CN115063975B - Short-time traffic flow data prediction method, system, computer equipment and storage medium - Google Patents
Short-time traffic flow data prediction method, system, computer equipment and storage medium Download PDFInfo
- Publication number
- CN115063975B CN115063975B CN202210648480.7A CN202210648480A CN115063975B CN 115063975 B CN115063975 B CN 115063975B CN 202210648480 A CN202210648480 A CN 202210648480A CN 115063975 B CN115063975 B CN 115063975B
- Authority
- CN
- China
- Prior art keywords
- traffic flow
- flow data
- predicted
- area
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
Abstract
The invention belongs to the technical field of traffic control, and discloses a short-time traffic flow data prediction method, a short-time traffic flow data prediction system, computer equipment and a storage medium, wherein the short-time traffic flow data prediction method comprises the following steps of: acquiring historical traffic flow data of each relevant area of the area to be predicted; according to the historical traffic flow data of the area to be predicted, periodic traffic flow data of the current period of the area to be predicted is obtained, and the historical traffic flow data of each relevant area of the area to be predicted is combined, so that the optimal relevant historical traffic flow data of the area to be predicted is obtained through a preset space-time feature selection method; external characteristic influence factor data of the current period of the area to be predicted are obtained, the external characteristic influence factor data of the current period of the area to be predicted, periodic traffic flow data of the current period and optimal related historical traffic flow data are input into a preset traffic flow prediction model, short-time predicted values of the traffic flow data of the area to be predicted are obtained, and accuracy of short-time traffic flow data prediction is effectively improved.
Description
Technical Field
The invention belongs to the technical field of traffic control, and relates to a short-time traffic flow data prediction method, a short-time traffic flow data prediction system, computer equipment and a storage medium.
Background
Along with the rapid development of economy and the increasing of the living standard of people, the scale of various vehicles, especially private cars, is continuously enlarged, so that various traffic problems are brought, and traffic congestion occurs frequently. How to avoid serious traffic accidents and how to reduce the probability of occurrence of traffic jams become important content of traffic flow research. In recent years, along with the development of intelligent traffic systems, the construction of ITS is considered as an effective method for reducing traffic jam and relieving urban traffic pressure at present, and the ITS not only can provide real-time road condition information for travelers to select a proper route, effectively relieve road jam and improve traffic facility service level, but also greatly improves traffic management service level while reducing cost input of traffic management departments, and plays a small role in reducing environmental pollution and guaranteeing road traffic safety. The traffic flow prediction is used as an important branch of ITS, can provide decision basis for traffic participants, and has great application prospect in the aspects of road network planning, congestion relief, traffic control and the like.
Since the first consideration of stephanides in 1981 was to solve a series of problems in the aspect of urban intelligent traffic control systems by utilizing an HA model, related research in the traffic flow prediction direction HAs been over forty years old, and short-time traffic flow data prediction methods are becoming mature and tend to saturate. Existing traffic flow prediction methods can be broadly divided into four categories: statistical method model based on mathematical logic, traditional machine learning prediction model, prediction model based on neural network and combination model. The statistical method model based on the mathematical logic is earlier used for traffic flow prediction, the HA model belongs to one of the models, and besides, the statistical method model also comprises ARIMA, VAR and other time sequence models, and the algorithm is easier to understand, but because the technology is limited at the time, the model prediction precision is low, and the high uncertainty and randomness of traffic flow data cannot be well captured, so the model is rarely used. Traditional machine learning prediction models including KNN, SVM and the like start from training samples, balance between interpretation of the model and effectiveness of results is achieved, but multidimensional sequence data and large-scale training samples cannot be processed well. Deep learning networks may use more data to achieve higher performance than traditional machine learning algorithms. In a neural network-based short-time traffic flow data prediction model, both convolutional neural networks and recurrent neural networks occupy a single place. However, with the continuous and deep development of related researches, the defects of a single traffic flow prediction model are also exposed, and some prediction models only consider the spatial correlation among short-time traffic flow data nodes and ignore the time dependence; and some only consider the time correlation among the sequence data, neglect the space dependence among regional nodes.
To capture features in both the temporal and spatial dimensions of traffic flow sequence data, researchers began to transfer points of interest to the combined model and use them for short-term traffic flow data predictions. Although the existing research basically considers the time correlation and the space correlation of traffic flow sequence data and tries to introduce external characteristic influencing factors and attention mechanisms, the same data set can generate different prediction results and prediction precision due to different combined models, and the prediction results are closely related to the training conditions of the models. Therefore, the short-time traffic flow data prediction model needs to be further improved in the aspect of prediction precision and practical application.
