Data-enhanced fine-grained few-sample traffic flow prediction method and system

Abstract

The invention discloses a data-enhanced fine-grained few-sample traffic flow prediction method and system. The method comprises the following steps: acquiring traffic flow data of a city; processing the traffic flow data to obtain a plurality of sub-data sets corresponding to the city, the city including a source area and a target area; determining corresponding spatio-temporal characteristic data in the plurality of sub-data sets; and carrying out model training according to the plurality of pieces of spatial-temporal characteristic data and the traffic flow data to obtain a trained transfer learning model, and carrying out traffic flow prediction on the target area according to the trained transfer learning model to obtain a traffic flow prediction result. According to the method, through refined data division, spatial-temporal feature extraction and migration prediction, fine-grained few-sample prediction of the traffic flow data is realized, and the accuracy and generalization ability of migration prediction are improved.

Description

Technical Field

[0001] The present application relates to the technical field of traffic flow prediction, and in particular to a data-enhanced fine-grained small-sample traffic flow prediction method, system, terminal and computer-readable storage medium. Background Art

[0002] The key technologies for few-sample traffic flow prediction include data enhancement technology, time series prediction technology, and transfer learning technology. Data enhancement technology usually generates new training samples by performing various transformations on a small number of existing samples, which can increase the diversity and quantity of data sets and reduce the risk of model overfitting. At present, most time series prediction technologies use deep learning such as neural networks to predict traffic flow. Transfer learning technology refers to the use of features extracted from data-rich source cities and their migration to data-scarce target cities, thereby improving the prediction performance of the model in the case of few samples. At present, adversarial multi-domain adaptive technology is used to solve the problem of network-wide traffic prediction, and a transferable traffic prediction framework is proposed: Domain Adversarial Spatiotemporal Network (DastNet). The DastNet algorithm realizes short-term traffic prediction across cities through a two-stage process.

[0003] However, common data enhancement methods such as rotation, scaling, cropping or adding noise, although they increase the diversity of samples to a certain extent, are still based on the characteristics and distribution of the original data, which may cause the model to show certain limitations when facing complex situations in the real world. Real data often contains a variety of unpredictable factors, such as emergencies, abnormal behaviors or environmental changes, which are difficult to simulate through simple data enhancement methods. In addition, in the selection of data sets, the migration of existing algorithms is based on two cities with very different geographical locations, but the differences in their urban layout, planning and design, and road network structure have a serious impact on traffic flow prediction, and effective migration cannot be achieved.

[0004] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0005] The main purpose of this application is to provide a data-enhanced fine-grained few-sample traffic flow prediction method and system, aiming to solve the problem that the existing technology of few-sample traffic flow prediction uses data enhancement to increase sample diversity, but cannot simulate complex situations such as emergencies, resulting in low prediction accuracy.

[0006] A first aspect of an embodiment of the present application provides a data-enhanced fine-grained few-sample traffic flow prediction method, which includes: acquiring traffic flow data of a city; processing the traffic flow data to obtain multiple sub-datasets corresponding to the city, wherein the city includes a source area and a target area; determining the spatiotemporal feature data corresponding to each of the multiple sub-datasets; performing model training based on the multiple spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, and performing traffic flow prediction on the target area based on the trained transfer learning model to obtain a traffic flow prediction result.

[0007] Optionally, in one embodiment of the present application, the acquiring of traffic flow data of a city specifically includes: acquiring geographic location information of the city; determining the distribution of sensor nodes in the city based on the geographic location information; and obtaining the traffic flow data of the city based on the sensor node distribution.

[0008] Optionally, in one embodiment of the present application, the traffic flow data includes current traffic flow data of the source area and historical traffic flow data of the target area; the processing of the traffic flow data to obtain multiple sub-data sets corresponding to the city specifically includes: dividing the current traffic flow data and the historical traffic flow data into regions to obtain multiple level regions; and dividing the multiple level regions into multiple sub-data sets according to the distribution of the sensor nodes.

[0009] Optionally, in one embodiment of the present application, the multiple sub-datasets include a first sub-dataset and a second sub-dataset, the target area is located in the first sub-dataset, and the spatiotemporal feature data include first spatiotemporal feature data and second spatiotemporal feature data; the determining of the spatiotemporal feature data corresponding to each of the multiple sub-datasets specifically includes: generating a first buffer zone in the first sub-dataset according to the sensor node distribution, and collecting the first spatiotemporal feature data in the first buffer zone; generating a second buffer zone in the second sub-dataset according to the sensor node distribution, and collecting the second spatiotemporal feature data in the second buffer zone.

