that some nodes and the connection between the nodes and a network are randomly deleted or ignored in the training process; specifically, in practical application, part of nodes and connections thereof of the penultimate fully-connected layer (the third fully-connected layer) in each module of the adjacent trend, the daily trend and the weekly trend are randomly lost.
And step S4, predicting the traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data and the trained traffic flow prediction model.
Specifically, the acquired real-time traffic flow space-time tensor and external factors in the real-time traffic data are input into a trained traffic flow prediction model, and the traffic flow prediction model outputs a traffic flow prediction tensor so as to realize the prediction of the traffic flow.
The real-time traffic flow space-time tensor includes a current traffic flow space-time tensor at a certain time, and a neighboring trend tensor, a daily trend tensor, and a week trend tensor corresponding to a nearest, an approaching, and a remote place at the time.
It should be noted that the protection scope of the traffic flow prediction method according to the present invention is not limited to the execution sequence of the steps illustrated in the embodiment, and all the solutions implemented by adding, subtracting, and replacing the steps in the prior art according to the principle of the present invention are included in the protection scope of the present invention.
As shown in fig. 7, in an embodiment, the traffic flow prediction system of the present invention includes a construction module 71, a construction module 72, a training module 73, and a prediction module 74.
The construction module 71 is configured to construct a traffic flow space-time tensor according to the traffic data.
The building module 72 is used for building a traffic flow prediction model based on the fuzzy convolution long-term and short-term memory network.
The training module 73 is configured to train the traffic flow prediction model based on the traffic flow space-time tensor and the traffic data, and obtain a trained traffic flow prediction model.
The prediction module 74 is configured to predict traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data, and the trained traffic flow prediction model.
It should be noted that the structures and principles of the building module 71, the building module 72, the training module 73, and the prediction module 74 correspond to the steps in the traffic flow prediction method one to one, and therefore, the description is omitted here.
It should be noted that the division of the modules of the above system is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. And these modules can be realized in the form of software called by processing element; or may be implemented entirely in hardware; and part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the x module may be a processing element that is set up separately, or may be implemented by being integrated in a chip of the system, or may be stored in a memory of the system in the form of program code, and the function of the x module may be called and executed by a processing element of the system. Other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element described herein may be an integrated circuit having signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.
For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as: one or more Application Specific Integrated Circuits (ASICs), or one or more Digital Signal Processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs), etc. For another example, when one of the above modules is implemented in the form of a Processing element scheduler code, the Processing element may be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).
The storage medium of the present invention stores thereon a computer program that realizes the above-described traffic flow prediction method when executed by a processor. The storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
As shown in fig. 8, the terminal of the present invention includes a processor 81 and a memory 82.
The memory 82 is used to store computer programs. Preferably, the memory 82 includes: various media that can store program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
The processor 81 is connected to the memory 82, and is configured to execute the computer program stored in the memory 82, so that the terminal executes the traffic flow prediction method.
Preferably, the Processor 81 may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; the integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, or discrete hardware components.
It should be noted that the traffic flow prediction system of the present invention can implement the traffic flow prediction method of the present invention, but the implementation apparatus of the traffic flow prediction method of the present invention includes, but is not limited to, the structure of the traffic flow prediction system described in the embodiment, and all the structural modifications and substitutions of the prior art made according to the principle of the present invention are included in the protection scope of the present invention.
In summary, the traffic flow prediction method, the system, the storage medium and the terminal of the invention provide an end-to-end traffic flow prediction model, which not only explores the spatial correlation of traffic flow, but also explores the correlation of traffic flow from three different scales of approaching trend, daily trend and weekly trend; the uncertainty of the traffic data is considered, and the data uncertainty is processed by fuzzy learning, wherein fuzzy rules can be adaptively learned without depending on human experience; when traffic flow prediction is carried out, the influence of time-space correlation, external factors and data uncertainty of the traffic flow is fully considered, and the deep convolution LSTM network and the fuzzy neural network are fused, so that the traffic flow prediction effect is more accurate and reliable. Therefore, the invention effectively overcomes various defects in the prior art and has high industrial utilization value.
The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (10)
1. A traffic flow prediction method is characterized by comprising the following steps:
constructing a traffic flow space-time tensor according to the traffic data;
building a traffic flow prediction model based on a fuzzy convolution long-term and short-term memory network;
training the traffic flow prediction model based on the traffic flow space-time tensor and the traffic data to obtain a trained traffic flow prediction model;
and predicting the traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data and the trained traffic flow prediction model.