Disclosure of Invention
The invention aims to overcome the defect of low short-time traffic flow data prediction accuracy in the prior art, and provides a short-time traffic flow data prediction method, a short-time traffic flow data prediction system, computer equipment and a storage medium.
In order to achieve the purpose, the invention is realized by adopting the following technical scheme:
in a first aspect of the present invention, a method for predicting short-term traffic flow data includes:
acquiring historical traffic flow data of each relevant area of the area to be predicted;
according to the historical traffic flow data of the area to be predicted, periodic traffic flow data of the current period of the area to be predicted is obtained, and the historical traffic flow data of each relevant area of the area to be predicted is combined, so that the optimal relevant historical traffic flow data of the area to be predicted is obtained through a preset space-time feature selection method;
external characteristic influence factor data of the current period of the area to be predicted are obtained, and the external characteristic influence factor data of the current period of the area to be predicted, periodic traffic flow data of the current period and optimal related historical traffic flow data are input into a preset traffic flow prediction model to obtain short-time predicted values of the traffic flow data of the area to be predicted.
Optionally, the periodic traffic data includes daily periodic traffic data, zhou Zhouqi periodic traffic data, and monthly periodic traffic data.
Optionally, the obtaining the optimal relevant historical traffic flow data of the area to be predicted by a preset space-time feature selection method includes: calculating Pearson correlation coefficients between the to-be-predicted area and historical traffic flow data of each correlation area of the to-be-predicted area, and selecting a first correlation area with a large preset number in front of the Pearson correlation coefficients as a target correlation area of the to-be-predicted area; and selecting the traffic flow data with a second preset number of historical moments before the correlation with the traffic flow data of the current time period by adopting a simulated annealing algorithm based on the target correlation area of the area to be predicted, and obtaining the optimal correlation historical traffic flow data of the area to be predicted.
Optionally, the preset traffic flow prediction model includes a first deep neural network layer, a second deep neural network layer, a first time attention mechanism network layer, a second time attention mechanism network layer, a fusion layer, a first fully connected neural network layer and a second fully connected neural network layer; the first deep neural network layer is sequentially connected with the first time attention mechanism network layer and the fusion layer, the second deep neural network layer is sequentially connected with the second time attention mechanism network layer and the fusion layer, and the fusion layer is sequentially connected with the first full-connection neural network layer and the second full-connection neural network layer; the input of the first deep neural network layer is the optimal relevant historical traffic flow data; the input of the second deep neural network layer is periodic traffic flow data of the current period of the area to be predicted; the external characteristic influence factor data of the current period of the region to be predicted is combined with the output of the fusion layer and then used as the input of the first fully-connected neural network layer; and the output of the second fully-connected neural network layer is a short-time predicted value of traffic flow data of the area to be predicted.
Optionally, the first deep neural network layer includes a feature extraction unit, the second deep neural network layer includes a plurality of feature extraction units, and the feature extraction unit includes a sequential convolutional neural network and a gating circulation unit network that are connected in turn.
Optionally, the traffic flow data is traffic flow data or traffic speed data.
Optionally, when the traffic flow data is traffic flow data, the external feature influence factor data includes date type data, weather data and traffic speed data of the current period; when the traffic flow data is traffic speed data, the external characteristic influence factor data includes date type data, weather data, and traffic flow data of the current period.
In a second aspect of the present invention, a short-term traffic flow data prediction system includes:
the data acquisition module is used for acquiring the to-be-predicted area and historical traffic flow data of each relevant area of the to-be-predicted area;
the feature extraction module is used for obtaining periodic traffic flow data of the current period of the region to be predicted according to the historical traffic flow data of the region to be predicted, and obtaining optimal relevant historical traffic flow data of the region to be predicted by combining the historical traffic flow data of each relevant region of the region to be predicted through a preset space-time feature selection method;
the prediction module is used for acquiring external feature influence factor data of the current period of the area to be predicted, inputting the external feature influence factor data of the current period of the area to be predicted, the periodic traffic flow data of the current period and the optimal correlation historical traffic flow data into a preset traffic flow prediction model, and obtaining a short-time predicted value of the traffic flow data of the area to be predicted.
In a third aspect of the present invention, a computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the short-term traffic flow data prediction method described above when executing the computer program.