[0010] Optionally, in one embodiment of the present application, the determining of the spatiotemporal feature data corresponding to each of the multiple sub-data sets further includes: when the distance between the source area and the target area exceeds a preset threshold range, using an inter-regional migration prediction method to perform model training; when the distance between the source area and the target area is within a preset threshold range, using an intra-regional migration prediction method to perform model training.

[0011] Optionally, in one embodiment of the present application, the trained transfer learning model is a first migration model or a second migration model; the model training is performed based on the multiple spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, specifically including: if an intra-regional migration prediction method is adopted, the first spatiotemporal feature data is processed to obtain a first spatiotemporal feature vector, and model training is performed based on the first spatiotemporal feature vector and the current traffic flow data to obtain the first migration model; if an inter-regional migration prediction method is adopted, the second spatiotemporal feature data is processed to obtain a second spatiotemporal feature vector, and model training is performed based on the second spatiotemporal feature vector and the current traffic flow data to obtain the second migration model.

[0012] Optionally, in one embodiment of the present application, the traffic flow prediction result includes traffic flow, vehicle speed and congestion situation; the traffic flow prediction is performed on the target area according to the trained transfer learning model to obtain the traffic flow prediction result, and then also includes: obtaining actual traffic flow data of the target area, and evaluating the traffic flow, vehicle speed and congestion situation of the target area according to the actual traffic flow data to obtain an evaluation result; optimizing the trained transfer learning model according to the evaluation result, and using the optimized transfer learning model as the trained transfer learning model.

[0013] The second aspect of the embodiment of the present application further provides a data-enhanced fine-grained few-sample traffic flow prediction system, wherein the data-enhanced fine-grained few-sample traffic flow prediction system includes:

[0014] A data acquisition module is used to obtain the traffic flow data of the city;

[0015] A data set partitioning module, used for processing the traffic flow data to obtain a plurality of sub-data sets corresponding to the city, wherein the city includes a source area and a target area;

[0016] A spatiotemporal knowledge extraction module, used to determine the spatiotemporal feature data corresponding to each of the plurality of sub-data sets;

[0017] The traffic flow prediction module is used to perform model training based on the multiple spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, and perform traffic flow prediction on the target area based on the trained transfer learning model to obtain a traffic flow prediction result.

[0018] The third aspect of an embodiment of the present application also provides a terminal, wherein the terminal includes: a memory, a processor, and a data-enhanced fine-grained few-sample traffic flow prediction program stored in the memory and executable on the processor, wherein the data-enhanced fine-grained few-sample traffic flow prediction program, when executed by the processor, implements the steps of the data-enhanced fine-grained few-sample traffic flow prediction method as described above.

[0019] The fourth aspect of an embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a data-enhanced fine-grained few-sample traffic flow prediction program, and when the data-enhanced fine-grained few-sample traffic flow prediction program is executed by a processor, the steps of the data-enhanced fine-grained few-sample traffic flow prediction method described above are implemented.

[0020] Beneficial effects: The present application provides a data-enhanced fine-grained few-sample traffic flow prediction method and system. The method realizes fine-grained few-sample prediction of traffic flow data through refined data partitioning, spatiotemporal feature extraction and migration prediction steps, and improves the accuracy and generalization ability of migration prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is a flow chart of a preferred embodiment of the data-enhanced fine-grained few-sample traffic flow prediction method of the present application;

[0023] Figure 2 It is a distribution map of sensor nodes in a certain region in a preferred embodiment of the data-enhanced fine-grained few-sample traffic flow prediction method of the present application;

[0024] Figure 3 It is a regional experimental research area division map in a preferred embodiment of the data-enhanced fine-grained few-sample traffic flow prediction method of the present application;

[0025] Figure 4 It is a structural diagram of a preferred embodiment of the data-enhanced fine-grained few-sample traffic flow prediction system of the present application;

[0026] Figure 5 This is a structural diagram of a preferred embodiment of the terminal of this application.