2. The traffic flow prediction method according to claim 1, characterized in that the formula for constructing the traffic flow space-time tensor from traffic data is:
wherein the content of the first and second substances,
and
respectively representing inflow and outflow of the regions of the ith row and the jth column in the tth time interval; p represents all GPS track sets in the t-th time interval; gkGeographical position, g, of the kth GPS point representing the track Trk∈ (i, j) denotes gkThe position is in the area where the ith row and the jth column are positioned; gk-1The geographic location of the (k-1) th GPS point representing the trajectory Tr;
denotes gk-1Not in the area of the ith row and the jth column; gk+1The geographic position of the (k +1) th GPS point representing the trajectory Tr;
denotes gk+1Not in the area of the ith row and the jth column;
the traffic flow of all areas in the t-th time interval of a city is expressed by a three-dimensional space-time tensor of 2 × I × J as a traffic flow space-time tensor Xt，Xt∈R2×I×J(ii) a Wherein
0 represents an inflow;
and 1 denotes the outflow.
3. The traffic flow prediction method according to claim 1, wherein the traffic flow prediction model includes an external factor module, an approach trend module, a daily trend module, a weekly trend module, and a fusion module;
wherein the external factor module comprises a first fully connected layer and a second fully connected layer; the second full connection layer is used for mapping the low-dimensional external factor prediction tensor output by the first full connection layer into a high-dimensional external factor prediction tensor;
the approach trend module, the daily trend module and the weekly trend module respectively comprise a fuzzy neural network, a deep neural network, a fusion layer and at least one third full-connection layer; the fusion layer is used for fusing the output of the fuzzy neural network and the output of the deep neural network; the third full connection layer is used for continuously learning the fusion result of the fusion layer and outputting a prediction tensor; the adjacent trend module, the daily trend module and the weekly trend module respectively output an adjacent prediction tensor, a daily prediction tensor and a weekly prediction tensor;
the fusion module is used for carrying out first fusion on the adjacent prediction tensor, the daily prediction tensor and the weekly prediction tensor to generate a first fusion result; and the external factor prediction tensor is used for fusing the first fusion result with the high-dimensional external factor prediction tensor again.
4. The traffic flow prediction method according to claim 3, characterized in that the fuzzy neural network includes an input layer, a fuzzy layer, and a fuzzy rule layer;
each node in the fuzzy layer represents a membership function, and the degree of membership of the input node to the fuzzy set is calculated; the formula of the membership function is as follows:
wherein u isiRepresenting a gaussian function; mu.siAnd σiRespectively representing the center and width, μ, of the Gaussian functioniAnd σiAre all learnable parameters;
representing the output of a node i in the l layers of fuzzification layers;
representing the output of a node k in the l-1 layer, wherein the node k is connected with a node i;
in the fuzzy rule layer, executing and operation, wherein the formula is as follows:
wherein the content of the first and second substances,
the output of a node j in the l-layer fuzzy layer is shown, and the node j is connected with a node m in the l + 1-layer fuzzy rule layer; n represents the number of nodes in the l-layer fuzzy layer connected with the node m in the l + 1-layer fuzzy rule layer;
representing the output of node m in the l +1 fuzzy rule layer.
5. The traffic flow prediction method according to claim 3, characterized in that the fusion layer is a fully connected layer, and the fusion formula is:
wherein the content of the first and second substances,ofand odRespectively representing the output of the fuzzy neural network and the output of the deep neural network; w is afAnd wdWeights representing an output of the fuzzy neural network and an output of the deep neural network, respectively;
represents a bias value; w is af、wdAnd
are all learnable parameters;
represents the output (o) of a l-1 layer fuzzy neural network through the fusion layerf)(l-1)And the output (o) of the l-1 layer deep neural networkd)(l-1)And performing fusion on the fusion result of the node i in the l-layer fusion layer after fusion.
6. The traffic flow prediction method according to claim 3, characterized in that the fusion module performs fusion operation in a manner based on a parameter matrix;
the first fused formula is:
wherein the content of the first and second substances,
representing a Hadamard product; xc、Xd、XwRespectively representing an adjacent prediction tensor, a daily prediction tensor and a weekly prediction tensor; wc、Wd、WwRespectively representing X for corresponding learnable parametersc、Xd、XwThe degree of influence of (c); xFConvRepresenting a first fusion result;
the formula for re-fusion is:
wherein, XExtAn extrinsic factor prediction tensor representing a high dimension;
and a traffic flow prediction tensor representing an output of the traffic flow prediction model.