In a fourth aspect of the present invention, a computer readable storage medium stores a computer program which, when executed by a processor, implements the steps of the short-time traffic flow data prediction method described above.
Compared with the prior art, the invention has the following beneficial effects:
according to the short-time traffic flow data prediction method, according to the historical traffic flow data of the to-be-predicted area, the historical traffic flow data of each relevant area of the to-be-predicted area is combined, the optimal relevant historical traffic flow data of the to-be-predicted area is obtained through a preset space-time feature selection method, and then the periodic traffic flow data of the current period of the to-be-predicted area and the external feature influence factor data of the current period of the to-be-predicted area are combined to be used as the input data of the traffic flow prediction model together to conduct short-time prediction of the traffic flow data, so that the model training cost is reduced, the model training efficiency is improved, meanwhile, the feature expression of the critical input data is enhanced, the accuracy of the short-time traffic flow data prediction is remarkably improved based on a multi-feature fusion mode, road operation efficiency can be effectively improved based on the prediction result, guarantee is provided for road traffic safety, management staff can take timely induction control measures to reduce or avoid occurrence of road traffic accidents, and the traffic jam conditions are of great significance and research value are provided for relieving and controlling the traveling efficiency.
Drawings
FIG. 1 is a flow chart of a short-term traffic flow data prediction method according to an embodiment of the present invention;
FIG. 2 is an exemplary diagram of input time series segments of periodic traffic data according to an embodiment of the present invention;
FIG. 3 is a detailed flowchart of a short-term traffic stream data prediction method according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a time-aware network architecture according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The invention is described in further detail below with reference to the attached drawing figures:
referring to fig. 1, in an embodiment of the present invention, a short-term traffic flow data prediction method is provided, which can construct a suitable deep network prediction model for short-term traffic flow prediction based on multi-feature fusion, and the result of the prediction model will improve road operation efficiency, provide a guarantee for road traffic safety, and enable a management department to take timely guidance control measures to reduce or avoid occurrence of road traffic accidents as much as possible. The method has important significance and research value for relieving and controlling traffic jam conditions and improving travel efficiency, and is a very urgent problem to be solved in traffic flow research. Specifically, the short-time traffic flow data prediction method comprises the following steps:
acquiring historical traffic flow data of each relevant area of the area to be predicted; according to the historical traffic flow data of the area to be predicted, periodic traffic flow data of the current period of the area to be predicted is obtained, and the historical traffic flow data of each relevant area of the area to be predicted is combined, so that the optimal relevant historical traffic flow data of the area to be predicted is obtained through a preset space-time feature selection method; external characteristic influence factor data of the current period of the area to be predicted are obtained, and the external characteristic influence factor data of the current period of the area to be predicted, periodic traffic flow data of the current period and optimal related historical traffic flow data are input into a preset traffic flow prediction model to obtain short-time predicted values of the traffic flow data of the area to be predicted.
In one possible embodiment, the periodic traffic data includes daily periodic traffic data, zhou Zhouqi periodic traffic data, and monthly periodic traffic data.
Specifically, referring to fig. 2, the time intervals are divided into the entry points, and the short-time prediction accuracy of the traffic flow data is improved by analyzing the dependency relationship among the daily periodicity, the weekly periodicity and the monthly periodicity of the traffic flow data.
In one possible implementation manner, the obtaining, by a preset space-time feature selection method, the optimal relevant historical traffic flow data of the area to be predicted includes: calculating Pearson correlation coefficients between the to-be-predicted area and historical traffic flow data of each correlation area of the to-be-predicted area, and selecting a first correlation area with a large preset number in front of the Pearson correlation coefficients as a target correlation area of the to-be-predicted area; and selecting the traffic flow data with a second preset number of historical moments before the correlation with the traffic flow data of the current time period by adopting a simulated annealing algorithm based on the target correlation area of the area to be predicted, and obtaining the optimal correlation historical traffic flow data of the area to be predicted.
Specifically, a space-time characteristic selection method based on a Filter formula is selected, the characteristics with low correlation are removed according to the score by calculating the correlation among the characteristics and giving out corresponding scores, and the selected optimal space-time sequence characteristics are input into a prediction model to be used as one of input data of a traffic flow prediction model.