[0027] Description of reference numerals:

[0028] 100. Data acquisition module; 200. Data set partitioning module; 300. Spatiotemporal knowledge extraction module; 400. Traffic flow prediction module. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical scheme and effect of the present application clearer and more specific, the technical scheme in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. The described embodiments are only possible technical implementations of the present application, not all possible implementations. Based on the embodiments in the present application, those skilled in the art can completely combine the embodiments of the present application to obtain other embodiments without creative work, and these embodiments are also within the scope of protection of the present application.

[0030] Among related technologies, DastNet is a knowledge transfer at the city level, which may perform well on specific types of transportation networks and urban structures, but its generalization ability in different types of cities or transportation modes still needs further verification.

[0031] The following describes the data-enhanced fine-grained few-sample traffic flow prediction method and system of the embodiment of the present application with reference to the accompanying drawings. In view of the problem that the few-sample traffic flow prediction in the above-mentioned related technology uses data enhancement to increase sample diversity, but cannot simulate complex situations such as emergencies, resulting in low prediction accuracy, the present application provides a data-enhanced fine-grained few-sample traffic flow prediction method. In this method, through refined data partitioning, spatiotemporal feature extraction and migration prediction steps, fine-grained few-sample prediction of traffic flow data is achieved, which solves the problems of few training data samples and difficult simulation of real data, and improves the accuracy and generalization ability of migration prediction. As a result, the technical problem that the few-sample traffic flow prediction in the related technology uses data enhancement to increase sample diversity, but cannot simulate complex situations such as emergencies, resulting in low prediction accuracy is solved.

[0032] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0033] The data-enhanced fine-grained few-sample traffic flow prediction method described in the preferred embodiment of the present application is as follows: Figure 1 As shown, the data-enhanced fine-grained few-sample traffic flow prediction method includes the following steps:

[0034] In step S101, traffic flow data of a city is obtained.

[0035] In a possible implementation, geographical location information of the city is acquired; sensor node distribution in the city is determined based on the geographical location information; and traffic flow data of the city is obtained based on the sensor node distribution.

[0036] In the embodiment of the present application, a new solution is proposed from the two aspects of data set selection and migration prediction. A regional traffic flow data set LargeST (large-scale traffic flow prediction benchmark data set) is selected as experimental data. Figure 2 The spatial distribution of sensor nodes in a certain area in China. The dataset contains all the traffic flow data of 8,600 sensor nodes in a certain period of time. This application can solve the problems of small number of training data samples and difficulty in simulating real data. It also customizes the division of the research area according to the location of the sensor nodes, and formulates two prediction schemes: intra-regional and inter-regional prediction, combining the spatial distribution characteristics of the research area, to achieve more fine-grained traffic flow prediction.

[0037] Specifically, traffic flow data in a certain area within a certain time period is obtained from the LargeST dataset; the dataset of the entire area is divided into multiple datasets (such as 5) according to the time period to facilitate subsequent experiments and analysis; according to administrative divisions, the dataset of a certain area is divided into multiple county-level areas (such as 22), and further divided into three major areas according to the spatial distribution of sensor nodes ( Figure 3 1, 2, and 3), and obtain multiple sub-datasets (can be two or three, Figure 3 There are three sub-datasets in the figure), each of which contains a specific number of counties and sensor nodes, as well as the corresponding traffic flow data, sensor node information and adjacency matrix.

[0038] It is worth noting that the Largest dataset is selected as the training data source in the embodiment of the present application. The dataset has a long traffic flow time span and contains a large number of nodes, which can more comprehensively cover the complex dynamic change characteristics of the actual traffic flow, thereby enhancing the generalization ability of the model and improving the prediction accuracy. Specifically, LargeST is a large-scale traffic flow prediction benchmark dataset, which contains four sub-datasets, covering more than 8,600 sensor data in different areas of a certain region. These sensors are distributed on different roads and intersections and can collect traffic flow information in real time. Therefore, the dataset naturally contains a large number of nodes (i.e., sensors) and can provide detailed meta-information, such as geographic location, road network, etc. The LargeST dataset integrates traffic flow data from multiple time periods to form a dataset with a long time span, which helps to more comprehensively understand the changing trends and laws of traffic flow. Due to the large number of sensors and their wide distribution, the LargeST dataset can cover a large geographical area and a complex road network, which enables the dataset to more comprehensively reflect the distribution and changes of traffic flow. The data collected by each sensor contains rich traffic flow information, such as vehicle speed, vehicle flow, congestion, etc.