7. The traffic flow prediction method according to claim 6, characterized by further comprising:
establishing a loss function, and training the traffic flow prediction model through the loss function until the value of the loss function is not reduced any more; the calculation formula of the loss function is as follows:
wherein theta represents all learnable parameters in the traffic flow prediction model; xtRepresenting a true traffic flow space-time tensor;
and randomly deleting or ignoring at least one node and connection thereof in the process of training the traffic flow prediction model.
8. A traffic flow prediction system, comprising: the system comprises a construction module, a building module, a training module and a prediction module;
the construction module is used for constructing a traffic flow space-time tensor according to the traffic data;
the building module is used for building a traffic flow prediction model based on the fuzzy convolution long-term and short-term memory network;
the training module is used for training the traffic flow prediction model based on the traffic flow space-time tensor and the traffic data to obtain a trained traffic flow prediction model;
the prediction module is used for predicting the traffic flow based on the real-time traffic flow space-time tensor, the real-time traffic data and the trained traffic flow prediction model.
9. A storage medium on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the traffic flow prediction method according to any one of claims 1 to 7.
10. A terminal, comprising: a processor and a memory;
the memory is used for storing a computer program;
the processor is configured to execute the computer program stored in the memory to cause the terminal to execute the traffic flow prediction method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010393242.7A CN111639791A (en) | 2020-05-11 | 2020-05-11 | Traffic flow prediction method, system, storage medium and terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010393242.7A CN111639791A (en) | 2020-05-11 | 2020-05-11 | Traffic flow prediction method, system, storage medium and terminal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111639791A true CN111639791A (en) | 2020-09-08 |
Family
ID=72329230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010393242.7A Pending CN111639791A (en) | 2020-05-11 | 2020-05-11 | Traffic flow prediction method, system, storage medium and terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111639791A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112099110A (en) * | 2020-09-17 | 2020-12-18 | 中国科学院海洋研究所 | Ocean internal wave forecasting method based on machine learning and remote sensing data |
| CN112489426A (en) * | 2020-11-26 | 2021-03-12 | 同济大学 | Urban traffic flow space-time prediction scheme based on graph convolution neural network |
| CN112561146A (en) * | 2020-12-08 | 2021-03-26 | 哈尔滨工程大学 | Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM |
| CN112926768A (en) * | 2021-01-27 | 2021-06-08 | 浙江工业大学 | Ground road lane-level traffic flow prediction method based on space-time attention mechanism |
| CN113077081A (en) * | 2021-03-26 | 2021-07-06 | 航天科工智能运筹与信息安全研究院(武汉)有限公司 | Traffic pollution emission prediction method |
| CN113591380A (en) * | 2021-07-28 | 2021-11-02 | 浙江大学 | Traffic flow prediction method, medium and equipment based on graph Gaussian process |
| CN114333311A (en) * | 2021-12-28 | 2022-04-12 | 南京民基悠步信息技术有限公司 | Comprehensive operation and maintenance management system based on Internet of things |
| EP4016412A1 (en) * | 2020-12-18 | 2022-06-22 | Imec VZW | Traffic prediction |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016095708A1 (en) * | 2014-12-16 | 2016-06-23 | 高德软件有限公司 | Traffic flow prediction method, and prediction model generation method and device |
| CN107230351A (en) * | 2017-07-18 | 2017-10-03 | 福州大学 | A kind of Short-time Traffic Flow Forecasting Methods based on deep learning |
| CN108346293A (en) * | 2018-04-18 | 2018-07-31 | 湖南大学 | A kind of arithmetic for real-time traffic flow Forecasting Approach for Short-term |
| CN110310479A (en) * | 2019-06-20 | 2019-10-08 | 云南大学 | A kind of Forecast of Urban Traffic Flow forecasting system and method |
| CN110633871A (en) * | 2019-09-25 | 2019-12-31 | 大连理工大学 | Regional traffic demand prediction method based on convolution long-term and short-term memory network |
-
2020
- 2020-05-11 CN CN202010393242.7A patent/CN111639791A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016095708A1 (en) * | 2014-12-16 | 2016-06-23 | 高德软件有限公司 | Traffic flow prediction method, and prediction model generation method and device |
| CN107230351A (en) * | 2017-07-18 | 2017-10-03 | 福州大学 | A kind of Short-time Traffic Flow Forecasting Methods based on deep learning |