The core objective of the spatial feature selection is to find a relevant area with strong relevance to the area to be predicted, so that the data sequence of the area to be predicted in any time period has strong relevance to the data sequence of the relevant area in the same time period. Because the short-time traffic flow data has a complex nonlinear relation, the correlation strength among all the related areas is measured by using the Pearson correlation coefficient, and the similarity matrix is formed by ranking the related areas from high to low according to the degree of correlation, so that the related areas with higher similarity with the areas to be predicted are screened out based on the similarity matrix, and the related areas with little meaning to the areas to be predicted are ignored. In order to avoid the problem that the similarity between the relevant areas may be different due to the different time periods of the input sequence, the traffic flow data at all times of each relevant area in the dataset is regarded as the characteristics of the relevant area to characterize the relevant area, and the time window is long so as to find the relevant area really similar to the area to be predicted.
The core goal of the time feature selection is to search a time period with strong correlation with a predicted time period, and search critical data which really plays a role, so that the predicted time period does not change the optimal time sequence data due to the addition of other space nodes, and the predicted time period has higher similarity with the selected optimal time sequence data period. The optimal time series data is determined on the basis of the found optimal spatial characteristics by a time characteristic selection algorithm (TFSABS, temporal Feature Selective Algorithm Based on Spatial) based on a simulated annealing algorithm, and finally, higher prediction accuracy is shown on the data set.
In one possible implementation, referring to fig. 3, the preset traffic flow prediction model includes a first deep neural network layer, a second deep neural network layer, a first time attention mechanism network layer, a second time attention mechanism network layer, a fusion layer, a first fully connected neural network layer, and a second fully connected neural network layerLaminating; the first deep neural network layer is sequentially connected with the first time attention mechanism network layer and the fusion layer, the second deep neural network layer is sequentially connected with the second time attention mechanism network layer and the fusion layer, and the fusion layer is sequentially connected with the first full-connection neural network layer and the second full-connection neural network layer; the input of the first deep neural network layer is the optimal relevant historical traffic flow data, which is expressed as X t The method comprises the steps of carrying out a first treatment on the surface of the The input of the second deep neural network layer is the periodic traffic flow data of the current period of the region to be predicted, which are respectively expressed as X m 、X w And X d The method comprises the steps of carrying out a first treatment on the surface of the The external characteristic influence factor data of the current period of the region to be predicted is combined with the output of the fusion layer and then used as the input of the first fully-connected neural network layer; the output of the second fully-connected neural network layer is a short-time predicted value of traffic flow data of the area to be predicted, which is expressed as
And comparing the traffic flow data with the actual traffic flow data Y, and optimizing the whole model by minimizing a loss function.
Optionally, the first deep neural network layer includes a feature extraction unit, the second deep neural network layer includes a plurality of feature extraction units, and the feature extraction unit includes a sequential convolutional neural network and a gating circulation unit network that are connected in turn.
Specifically, the spatial dependence relationship among the nodes is captured by using a time sequence convolution network (Temporal Convolutional Network, TCN), and the causal relationship in the spatial dimension of traffic flow data is taken into consideration by adding a causal convolution neural network. The output of the TCN network is used as the input of a gating circulation unit (Gated Recurrent Unit, GRU), the GRU network is utilized to capture the time dependence relationship among all areas, and the change trend of the historical traffic flow data is still reserved while the traffic flow data of the current period is captured. Referring to fig. 4, the different effects of different historical time periods on the predicted time period are taken into account by the time-attention-introducing mechanism network weighting each historical time period. Before the preset traffic flow prediction model is used, training and testing can be completed based on a historical data set before the preset traffic flow prediction model is used.
In one possible embodiment, the traffic flow data is traffic flow data or traffic speed data. Based on the short-time traffic flow data prediction method, the short-time prediction of traffic flow data or traffic speed data can be realized, and the short-time prediction of traffic flow data at the next time or at the following times can be realized.
In one possible embodiment, when the traffic flow data is traffic flow data, the external characteristic influence factor data includes date type data, weather data, and traffic speed data of a current period; when the traffic flow data is traffic speed data, the external characteristic influence factor data includes date type data, weather data, and traffic flow data of the current period.
Specifically, in addition to the data of common external feature influencing factors such as date type data (such as workdays and holidays), weather data and the like, traffic speed data or traffic flow data of a current period are added, namely, when the traffic flow data is predicted, the traffic speed data of the current period is taken as one of external feature influencing factors; when predicting traffic speed data, traffic flow data of the current period is taken as one of external characteristic influencing factors. Representing some external characteristic factor of the current period as a vector e ext ∈R l Where l denotes the length of certain external feature influence factor data. The method comprises the steps of processing discrete data, such as date type data, weather data and other external characteristic influence factor data, into binary vectors through a one-hot coding mode; and carrying out normalization processing on continuous data such as speed, temperature, humidity, wind speed and the like through Min-Max operation, and finally realizing the reaction of external characteristic factor data to the region to be predicted based on the two-layer fully connected neural network.