[0039] In step S102, the traffic flow data is processed to obtain a plurality of sub-datasets corresponding to the city, wherein the city includes a source area and a target area.

[0040] In a possible implementation, the traffic flow data includes current traffic flow data of the source area and historical traffic flow data of the target area. The current traffic flow data and the historical traffic flow data are divided into regions to obtain multiple level regions; and the multiple level regions are divided into multiple sub-data sets according to the distribution of the sensor nodes.

[0041] It can be understood that in this preprocessing process, the original traffic flow data (from multiple sensor nodes and containing rich spatiotemporal information) are cleaned and sorted to remove outliers and missing values, and the data set is divided into multiple sub-datasets according to the administrative divisions or the spatial distribution of sensor nodes, each of which contains a specific number of counties and sensor nodes.

[0042] Furthermore, traffic flow data in a certain area within a certain time period is obtained, and the entire data set of a certain area is divided into 5 data sets according to time, and one of the data sets is extracted for the experiment. According to the administrative division, the data set of a certain area is divided into 22 county-level areas, and further according to the spatial distribution of sensor nodes, such as Figure 3It is divided into three large areas and three sub-datasets are obtained. The specific information is shown in Table 1. Area 1 includes 10 counties and 2358 nodes; Area 2 includes 6 counties and 796 nodes; Area 3 includes 6 counties and 4550 nodes. Each county contains three files: traffic flow data, sensor node information and its corresponding adjacency matrix.

[0043] Table 1: Sub-dataset information

[0044] Serial number node Time Range interval scale 1 2358 2017 / 1 / 1-2021 / 12 / 31 5min 525888 2 796 2017 / 1 / 1-2021 / 12 / 31 5min 525888 3 4550 2017 / 1 / 1-2021 / 12 / 31 5min 525888

[0045] In step S103, the spatiotemporal feature data corresponding to each of the plurality of sub-data sets are determined.

[0046] In a possible implementation, the plurality of sub-datasets include a first sub-dataset and a second sub-dataset, the target area is located in the first sub-dataset, and the spatiotemporal feature data include first spatiotemporal feature data and second spatiotemporal feature data. According to the sensor node distribution, a first buffer zone is generated in the first sub-dataset, and the first spatiotemporal feature data in the first buffer zone is collected; according to the sensor node distribution, a second buffer zone is generated in the second sub-dataset, and the second spatiotemporal feature data in the second buffer zone is collected.

[0047] Specifically, by introducing spatiotemporal knowledge, unique spatiotemporal characteristics are determined for each sub-dataset (or node). This includes using ArcGIS software (Arc Geographic Information System) to generate circular buffers, define the independent area range of each node, and collect spatiotemporal characteristic data related to the node within this range, such as traffic flow, vehicle speed, and congestion. These characteristic data reflect the fine-grained spatiotemporal differences within the region and provide key information for subsequent migration prediction.

[0048] Further, in the process of determining the regional scope, this application proposes a refined regional division method when introducing spatiotemporal knowledge to improve the accuracy and generalization ability of migration prediction. The specific operation steps are as follows: First, with the help of ArcGIS software, a circular buffer with a radius of 2.5 kilometers is generated with the location of the sensor node as the center, so as to define the independent regional scope of each node, and collect spatiotemporal feature data related to the node within each regional scope. Since the buffer of each node only covers its unique spatial area, the characteristics of each node are unique, which fully reflects the fine-grained spatiotemporal differences in the region. Through this method, each node can obtain unique regional characteristics, thereby providing support for more accurate migration prediction, especially for dynamic and complex traffic flow scenarios. It can be understood that the buffer area range of this embodiment is constructed based on a circular area with a radius of 2.5 kilometers. According to the needs of the specific prediction task, if the information of interest has different feature distributions or spatial coverage requirements, the range can be appropriately adjusted according to the characteristics of the target prediction. This flexible range setting method can ensure that the regional division is more consistent with the target characteristics of the prediction task, thereby improving the applicability and prediction accuracy of the model in specific task scenarios.