| CN108346293A (en) * | 2018-04-18 | 2018-07-31 | 湖南大学 | A kind of arithmetic for real-time traffic flow Forecasting Approach for Short-term |
| CN110310479A (en) * | 2019-06-20 | 2019-10-08 | 云南大学 | A kind of Forecast of Urban Traffic Flow forecasting system and method |
| CN110633871A (en) * | 2019-09-25 | 2019-12-31 | 大连理工大学 | Regional traffic demand prediction method based on convolution long-term and short-term memory network |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112099110A (en) * | 2020-09-17 | 2020-12-18 | 中国科学院海洋研究所 | Ocean internal wave forecasting method based on machine learning and remote sensing data |
| CN112099110B (en) * | 2020-09-17 | 2021-06-22 | 中国科学院海洋研究所 | Ocean internal wave forecasting method based on machine learning and remote sensing data |
| CN112489426A (en) * | 2020-11-26 | 2021-03-12 | 同济大学 | Urban traffic flow space-time prediction scheme based on graph convolution neural network |
| CN112489426B (en) * | 2020-11-26 | 2021-11-09 | 同济大学 | Urban traffic flow space-time prediction scheme based on graph convolution neural network |
| CN112561146A (en) * | 2020-12-08 | 2021-03-26 | 哈尔滨工程大学 | Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM |
| CN112561146B (en) * | 2020-12-08 | 2023-04-18 | 哈尔滨工程大学 | Large-scale real-time traffic flow prediction method based on fuzzy logic and depth LSTM |
| EP4016412A1 (en) * | 2020-12-18 | 2022-06-22 | Imec VZW | Traffic prediction |
| WO2022129421A1 (en) * | 2020-12-18 | 2022-06-23 | Imec Vzw | Traffic prediction |
| CN112926768A (en) * | 2021-01-27 | 2021-06-08 | 浙江工业大学 | Ground road lane-level traffic flow prediction method based on space-time attention mechanism |
| CN113077081A (en) * | 2021-03-26 | 2021-07-06 | 航天科工智能运筹与信息安全研究院(武汉)有限公司 | Traffic pollution emission prediction method |
| CN113591380A (en) * | 2021-07-28 | 2021-11-02 | 浙江大学 | Traffic flow prediction method, medium and equipment based on graph Gaussian process |
| CN114333311A (en) * | 2021-12-28 | 2022-04-12 | 南京民基悠步信息技术有限公司 | Comprehensive operation and maintenance management system based on Internet of things |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111639791A (en) | Traffic flow prediction method, system, storage medium and terminal | |
| Liu et al. | Dynamic spatial-temporal representation learning for traffic flow prediction | |
| Luo et al. | ESTNet: embedded spatial-temporal network for modeling traffic flow dynamics | |
| CN115240425B (en) | Traffic prediction method based on multi-scale space-time fusion graph network | |
| CN111009129B (en) | Urban road traffic flow prediction method and device based on space-time deep learning model | |
| CN110570035B (en) | People flow prediction system for simultaneously modeling space-time dependency and daily flow dependency | |
| CN113673769B (en) | Traffic flow prediction method of graph neural network based on multivariate time sequence interpolation | |
| EP4016412A1 (en) | Traffic prediction | |
| CN114944053B (en) | Traffic flow prediction method based on space-time hypergraph neural network | |
| He et al. | STNN: A spatio-temporal neural network for traffic predictions | |
| CN114299723A (en) | Traffic flow prediction method | |
| CN116895157A (en) | Traffic flow prediction method, system and medium based on trend time-space diagram convolution | |
| Wu et al. | Msstn: Multi-scale spatial temporal network for air pollution prediction | |
| CN112862177A (en) | Urban area concentration degree prediction method, equipment and medium based on deep neural network | |
| CN115660135A (en) | Traffic flow prediction method and system based on Bayes method and graph convolution | |
| Tang et al. | Spatio-temporal latent graph structure learning for traffic forecasting | |
| Cao et al. | A hybrid deep learning-based traffic forecasting approach integrating adjacency filtering and frequency decomposition | |
| Zhang et al. | Boosted trajectory calibration for traffic state estimation | |
| CN117195976A (en) | Traffic flow prediction method and system based on layered attention | |
| CN112382081A (en) | Traffic flow prediction method based on multiple tasks | |
| CN112529294A (en) | Training method, medium and equipment for individual random trip destination prediction model | |
| Liu et al. | Acfm: A dynamic spatial-temporal network for traffic prediction | |
| ABBAS | A survey of research into artificial neural networks for crime prediction | |
| Chen et al. | Next location prediction with a graph convolutional network based on a seq2seq framework | |
| CN114819253A (en) | Urban crowd gathering hotspot area prediction method, system, medium and terminal |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200908 |
|
| RJ01 | Rejection of invention patent application after publication |