According to the short-time traffic flow data prediction method, aiming at the situation that the current traffic flow prediction model is excessive in input data but useless, the optimal relevant historical traffic flow data of the area to be predicted is selected by adding a space-time feature selection algorithm. Based on the short-time traffic flow data set adopted by the existing traffic flow prediction model, the time span is short, the consideration of the space dependence relationship between adjacent weeks can only be considered at most, the periodic traffic flow data of the current period of the area to be predicted can be obtained by dividing the period into the access points and considering the month periodicity for the first time; based on this, four temporal characteristics for short-term traffic flow data: modeling is carried out on the dependence relationship among the optimal relevant historical traffic flow data, the daily periodic traffic flow data, the Zhou Zhouqi traffic flow data and the monthly periodic traffic flow data, so that the training cost of the traffic flow prediction model is reduced, the training efficiency of the traffic flow prediction model is improved, and the characteristic expression of the key input data is enhanced. Aiming at the problem that the existing traffic flow prediction model does not consider the potential influence between the traffic flow and the traffic speed in the same time period, the influence of the traffic flow data or the traffic speed data in the current time period is considered while the conventional external characteristic factors such as date type data, weather data and the like are considered, the relation between macroscopic parameters is added in the prediction problem, the applicability and generalization of the traffic flow prediction model are improved, and the traffic flow prediction model has a certain convincing effect.
In a further embodiment of the invention, the prediction based on the short-time traffic flow data prediction method is performed by taking the data sets of two road types of expressways and urban roads as research objects.
The expressway data set adopts a public real-time traffic flow data set collected by the PeMS system, wherein the public real-time traffic flow data set comprises traffic flow data and traffic speed data, and external characteristic information such as weather data and the like is added on the basis of the data set through a web crawler technology based on a Dark Sky platform; the urban road data set adopts video data disclosed in expressway areas with intersections and without signal lamp guidance in certain two rings in certain city; the data sets of the two road types are independently trained and tested, and the short-time traffic flow data prediction method has obvious advantages and performance in the aspect of single-step prediction and multi-step prediction of short-time traffic flow data through a comparison experiment.
The following are device embodiments of the present invention that may be used to perform method embodiments of the present invention. For details not disclosed in the apparatus embodiments, please refer to the method embodiments of the present invention.
In still another embodiment of the present invention, a short-term traffic flow data prediction system is provided, which can be used to implement the short-term traffic flow data prediction method described above, and specifically, the short-term traffic flow data prediction system includes a data acquisition module, a feature extraction module, and a prediction module.
The data acquisition module is used for acquiring historical traffic flow data of the areas to be predicted and all relevant areas of the areas to be predicted; the feature extraction module is used for obtaining periodic traffic flow data of the current period of the region to be predicted according to the historical traffic flow data of the region to be predicted, and obtaining optimal relevant historical traffic flow data of the region to be predicted by combining the historical traffic flow data of each relevant region of the region to be predicted through a preset space-time feature selection method; the prediction module is used for acquiring external feature influence factor data of the current period of the area to be predicted, inputting the external feature influence factor data of the current period of the area to be predicted, the periodic traffic flow data of the current period and the optimal related historical traffic flow data into a preset traffic flow prediction model, and obtaining a short-time predicted value of the traffic flow data of the area to be predicted.
In one possible embodiment, the periodic traffic data includes daily periodic traffic data, zhou Zhouqi periodic traffic data, and monthly periodic traffic data.
In one possible implementation manner, the obtaining, by a preset space-time feature selection method, the optimal relevant historical traffic flow data of the area to be predicted includes: calculating Pearson correlation coefficients between the to-be-predicted area and historical traffic flow data of each correlation area of the to-be-predicted area, and selecting a first correlation area with a large preset number in front of the Pearson correlation coefficients as a target correlation area of the to-be-predicted area; and selecting the traffic flow data with a second preset number of historical moments before the correlation with the traffic flow data of the current time period by adopting a simulated annealing algorithm based on the target correlation area of the area to be predicted, and obtaining the optimal correlation historical traffic flow data of the area to be predicted.