[0049] It is worth noting that this application customizes the regional range according to the geographical distribution of sensor nodes to achieve more fine-grained migration prediction. Through the refined division of node areas, the method can more effectively capture the unique characteristics of different regions. At the same time, in order to verify the effectiveness of this method, comparative experimental schemes between regions and within regions are designed to analyze the impact of spatial distances between different regions on migration prediction performance. Through the above improvements, this application can improve the accuracy and applicability of traffic flow prediction in a small sample environment, and is suitable for complex and dynamic urban traffic flow scenarios.

[0050] In step S104, model training is performed based on the plurality of spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, and traffic flow prediction is performed on the target area based on the trained transfer learning model to obtain a traffic flow prediction result.

[0051] It should be noted that transfer learning is a machine learning method that uses the knowledge learned in one task to help improve the learning effect of another task. In traffic flow prediction, transfer learning can use the traffic flow data of the source area to predict the traffic flow in the target area.

[0052] In a possible implementation, when the distance between the source area and the target area exceeds a preset threshold range, the inter-regional migration prediction method is used for model training; when the distance between the source area and the target area is within a preset threshold range, the intra-regional migration prediction method is used for model training.

[0053] In a possible implementation, the trained transfer learning model is a first transfer model or a second transfer model. If the intra-regional transfer prediction method is adopted, the first spatiotemporal feature data is processed to obtain a first spatiotemporal feature vector, and a model is trained based on the first spatiotemporal feature vector and the current traffic flow data to obtain the first transfer model; if the inter-regional transfer prediction method is adopted, the second spatiotemporal feature data is processed to obtain a second spatiotemporal feature vector, and a model is trained based on the second spatiotemporal feature vector and the current traffic flow data to obtain the second transfer model.

[0054] Specifically, intra-regional migration prediction: within the same sub-dataset (or study area), a specific county is selected as the target area and the remaining counties are selected as the source area. The spatiotemporal feature data and traffic flow data of the source area are used to train the transfer learning model, and the trained model (first migration model) is applied to the target area to perform traffic flow prediction.

[0055] Inter-regional migration prediction: Across different sub-datasets (or study areas), select the target area and the source area, extract the spatiotemporal feature data and traffic flow data of the source area, and perform appropriate preprocessing and conversion. Use the transfer learning model (the second transfer model) to transfer the knowledge of the source area to the target area for traffic flow prediction.

[0056] Furthermore, in the process of migration prediction, this application deeply analyzes the impact of neighboring areas and distant areas on the target area, and designs two sets of experiments for inter-regional migration prediction and intra-regional migration prediction to evaluate the impact of areas at different spatial distances on the prediction effect of the target area. In the intra-regional migration prediction method, according to the above-divided data set, Region 1 selects County A as the target area, Region 2 selects County B as the target area, Region 3 selects County C as the target area, and the remaining counties are source areas to conduct intra-regional prediction experiments; in the inter-regional migration prediction method, according to the above-divided data set, Region 1 predicts County B in Region 2, Region 2 predicts County C in Region 3, and Region 3 predicts County A in Region 1.

[0057] According to the divided data set, Region 1 selects County A as the target area, Region 2 selects County B as the target area, Region 3 selects County C as the target area, and the remaining counties are source areas for intra-regional prediction experiments; Region 1 predicts B in Region 2, and so on for inter-regional predictions.

[0058] Table 2: MAE values ​​(mean absolute error) for inter-regional predictions

[0059]

[0060] As shown in Table 2, for each target domain, the experimental results between regions and within regions are not much different, and both can be well predicted. For County A and County C, the prediction results within the region are better than the prediction results between regions, and the prediction results between regions in County B are better than the prediction results within the region. Observing the node information, the total number of nodes in Region 1 is 2358, and the number of nodes in Region 2 is 796.

[0061] Therefore, the present application judges whether the distance between the source area and the target area exceeds a preset threshold range, thereby determining whether to perform an intra-region migration prediction method or an inter-region migration prediction method, thereby improving the prediction accuracy.

[0062] In a possible implementation, the traffic flow prediction result includes traffic flow, vehicle speed, and congestion. After obtaining the traffic flow prediction result, the actual traffic flow data of the target area is obtained, and the traffic flow, vehicle speed, and congestion of the target area are evaluated according to the actual traffic flow data to obtain an evaluation result; the trained transfer learning model is optimized according to the evaluation result, and the optimized transfer learning model is used as the trained transfer learning model.

[0063] Specifically, based on the output of the transfer learning model, the traffic flow prediction results of the target area are generated. These results include traffic flow, vehicle speed, congestion, etc. in the future period; the prediction results are evaluated using appropriate evaluation indicators (such as accuracy, recall rate, F1 score, etc.), and the prediction performance and accuracy of the model are analyzed by comparing the prediction results with the actual traffic flow data; based on the evaluation results, the transfer learning model is optimized and adjusted, and different model structures, parameter settings and feature selection methods are tried to improve the prediction performance and accuracy.

[0064] After determining the spatiotemporal feature data of each sub-dataset, this application uses these feature data to predict the traffic flow prediction results of the target area through the transfer learning method. This method makes full use of the traffic flow data and spatiotemporal feature information of the source area, and improves the accuracy and reliability of the traffic flow prediction in the target area.

[0065] Next, the data-enhanced fine-grained few-sample traffic flow prediction system proposed in accordance with the embodiments of the present application will be described with reference to the accompanying drawings.

[0066] Figure 4 It is a structural diagram of the data-enhanced fine-grained few-sample traffic flow prediction system of an embodiment of the present application.

[0067] like Figure 4 As shown, the data-enhanced fine-grained few-sample traffic flow prediction system includes: a data acquisition module 100, a data set partitioning module 200, a spatiotemporal knowledge extraction module 300 and a traffic flow prediction module 400.

[0068] Specifically, the data acquisition module 100 is used to acquire the traffic flow data of the city;

[0069] A data set partitioning module 200 is used to process the traffic flow data to obtain a plurality of sub-data sets corresponding to the city, wherein the city includes a source area and a target area;

[0070] A spatiotemporal knowledge extraction module 300, used to determine the spatiotemporal feature data corresponding to each of the plurality of sub-data sets;

[0071] The traffic flow prediction module 400 is used to perform model training based on the multiple spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, and perform traffic flow prediction on the target area based on the trained transfer learning model to obtain a traffic flow prediction result.

[0072] Figure 5 This is a structural diagram of a terminal provided in an embodiment of the present application. The terminal may include:

[0073] A memory 501 , a processor 502 , and a computer program stored in the memory 501 and executable on the processor 502 .

[0074] When the processor 502 executes the program, the data-enhanced fine-grained few-sample traffic flow prediction method provided in the above embodiment is implemented.

[0075] Furthermore, the terminal further includes:

[0076] The communication interface 503 is used for communication between the memory 501 and the processor 502 .

[0077] The memory 501 is used to store computer programs that can be executed on the processor 502 .

[0078] The memory 501 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0079] If the memory 501, the processor 502 and the communication interface 503 are implemented independently, the communication interface 503, the memory 501 and the processor 502 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EIS) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0080] Optionally, in a specific implementation, if the memory 501, the processor 502 and the communication interface 503 are integrated on a chip, the memory 501, the processor 502 and the communication interface 503 can communicate with each other through an internal interface.

[0081] The processor 502 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0082] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above data-enhanced fine-grained few-sample traffic flow prediction method is implemented.

[0083] One embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the present application Figure 1 The data-enhanced fine-grained few-sample traffic flow prediction method provided by any embodiment in the corresponding embodiment.

[0084] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0085] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0086] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0087] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, a "computer-readable storage medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable storage media include the following: an electrical connection with one or N wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable storage medium may even be paper or other suitable medium on which the program is printed, since the program can be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary, and then storing it in a computer memory.

[0088] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0089] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0090] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0091] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

[0092] It should be understood that the application of the present application is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to this application.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A data-enhanced fine-grained few-sample traffic flow prediction method, characterized in that: The data-enhanced fine-grained few-sample traffic flow prediction method includes: Get traffic flow data for the city; Processing the traffic flow data to obtain a plurality of sub-datasets corresponding to the city, wherein the city includes a source area and a target area; Determine the spatiotemporal feature data corresponding to each of the plurality of sub-data sets; Model training is performed based on the plurality of spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, and traffic flow prediction is performed on the target area based on the trained transfer learning model to obtain a traffic flow prediction result.

2. The data-enhanced fine-grained few-sample traffic flow prediction method according to claim 1, characterized in that: The obtaining of the traffic flow data of the city specifically includes: Obtaining geographic location information of the city; Determining the distribution of sensor nodes in the city according to the geographical location information; The traffic flow data of the city is obtained according to the distribution of the sensor nodes.

3. The data-enhanced fine-grained few-sample traffic flow prediction method according to claim 2 is characterized in that: The traffic flow data includes current traffic flow data of the source area and historical traffic flow data of the target area; The processing of the traffic flow data to obtain a plurality of sub-data sets corresponding to the city specifically includes: Dividing the current traffic flow data and the historical traffic flow data into regions to obtain multiple levels of regions; According to the distribution of the sensor nodes, a plurality of the level areas are divided into a plurality of sub-data sets.

4. The data-enhanced fine-grained few-sample traffic flow prediction method according to claim 2, characterized in that: The plurality of sub-datasets include a first sub-dataset and a second sub-dataset, the target area is located in the first sub-dataset, and the spatiotemporal feature data include first spatiotemporal feature data and second spatiotemporal feature data; The determining of the spatiotemporal feature data corresponding to each of the plurality of sub-data sets specifically includes: According to the sensor node distribution, a first buffer zone is generated in the first sub-data set, and first spatiotemporal feature data in the first buffer zone is collected; A second buffer is generated in the second sub-data set according to the sensor node distribution, and second spatiotemporal feature data in the second buffer is collected.

5. The data-enhanced fine-grained few-sample traffic flow prediction method according to claim 4, characterized in that: The step of determining the spatiotemporal feature data corresponding to each of the plurality of sub-data sets further comprises: When the distance between the source area and the target area exceeds a preset threshold range, the model training is performed using an inter-region migration prediction method; When the distance between the source area and the target area is within a preset threshold range, the model training is performed using an intra-region migration prediction method.

6. The data-enhanced fine-grained few-sample traffic flow prediction method according to claim 5, characterized in that: The trained transfer learning model is the first transfer model or the second transfer model; The model training is performed according to the plurality of spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, specifically comprising: If the intra-regional migration prediction method is adopted, the first spatiotemporal feature data is processed to obtain a first spatiotemporal feature vector, and model training is performed based on the first spatiotemporal feature vector and the current traffic flow data to obtain the first migration model; If the inter-regional migration prediction method is adopted, the second spatiotemporal feature data is processed to obtain a second spatiotemporal feature vector, and model training is performed based on the second spatiotemporal feature vector and the current traffic flow data to obtain the second migration model.

7. The data-enhanced fine-grained few-sample traffic flow prediction method according to claim 1, characterized in that: The traffic flow prediction results include traffic flow, vehicle speed and congestion conditions; The method further includes: performing traffic flow prediction on the target area according to the trained transfer learning model to obtain a traffic flow prediction result; and then: Acquire actual traffic flow data of the target area, According to the actual traffic flow data, the traffic flow, the vehicle speed and the congestion situation in the target area are evaluated to obtain an evaluation result; The trained transfer learning model is optimized according to the evaluation result, and the optimized transfer learning model is used as the trained transfer learning model.

8. A data-enhanced fine-grained few-sample traffic flow prediction system, characterized in that: The data-enhanced fine-grained few-sample traffic flow prediction system includes: A data acquisition module, used to acquire the traffic flow data of the city; A data set partitioning module, used for processing the traffic flow data to obtain a plurality of sub-data sets corresponding to the city, wherein the city includes a source area and a target area; A spatiotemporal knowledge extraction module, used to determine the spatiotemporal feature data corresponding to each of the plurality of sub-data sets; The traffic flow prediction module is used to perform model training based on the multiple spatiotemporal feature data and the traffic flow data to obtain a trained transfer learning model, and perform traffic flow prediction on the target area based on the trained transfer learning model to obtain a traffic flow prediction result.

9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a data-enhanced fine-grained few-sample traffic flow prediction program stored in the memory and executable on the processor. When the data-enhanced fine-grained few-sample traffic flow prediction program is executed by the processor, the steps of the data-enhanced fine-grained few-sample traffic flow prediction method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a data-enhanced fine-grained few-sample traffic flow prediction program, which, when executed by a processor, implements the steps of the data-enhanced fine-grained few-sample traffic flow prediction method as described in any one of claims 1-7.

Images