In one possible implementation manner, the preset traffic flow prediction model comprises a first deep neural network layer, a second deep neural network layer, a first time attention mechanism network layer, a second time attention mechanism network layer, a fusion layer, a first fully-connected neural network layer and a second fully-connected neural network layer; the first deep neural network layer is sequentially connected with the first time attention mechanism network layer and the fusion layer, the second deep neural network layer is sequentially connected with the second time attention mechanism network layer and the fusion layer, and the fusion layer is sequentially connected with the first full-connection neural network layer and the second full-connection neural network layer; the input of the first deep neural network layer is the optimal relevant historical traffic flow data; the input of the second deep neural network layer is periodic traffic flow data of the current period of the area to be predicted; the external characteristic influence factor data of the current period of the region to be predicted is combined with the output of the fusion layer and then used as the input of the first fully-connected neural network layer; and the output of the second fully-connected neural network layer is a short-time predicted value of traffic flow data of the area to be predicted.
In a possible implementation manner, the first deep neural network layer comprises a feature extraction unit, the second deep neural network layer comprises a plurality of feature extraction units, and the feature extraction units comprise a time sequence convolution neural network and a gating circulation unit network which are connected in sequence.
In one possible embodiment, the traffic flow data is traffic flow data or traffic speed data.
In one possible embodiment, when the traffic flow data is traffic flow data, the external characteristic influence factor data includes date type data, weather data, and traffic speed data of a current period; when the traffic flow data is traffic speed data, the external characteristic influence factor data includes date type data, weather data, and traffic flow data of the current period.
All relevant contents of each step involved in the foregoing embodiment of the short-time traffic flow data prediction method may be cited to the functional description of the functional module corresponding to the short-time traffic flow data prediction system in the embodiment of the present invention, which is not described herein.
The division of the modules in the embodiments of the present invention is schematically only one logic function division, and there may be another division manner in actual implementation, and in addition, each functional module in each embodiment of the present invention may be integrated in one processor, or may exist separately and physically, or two or more modules may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules.
In yet another embodiment of the present invention, a computer device is provided that includes a processor and a memory for storing a computer program including program instructions, the processor for executing the program instructions stored by the computer storage medium. The processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processor, digital signal processor (Digital Signal Processor, DSP), application specific integrated circuit (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate array (FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc., which are the computational core and control core of the terminal adapted to implement one or more instructions, in particular to load and execute one or more instructions in a computer storage medium to implement the corresponding method flow or corresponding functions; the processor provided by the embodiment of the invention can be used for the operation of the short-time traffic flow data prediction method.
In yet another embodiment of the present invention, a storage medium, specifically a computer readable storage medium (Memory), is a Memory device in a computer device, for storing a program and data. It is understood that the computer readable storage medium herein may include both built-in storage media in a computer device and extended storage media supported by the computer device. The computer-readable storage medium provides a storage space storing an operating system of the terminal. Also stored in the memory space are one or more instructions, which may be one or more computer programs (including program code), adapted to be loaded and executed by the processor. The computer readable storage medium herein may be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as at least one magnetic disk memory. One or more instructions stored in a computer-readable storage medium may be loaded and executed by a processor to implement the corresponding steps of the method for predicting short-term traffic flow data in the above embodiments.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
Claims (8)
1. A short-term traffic stream data prediction method, comprising:
acquiring historical traffic flow data of each relevant area of the area to be predicted;
according to the historical traffic flow data of the area to be predicted, periodic traffic flow data of the current period of the area to be predicted is obtained, and the historical traffic flow data of each relevant area of the area to be predicted is combined, so that the optimal relevant historical traffic flow data of the area to be predicted is obtained through a preset space-time feature selection method;
external characteristic influence factor data of the current period of the area to be predicted are obtained, and the external characteristic influence factor data of the current period of the area to be predicted, periodic traffic flow data of the current period and optimal related historical traffic flow data are input into a preset traffic flow prediction model to obtain a short-time predicted value of the traffic flow data of the area to be predicted;
the preset traffic flow prediction model comprises a first deep neural network layer, a second deep neural network layer, a first time attention mechanism network layer, a second time attention mechanism network layer, a fusion layer, a first full-connection neural network layer and a second full-connection neural network layer; the first deep neural network layer is sequentially connected with the first time attention mechanism network layer and the fusion layer, the second deep neural network layer is sequentially connected with the second time attention mechanism network layer and the fusion layer, and the fusion layer is sequentially connected with the first full-connection neural network layer and the second full-connection neural network layer;
the input of the first deep neural network layer is the optimal relevant historical traffic flow data; the input of the second deep neural network layer is periodic traffic flow data of the current period of the area to be predicted; the external characteristic influence factor data of the current period of the region to be predicted is combined with the output of the fusion layer and then used as the input of the first fully-connected neural network layer; the output of the second fully-connected neural network layer is a short-time predicted value of traffic flow data of the area to be predicted;
the first deep neural network layer comprises a feature extraction unit, the second deep neural network layer comprises a plurality of feature extraction units, and the feature extraction units comprise a time sequence convolution neural network and a gating circulation unit network which are sequentially connected.
2. The short-term traffic stream data prediction method according to claim 1, wherein the periodic traffic stream data includes daily periodic traffic stream data, zhou Zhouqi traffic stream data, and monthly periodic traffic stream data.
3. The short-term traffic flow data prediction method according to claim 1, wherein the obtaining, by a preset space-time feature selection method, the optimal relevant historical traffic flow data of the area to be predicted includes:
calculating Pearson correlation coefficients between the to-be-predicted area and historical traffic flow data of each correlation area of the to-be-predicted area, and selecting a first correlation area with a large preset number in front of the Pearson correlation coefficients as a target correlation area of the to-be-predicted area;
and selecting the traffic flow data with a second preset number of historical moments before the correlation with the traffic flow data of the current time period by adopting a simulated annealing algorithm based on the target correlation area of the area to be predicted, and obtaining the optimal correlation historical traffic flow data of the area to be predicted.
4. The short-term traffic flow data prediction method according to claim 1, wherein the traffic flow data is traffic flow data or traffic speed data.
5. The short-time traffic flow data prediction method according to claim 4, wherein when the traffic flow data is traffic flow data, the external characteristic influence factor data includes date type data, weather data, and traffic speed data of a current period; when the traffic flow data is traffic speed data, the external characteristic influence factor data includes date type data, weather data, and traffic flow data of the current period.
6. A short-term traffic stream data prediction system, comprising:
the data acquisition module is used for acquiring the to-be-predicted area and historical traffic flow data of each relevant area of the to-be-predicted area;
the feature extraction module is used for obtaining periodic traffic flow data of the current period of the region to be predicted according to the historical traffic flow data of the region to be predicted, and obtaining optimal relevant historical traffic flow data of the region to be predicted by combining the historical traffic flow data of each relevant region of the region to be predicted through a preset space-time feature selection method;
the prediction module is used for acquiring external characteristic influence factor data of the current period of the area to be predicted, inputting the external characteristic influence factor data of the current period of the area to be predicted, the periodic traffic flow data of the current period and the optimal correlation historical traffic flow data into a preset traffic flow prediction model, and obtaining a short-time predicted value of the traffic flow data of the area to be predicted;
the preset traffic flow prediction model comprises a first deep neural network layer, a second deep neural network layer, a first time attention mechanism network layer, a second time attention mechanism network layer, a fusion layer, a first full-connection neural network layer and a second full-connection neural network layer; the first deep neural network layer is sequentially connected with the first time attention mechanism network layer and the fusion layer, the second deep neural network layer is sequentially connected with the second time attention mechanism network layer and the fusion layer, and the fusion layer is sequentially connected with the first full-connection neural network layer and the second full-connection neural network layer;
the input of the first deep neural network layer is the optimal relevant historical traffic flow data; the input of the second deep neural network layer is periodic traffic flow data of the current period of the area to be predicted; the external characteristic influence factor data of the current period of the region to be predicted is combined with the output of the fusion layer and then used as the input of the first fully-connected neural network layer; the output of the second fully-connected neural network layer is a short-time predicted value of traffic flow data of the area to be predicted;
the first deep neural network layer comprises a feature extraction unit, the second deep neural network layer comprises a plurality of feature extraction units, and the feature extraction units comprise a time sequence convolution neural network and a gating circulation unit network which are sequentially connected.
7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the short-term traffic flow data prediction method according to any one of claims 1 to 5 when the computer program is executed.
8. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the short-term traffic flow data prediction method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210648480.7A CN115063975B (en) | 2022-06-09 | 2022-06-09 | Short-time traffic flow data prediction method, system, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210648480.7A CN115063975B (en) | 2022-06-09 | 2022-06-09 | Short-time traffic flow data prediction method, system, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115063975A CN115063975A (en) | 2022-09-16 |
| CN115063975B true CN115063975B (en) | 2023-07-11 |
Family
ID=83201078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210648480.7A Active CN115063975B (en) | 2022-06-09 | 2022-06-09 | Short-time traffic flow data prediction method, system, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115063975B (en) |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4353862B2 (en) * | 2004-06-30 | 2009-10-28 | パナソニック株式会社 | Traffic flow data prediction apparatus and traffic flow data prediction method |
| US8755991B2 (en) * | 2007-01-24 | 2014-06-17 | Tomtom Global Assets B.V. | Method and structure for vehicular traffic prediction with link interactions and missing real-time data |
| BR102017019865A2 (en) * | 2017-09-15 | 2019-04-16 | Velsis Sistemas E Tecnologia Viaria S/A | PREDICTIVE, INTEGRATED AND INTELLIGENT SYSTEM FOR TRAFFIC TRAFFIC TIME CONTROL |
| CN110223510B (en) * | 2019-04-24 | 2021-03-26 | 长安大学 | Multi-factor short-term traffic flow prediction method based on neural network LSTM |
| CN113192315B (en) * | 2020-01-14 | 2022-08-23 | 香港理工大学深圳研究院 | Traffic flow distribution prediction method, prediction device and terminal equipment |
| CN111524349B (en) * | 2020-04-14 | 2021-05-25 | 长安大学 | Context feature injected multi-scale traffic flow prediction model establishing method and using method |
| CN111968375B (en) * | 2020-08-27 | 2021-08-10 | 北京嘀嘀无限科技发展有限公司 | Traffic flow prediction method and device, readable storage medium and electronic equipment |
| CN113674525B (en) * | 2021-07-30 | 2022-08-16 | 长安大学 | Signalized intersection vehicle queuing length prediction method based on sparse data |
-
2022
- 2022-06-09 CN CN202210648480.7A patent/CN115063975B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115063975A (en) | 2022-09-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Long-term traffic prediction based on lstm encoder-decoder architecture | |
| Chen et al. | Forecasting the short-term metro ridership with seasonal and trend decomposition using loess and LSTM neural networks | |
| CN109034448B (en) | Trajectory prediction method based on vehicle trajectory semantic analysis and deep belief network | |
| Jin et al. | Spatio-temporal recurrent convolutional networks for citywide short-term crowd flows prediction | |
| Zhang et al. | A novel Encoder-Decoder model based on read-first LSTM for air pollutant prediction | |
| Li et al. | Graph CNNs for urban traffic passenger flows prediction | |
| CN110555989B (en) | Xgboost algorithm-based traffic prediction method | |
| CN112863182B (en) | Cross-modal data prediction method based on transfer learning | |
| Lu et al. | Lane-level traffic speed forecasting: A novel mixed deep learning model | |
| CN115691137B (en) | Multi-modal data prediction method based on causal Markov model | |
| Lee et al. | Long short-term memory recurrent neural network for urban traffic prediction: A case study of seoul | |
| CN115148019A (en) | Early warning method and system based on holiday congestion prediction algorithm | |
| CN115496257A (en) | Short-term vehicle speed prediction based on space-time fusion | |
| Peng et al. | A forecast model of tourism demand driven by social network data | |
| Hsueh et al. | A short-term traffic speed prediction model based on LSTM networks | |
| CN115206092A (en) | Traffic prediction method of BiLSTM and LightGBM model based on attention mechanism | |
| Haputhanthri et al. | Short-term traffic forecasting using LSTM-based deep learning models | |
| Li et al. | Parking demand forecasting based on improved complete ensemble empirical mode decomposition and GRU model | |
| CN113838303B (en) | Parking lot recommendation method and device, electronic equipment and storage medium | |
| Lai et al. | Taxi demand prediction with LSTM-based combination model | |
| Liu et al. | Hourly traffic flow forecasting using a new hybrid modelling method | |
| CN115063975B (en) | Short-time traffic flow data prediction method, system, computer equipment and storage medium | |
| Sun et al. | Ada-STNet: A Dynamic AdaBoost Spatio-Temporal Network for Traffic Flow Prediction | |
| Pan et al. | Traffic speed prediction based on time classification in combination with spatial graph convolutional network | |
| CN114463978B (en) | Data monitoring method based on track traffic information processing 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |