Traffic flow high-resolution prediction method, device and equipment based on multi-model fusion

Abstract

The application discloses a traffic flow high-resolution prediction method and device based on multi-model fusion and equipment, and belongs to the technical field of traffic prediction. The application obtains real-time traffic data and historical traffic data of a target road section, constructs a multi-time scale feature set based on the real-time traffic data and the historical traffic data, inputs the multi-time scale feature set into a SVM model and a speed-flow model designed based on the multi-time scale feature respectively, predicts the flow of the target road section at a to-be-predicted time, obtains first predicted flow data predicted by the SVM model and second predicted flow data predicted by the speed-flow model, and adaptively fuses the first predicted flow data and the second predicted flow data to obtain target predicted flow of the target road section at the to-be-predicted time. The application fully combines the data driving advantage of the SVM model and the physical law advantage of the speed-flow model, and realizes high-resolution and high-precision prediction of traffic flow.

Description

Technical Field

[0001] This invention relates to the field of traffic prediction technology, and specifically to a method, apparatus, and device for high-resolution traffic flow prediction based on multi-model fusion. Background Technology

[0002] Traffic flow prediction, as one of the core technologies of modern Intelligent Transportation Systems (ITS), plays a crucial role in alleviating traffic congestion, improving road network efficiency, optimizing traffic management and control strategies, and reducing traffic accidents, energy consumption, and carbon emissions. With accelerating urbanization and continuously growing travel demand, the need for high-precision, real-time traffic flow prediction is becoming increasingly urgent.

[0003] Thanks to the rapid development of the Internet of Things (IoT) and intelligent sensing technologies, traffic data collection has become unprecedentedly abundant and convenient. Vehicle-mounted sensors (such as GPS and OBD), roadside sensing devices (such as radar, geomagnetic coils, and lidar), and a network of traffic cameras throughout the city can capture core traffic flow parameters such as vehicle speed, flow rate, density, and occupancy in real time and from all angles, providing a solid data foundation for predictive models.

[0004] Traditional traffic flow forecasting methods, such as historical averages, time series models (ARIMA), Kalman filtering, and statistical regression models, have played an important role over the past few decades. However, these methods often exhibit significant limitations when dealing with the inherent high nonlinearity, strong spatiotemporal dependence, complex dynamic evolution characteristics, and sudden disturbances of traffic systems. Their forecasting accuracy and robustness are insufficient to meet the needs of refined management and control.

[0005] In recent years, breakthroughs in artificial intelligence technologies, particularly deep learning (such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs) and their variants LSTM / GRU, and Graph Neural Networks (GNNs)) and reinforcement learning, have brought revolutionary progress to traffic flow prediction. These methods can automatically learn complex feature representations and dynamic patterns from massive, high-dimensional spatiotemporal data, demonstrating powerful capabilities in handling nonlinear relationships and large-scale data, and significantly improving prediction accuracy.

[0006] However, current mainstream deep learning methods still face key challenges in the deep fusion of multi-source heterogeneous data, especially in the failure to specifically and deeply fuse historical time data at different time scales that affect traffic flow. Summary of the Invention

[0007] To address the aforementioned shortcomings in existing technologies, this invention provides a method, apparatus, and device for high-resolution traffic flow prediction based on multi-model fusion. By fully combining the data-driven advantages of the SVM model and the physical law advantages of the speed-flow model, high-resolution and high-precision traffic flow prediction is achieved.

[0008] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0009] In a first aspect, the present invention provides a high-resolution traffic flow prediction method based on the fusion of an SVM model and a speed-flow model, comprising:

[0010] Acquire real-time and historical traffic data for the target road segment, wherein the traffic data includes at least the speed of at least one vehicle in the target road segment and the corresponding time.

[0011] Based on the real-time traffic data and the historical traffic data, a multi-timescale feature set is constructed. The multi-timescale feature set consists of at least some features from multiple timescale traffic feature datasets, and the number of traffic feature data at each timescale in the multi-timescale feature set is related to the target predicted traffic flow at the time to be predicted. Each timescale traffic feature dataset corresponds to one timescale. The traffic feature data in the timescale traffic feature dataset is calculated based on the real-time traffic data and the historical traffic data.

[0012] The multi-time-scale feature set is respectively input into the SVM model and the speed-flow model designed based on the multi-time-scale features to predict the traffic flow of the target road segment at the time to be predicted, so as to obtain the first predicted traffic flow data predicted by the SVM model and the second predicted traffic flow data predicted by the speed-flow model.

[0013] The first predicted traffic data and the second predicted traffic data are adaptively fused to obtain the target predicted traffic flow of the target road segment at the time to be predicted.

[0014] In some embodiments, it also includes:

[0015] Based on the real-time traffic data and the historical traffic data, obtain traffic feature datasets at multiple time scales corresponding to the time to be predicted;

[0016] For each time-scale traffic feature dataset, obtain the discrete mutual information between each traffic feature data in the time-scale traffic feature dataset and the target predicted traffic flow;

[0017] Based on the discrete mutual information, the importance score of the traffic feature data is determined;

[0018] Based on the number of assigned features corresponding to the traffic feature dataset at the time scale where the traffic feature data is located and the importance score of the traffic feature data, the traffic feature data used to form the multi-time scale feature set is determined.

[0019] In some embodiments, it also includes:

[0020] For each of the time-scale traffic feature datasets, determine the average mutual information corresponding to the time-scale traffic feature dataset;

[0021] The number of assigned features corresponding to each of the time-scale traffic feature datasets is determined based on the average mutual information and the total number of features of the multiple time-scale traffic feature datasets.

[0022] In some embodiments, the SVM model designed based on multi-timescale features adopts an RBF kernel function based on scale-corresponding weights, wherein the scale-corresponding weights are determined by the average mutual information of the time-scale traffic feature dataset.

[0023] In some embodiments, a velocity-flow model designed based on multi-timescale features is used to predict flow rate, resulting in the second predicted flow rate data, including:

[0024] The predicted probability of each traffic state is calculated based on the multi-timescale feature set.

[0025] For each traffic state, the predicted state traffic flow is calculated using the multi-scale regression sub-model corresponding to that state.

[0026] The second predicted flow data is obtained by weighting all the predicted flow values ​​according to their posterior probabilities.

[0027] In some embodiments, state flow prediction is performed based on scale-corresponding weights.

[0028] Secondly, the present invention provides a high-resolution traffic flow prediction device based on the fusion of an SVM model and a speed-flow model, comprising:

[0029] The acquisition module is used to acquire real-time traffic data and historical traffic data of the target road segment. The traffic data includes at least the speed of at least one vehicle in the target road segment and the corresponding time.

[0030] A construction module is used to construct a multi-timescale feature set based on the real-time traffic data and the historical traffic data. The multi-timescale feature set consists of at least some features from multiple timescale traffic feature datasets, and the number of traffic feature data at each time scale in the multi-timescale feature set is related to the target predicted traffic flow at the time to be predicted. Each timescale traffic feature dataset corresponds to one time scale. The traffic feature data in the timescale traffic feature dataset is calculated based on the real-time traffic data and the historical traffic data.

[0031] The prediction module is used to input the multi-time-scale feature set into the SVM model and the speed-flow model designed based on the multi-time-scale features respectively to predict the traffic flow of the target road segment at the time to be predicted, and to obtain the first predicted traffic flow data predicted by the SVM model and the second predicted traffic flow data predicted by the speed-flow model.

[0032] The fusion module is used to adaptively fuse the first predicted traffic data and the second predicted traffic data to obtain the target predicted traffic of the target road segment at the time to be predicted.

[0033] Thirdly, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method as described in the present invention.

[0034] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method as described in the present invention.

[0035] Fifthly, the present invention provides a computer program product, including a computer program, characterized in that the computer program, when executed by a processor, implements the method as described in the present invention.

[0036] This invention proposes a high-resolution traffic flow prediction method, apparatus, and related equipment based on the fusion of SVM and velocity-flow models. By acquiring real-time and historical traffic data of the target road segment, a multi-time-scale feature set is constructed based on this data. This feature set is then input into both the SVM and velocity-flow models designed based on these features. Traffic flow prediction is performed on the target road segment at the predicted time, resulting in first predicted traffic flow data from the SVM model and second predicted traffic flow data from the velocity-flow model. The first and second predicted traffic flow data are then adaptively fused to obtain the target predicted traffic flow for the target road segment at the predicted time. This method fully combines the data-driven advantages of the SVM model with the physical law advantages of the velocity-flow model, achieving high-resolution and high-precision traffic flow prediction.

[0037] The present invention has the following beneficial effects:

[0038] This application divides real-time and historical traffic data into multiple time-scale traffic feature datasets, enabling multi-scale classification analysis of traffic data features to obtain expressive information reflecting different types of target road segments. Simultaneously, by selecting the final multi-time-scale feature set for traffic flow prediction through discrete mutual information and importance scoring, it effectively selects reasonable traffic feature data for target traffic flow prediction, significantly reducing the amount of redundant feature data, effectively avoiding the influence of redundant data on the final prediction data, and improving the accuracy of target traffic flow prediction.

[0039] This application uses an SVM model designed based on multi-timescale features to enhance the influence of information-rich scales on the predicted traffic flow during analysis and prediction, while suppressing less information-rich scales. Moreover, the weights can be updated periodically, so that even if the interval between the feature and the time to be predicted is large during the long-term prediction process, such as the ultra-macro scale, the feature still has a high scale weight when the target road segment has strong regularity, thereby improving the influence of the corresponding scale feature data on the prediction results.

[0040] This application utilizes a speed-flow model designed based on multi-timescale features for traffic flow prediction. This improved model automatically identifies traffic state patterns while maintaining the fundamental laws governing the speed-flow relationship, ensuring high physical plausibility in the prediction results. Furthermore, by combining multi-timescale features and corresponding weights, it emphasizes feature data at different time scales under different traffic conditions and uses probability weighting to reduce boundary effects in the prediction results, thereby improving the accuracy and adaptability of traffic flow prediction using the speed-flow model. Attached Figure Description

[0041] Figure 1 The diagram illustrates the implementation environment architecture of the high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model provided in this application embodiment.

[0042] Figure 2 The diagram illustrates a flowchart of a high-resolution traffic flow prediction method based on the fusion of an SVM model and a speed-flow model, according to an embodiment of this application.

[0043] Figure 3 A schematic diagram of the structure of a high-resolution traffic flow prediction device based on the fusion of an SVM model and a speed-flow model provided in an embodiment of this application is shown.

[0044] Figure 4 A schematic diagram of the structure of a computer system suitable for implementing an electronic device or server according to embodiments of this application is shown. Detailed Implementation

[0045] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0046] For the specific implementation environment of the high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model proposed in this application, please refer to [link / reference needed]. Figure 1 . Figure 1 The diagram illustrates the implementation environment architecture of the high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model provided in this application embodiment.

[0047] like Figure 1 As shown, the implementation environment architecture includes: data acquisition device 101 and server 102.

[0048] The data acquisition device 101 is used to acquire key traffic data such as the speed of at least one vehicle within the road segment. The data acquisition device 101 may be an on-board terminal installed on a vehicle that can acquire vehicle driving data, a mobile device with speed acquisition function carried by the driver and passengers in the vehicle, or a speed acquisition device set up on the target road segment, but is not limited to these.

[0049] Server 102 communicates with data acquisition device 101 to obtain traffic data of the target road segment, and executes the high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model provided in this application embodiment according to the traffic data, so as to predict the traffic flow of the target road segment and the time to be predicted, and obtain the target predicted traffic flow.

[0050] Server 102 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0051] The data acquisition device 101 and the server 102 are connected directly or indirectly via wired or wireless communication. Optionally, the aforementioned wireless or wired network uses standard communication technologies and / or protocols. The network is typically the Internet, but can also be any network, including but not limited to any combination of Local Area Network (LAN), Metropolitan Area Network (MAN), Wide Area Network (WAN), mobile, wired or wireless network, private network, or virtual private network.

[0052] The high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model proposed in this application can be implemented by a high-resolution traffic flow prediction device based on the fusion of SVM model and speed-flow model, which can be installed on a terminal device or server.

[0053] To further illustrate the technical solutions provided in the embodiments of this application, a detailed description is provided below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiments of this application provide method operation instruction steps as shown in the following embodiments or drawings, the method may include more or fewer operation instruction steps based on conventional or non-creative effort. In steps where there is no logically necessary causal relationship, the execution order of these steps is not limited to the execution order provided in the embodiments of this application. In actual processing or when the device executes the method, it may be executed sequentially or in parallel according to the method shown in the embodiments or drawings.

[0054] It should be noted that the acquisition or use of traffic data in this application embodiment requires the user's consent. Relevant traffic data can only be obtained after the user's authorization, and the acquisition or use of traffic data complies with the provisions of relevant laws and regulations.

[0055] Please refer to Figure 2 , Figure 2 This diagram illustrates a flowchart of a high-resolution traffic flow prediction method based on the fusion of an SVM model and a speed-flow model, according to an embodiment of this application. Figure 2 As shown, the method includes:

[0056] Step 201: Obtain real-time traffic data and historical traffic data for the target road segment. The traffic data shall include at least the speed of at least one vehicle in the target road segment and the corresponding time.

[0057] It should be noted that the target road segment is the road segment for which traffic flow prediction is to be performed. Real-time traffic data is data characterizing the traffic conditions within the target road segment, including but not limited to congestion, vehicle speed, and time. Historical traffic data is real-time traffic data received and stored prior to the target road segment. Real-time traffic data for the target road segment can be obtained through vehicle-mounted terminals communicating with the server, mobile devices with speed acquisition capabilities carried by drivers and passengers, speed acquisition devices installed on the target road segment, etc., and this application does not impose specific limitations on these methods.

[0058] Step 202: Construct a multi-timescale feature set based on real-time traffic data and historical traffic data; the multi-timescale feature set consists of at least some features from multiple timescale traffic feature datasets, and the number of traffic feature data at each time scale in the multi-timescale feature set is related to the target predicted traffic flow at the time to be predicted, and each timescale traffic feature dataset corresponds to one time scale; the traffic feature data in the timescale traffic feature dataset is calculated based on real-time traffic data and historical traffic data.

[0059] It should be noted that various time scales can be set according to the required prediction accuracy. For example, multiple time scales can include microscale, mesoscale, macroscale, and ultra-macroscale. Among them, the microscale can be 5-30 minutes, which can reflect the instantaneous fluctuations of traffic flow changes on the target road segment and the dispersion effect of the vehicle fleet. The mesoscale can be 1-3 hours, which can reflect the short-term trend of traffic flow on the target road segment and the impact of events. The macroscale can be 6-12 hours, which can reflect the periodic patterns and daily cycle patterns of traffic flow changes on the target road segment. The ultra-macroscale can be 12-24 hours, which can reflect the seasonal patterns and long-term trends of traffic flow changes on the target road segment.

[0060] Furthermore, the micro-scale traffic feature dataset is constructed as follows:

[0061]

[0062] in, This is a dataset of traffic features at a microscale. Let i be the flow rate at the i-th time step within the microscopic scale. Let be the velocity at the i-th time step within the microscopic scale. For the first-order difference of the flow, , For the second difference of the flow, .

[0063] The mesoscale traffic feature dataset was constructed as follows:

[0064]

[0065] in, This is a dataset of traffic features at the mesoscale. Let be the average flow over the past k time steps. Let the standard deviation of the flow over the past k time steps be denoted as . Let be the slope of the traffic trend over the past k time steps. Flow variation coefficient k is the size of the time window.

[0066] The macro-scale traffic feature dataset is constructed as follows:

[0067]

[0068] in, This is a macro-scale traffic feature dataset. This represents the mean flow rate corresponding to the minimum value on a macroscopic scale. Taking a macroscopic scale of 6-12 hours as an example, min = 6 hours = 72 time steps. , This represents the average flow rate corresponding to the maximum value on a macroscopic scale, where max = 12 hours = 144 time steps. , For traffic anomaly, , Daily year-on-year change rate .

[0069] The ultra-macroscale traffic feature dataset is constructed as follows:

[0070]

[0071] in, This is a dataset of traffic features at a very macroscopic scale. This is the historical average traffic volume for the same period. Historical traffic trends for the same period Period ratio, , Periodic deviation, 288 represents the number of steps taken in a day.

[0072] In one feasible embodiment, multiple time-scale traffic feature datasets corresponding to the time to be predicted are obtained based on real-time traffic data and historical traffic data; for each time-scale traffic feature dataset, discrete mutual information between each traffic feature data in the time-scale traffic feature dataset and the target predicted flow is obtained; the importance score of the traffic feature data is determined based on the discrete mutual information; and the traffic feature data used to form the multi-time-scale feature set is determined based on the number of assigned features corresponding to the traffic feature dataset of the time-scale where the traffic feature data is located and the importance score of the traffic feature data.

[0073] Specifically, based on the construction method of traffic feature datasets at each time scale as described above, and based on real-time traffic data and historical traffic data, each traffic feature data in the traffic feature data at each time scale is obtained by calculation, thereby obtaining multiple traffic feature datasets corresponding to the time to be predicted.

[0074] Furthermore, for each time-scale traffic feature dataset, the discrete probability between each traffic feature data point and the target traffic flow is obtained, i.e., the probability of simultaneous occurrence, which serves as the discrete mutual information between the traffic feature data and the target predicted traffic flow. Specifically, the discrete mutual information can be calculated using the following formula:

[0075]

[0076] in, Let be the discrete mutual information between the i-th feature at scale s and the target predicted flow Y. , , Let i be the i-th feature at scale s. Let be the joint probability between the i-th feature at scale s and the target predicted flow Y. Let be the marginal probability between the i-th feature at scale s and the target predicted flow Y. Let Y be the range of values ​​for the target flow rate.

[0077] Then, for each traffic feature data in the time-scale traffic feature dataset, an importance score is calculated, which can be done using the following formula:

[0078]

[0079] in, The importance score for the i-th feature at scale s. Let be the discrete mutual information between the i-th feature at scale s and the target predicted flow Y. Let be the j-th feature at scale s.

[0080] Furthermore, the number of assigned features corresponding to the traffic feature dataset at the specified time scale is determined. Specifically, for each time scale traffic feature dataset, the average mutual information corresponding to that time scale traffic feature dataset is determined. Based on the average mutual information and the total number of features across multiple time scale traffic feature datasets, the number of assigned features corresponding to each time scale traffic feature dataset is determined.

[0081] For example, the number of assigned features can be determined using the following formula:

[0082]

[0083] in, The number of assigned features corresponding to the traffic feature dataset at scale s is . The total number of traffic features Let be the average mutual information of traffic feature data at scale s, and let be the sum of mutual information of traffic feature datasets at multiple time scales.

[0084] After obtaining the number of assigned features for each time-scale traffic feature dataset, the number of traffic features with the highest importance scores in that time-scale traffic feature dataset are used as the traffic feature data in the multi-time-scale feature set. For example, if the number of assigned features corresponding to the micro-scale is 2, then the two traffic feature data with the highest importance scores in the micro-scale traffic feature dataset will be used as the traffic feature data in the multi-time-scale feature set.

[0085] Therefore, this application embodiment divides real-time traffic data and historical traffic data into multiple time-scale traffic feature datasets, enabling multi-scale classification analysis of traffic data features to obtain expressive information reflecting different types of target road segments. Simultaneously, by selecting the final multi-time-scale feature set for traffic flow prediction through discrete mutual information and importance scoring, it effectively selects reasonable traffic feature data for target traffic flow prediction, significantly reducing the amount of redundant feature data, effectively avoiding the influence of redundant data on the final prediction data, and improving the accuracy of target traffic flow prediction.

[0086] Step 203: Input the multi-time-scale feature set into the SVM model and the speed-flow model designed based on the multi-time-scale features respectively to predict the traffic flow of the target road segment at the time to be predicted, and obtain the first predicted traffic flow data predicted by the SVM model and the second predicted traffic flow data predicted by the speed-flow model.

[0087] In a feasible embodiment, the SVM model designed based on multi-timescale features adopts the RBF kernel function based on scale correspondence correction, and the scale correspondence weights are determined by the average mutual information of the time-scale traffic feature dataset.

[0088] For example, the prediction function of an SVM model designed based on multi-timescale features is as follows:

[0089]

[0090] in, This is the first predicted traffic data. For multi-scale RBF kernel functions, , For Lagrange multipliers, This is a bias term.

[0091] Furthermore, the multi-scale RBF kernel function can be expressed as follows:

[0092]

[0093] in, The weights corresponding to scale s, Let be the RBF kernel parameters at scale s. This is a subset of the input features at scale s. This is a subset of the features of the support vectors at scale s.

[0094] Furthermore, The following formula is used for calculation:

[0095]

[0096] in, Let s be the average mutual information of scale s, and T be the temperature parameter. When T=1, the weights are directly allocated proportionally. When T<1, the weights of scales with high mutual information will be amplified. When T>1, the weight distribution will be more uniform.

[0097] Therefore, by using an SVM model designed based on multi-timescale features, it is possible to enhance the influence of information-rich scales on the target predicted traffic flow during analysis and prediction, while suppressing scales with less information. Moreover, the weights can be updated periodically, so that even if the interval between the feature and the time to be predicted is large during the long-term prediction process, such as the ultra-macro scale, the feature still has a high scale weight when the target road segment has strong regularity, thereby improving the influence of the corresponding scale feature data on the prediction results.

[0098] On the other hand, a speed-flow model designed based on multi-timescale features is used to predict traffic flow and obtain second predicted traffic flow data. This includes: calculating the predicted probability of each traffic state based on the multi-timescale feature set; calculating the predicted traffic flow value for each traffic state using the corresponding multi-scale regression sub-model; and weighting all the predicted traffic flow values ​​by posterior probability to obtain the second predicted traffic flow data.

[0099] Specifically, state flow prediction is performed based on the scale-corresponding weights.

[0100] Specifically, the velocity-flow model designed based on multi-timescale features can use the following prediction formula:

[0101]

[0102] in, For the second predicted traffic data, K is the total number of state types, and k is the state type. Let be the posterior probability of state k. This is the predicted state flow rate for state k.

[0103] Furthermore, the following formula is used for calculation. :

[0104]

[0105] in, This represents the predicted state flow rate under state k. Let be the weights corresponding to scale s under state k. For multi-scale regression sub-models, This represents the current speed in real-time traffic data.

[0106] Therefore, this application's embodiments utilize a speed-flow model designed based on multi-timescale features for traffic flow prediction. This improved model automatically identifies traffic state patterns while maintaining the fundamental laws governing the speed-flow relationship, ensuring high physical plausibility in the prediction results. Furthermore, by combining multi-timescale features and corresponding weights, it can focus on feature data at different time scales under different traffic conditions and reduce boundary effects in the prediction results through probability weighting, thereby improving the accuracy and adaptability of traffic flow prediction using the speed-flow model.

[0107] Step 204: Adaptively fuse the first predicted traffic data and the second predicted traffic data to obtain the target predicted traffic flow of the target road segment at the time to be predicted.

[0108] Preferably, the prediction time can be 5 minutes. That is, in this embodiment of the application, the prediction resolution is improved by setting a prediction frequency of 5 minutes, thereby achieving high-resolution prediction.

[0109] In summary, the high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model proposed in this application acquires real-time and historical traffic data of the target road segment. Based on the real-time and historical traffic data, a multi-time-scale feature set is constructed. The multi-time-scale feature set is then input into the SVM model and speed-flow model designed based on the multi-time-scale features, respectively, to predict the traffic flow of the target road segment at the time to be predicted. This yields the first predicted traffic flow data predicted by the SVM model and the second predicted traffic flow data predicted by the speed-flow model. The first and second predicted traffic flow data are then adaptively fused to obtain the target predicted traffic flow of the target road segment at the time to be predicted. This method fully combines the data-driven advantages of the SVM model and the physical law advantages of the speed-flow model, achieving high-resolution and high-precision prediction of traffic flow.

[0110] It should be noted that although the operation of the method of the present invention is described in a specific order in the accompanying drawings, this does not require or imply that the operations must be performed in that specific order, or that all the operations shown must be performed in order to achieve the desired result.

[0111] Figure 3 A schematic diagram of the structure of a high-resolution traffic flow prediction device based on the fusion of an SVM model and a speed-flow model provided in an embodiment of this application is shown.

[0112] like Figure 3 As shown, the high-resolution traffic flow prediction device 10 based on the fusion of SVM model and speed-flow model includes:

[0113] The acquisition module 11 is used to acquire real-time traffic data and historical traffic data of the target road segment. The traffic data includes at least the speed of at least one vehicle in the target road segment and the corresponding time.

[0114] The construction module 12 is used to construct a multi-timescale feature set based on the real-time traffic data and the historical traffic data. The multi-timescale feature set consists of at least some features from multiple timescale traffic feature datasets, and the number of traffic feature data at each time scale in the multi-timescale feature set is related to the target predicted traffic flow at the time to be predicted. Each timescale traffic feature dataset corresponds to one time scale. The traffic feature data in the timescale traffic feature dataset is calculated based on the real-time traffic data and the historical traffic data.

[0115] Prediction module 13 is used to input the multi-time-scale feature set into the SVM model and the speed-flow model designed based on the multi-time-scale features respectively to predict the flow of the target road segment at the time to be predicted, and to obtain the first predicted flow data predicted by the SVM model and the second predicted flow data predicted by the speed-flow model.

[0116] The fusion module 14 is used to adaptively fuse the first predicted traffic data and the second predicted traffic data to obtain the target predicted traffic of the target road segment at the time to be predicted.

[0117] In some embodiments, the construction module 12 is specifically used for:

[0118] Based on the real-time traffic data and the historical traffic data, obtain traffic feature datasets at multiple time scales corresponding to the time to be predicted;

[0119] For each time-scale traffic feature dataset, obtain the discrete mutual information between each traffic feature data in the time-scale traffic feature dataset and the target predicted traffic flow;

[0120] Based on the discrete mutual information, the importance score of the traffic feature data is determined;

[0121] Based on the number of assigned features corresponding to the traffic feature dataset at the time scale where the traffic feature data is located and the importance score of the traffic feature data, the traffic feature data used to form the multi-time scale feature set is determined.

[0122] In some embodiments, the construction module 12 is specifically used for:

[0123] For each of the time-scale traffic feature datasets, determine the average mutual information corresponding to the time-scale traffic feature dataset;

[0124] The number of assigned features corresponding to each of the time-scale traffic feature datasets is determined based on the average mutual information and the total number of features of the multiple time-scale traffic feature datasets.

[0125] In some embodiments, the SVM model designed based on multi-timescale features adopts an RBF kernel function based on scale-corresponding weights, wherein the scale-corresponding weights are determined by the average mutual information of the time-scale traffic feature dataset.

[0126] In some embodiments, the prediction module 13 is specifically used for:

[0127] The predicted probability of each traffic state is calculated based on the multi-timescale feature set.

[0128] For each traffic state, the predicted state traffic flow is calculated using the multi-scale regression sub-model corresponding to that state.

[0129] The second predicted flow data is obtained by weighting all the predicted flow values ​​according to their posterior probabilities.

[0130] In some embodiments, state flow prediction is performed based on scale-corresponding weights.

[0131] It should be understood that the modules or modules described in the high-resolution traffic flow prediction device 10 based on the fusion of SVM model and speed-flow model are similar to those in the reference model. Figure 2The steps in the described method correspond accordingly. Therefore, the operations and features described above for the method are also applicable to the high-resolution traffic flow prediction device 10 based on the fusion of SVM model and speed-flow model, and its included modules, and will not be repeated here. The high-resolution traffic flow prediction device 10 based on the fusion of SVM model and speed-flow model can be pre-implemented in the browser or other secure applications of an electronic device, or it can be loaded into the browser or other secure applications of an electronic device through download or other means. The corresponding modules in the high-resolution traffic flow prediction device 10 based on the fusion of SVM model and speed-flow model can cooperate with the modules in the electronic device to implement the solution of the embodiments of this application.

[0132] The division of modules or units mentioned in the detailed description above is not mandatory. In fact, according to the embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0133] The following is for reference. Figure 4 , Figure 4 A schematic diagram of the structure of a computer system suitable for implementing the embodiments of this application is shown.

[0134] like Figure 4 As shown, the computer system 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 402 or programs loaded from storage section 408 into random access memory (RAM) 403. RAM 403 also stores various programs and data required for the system's operating instructions. CPU 401, ROM 402, and RAM 403 are interconnected via bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0135] The following components are connected to I / O interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, modem, etc. The communication section 409 performs communication processing via a network such as the Internet. Drive 410 is also connected to I / O interface 405 as needed. Removable media 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 410 as needed so that computer programs read from them can be installed into storage section 408 as needed.

[0136] Specifically, according to embodiments of this application, the flowchart above refers to... Figure 2 The described process can be implemented as a computer software program. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such an embodiment, the computer program contains program code for performing the methods shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by central processing unit (CPU) 401, it performs the functions defined in the system of this application.

[0137] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operational instructions of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two connected blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified functions or operational instructions, or using a combination of dedicated hardware and computer instructions.

[0139] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be housed in a processor; for example, a processor can be described as including an acquisition module, a construction module, a prediction module, and a fusion module. The names of these units or modules do not necessarily limit the specific unit or module itself. For example, the acquisition module can also be described as "acquiring real-time traffic data and historical traffic data of a target road segment, wherein the traffic data includes at least the speed of at least one vehicle within the target road segment and the corresponding time."

[0140] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments, or may exist independently and not assembled into the electronic device. The computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the high-resolution traffic flow prediction method based on the fusion of SVM model and speed-flow model described in this application.

[0141] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

[0142] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A high-resolution traffic flow prediction method based on multi-model fusion, characterized in that, include: Acquire real-time and historical traffic data for the target road segment, wherein the traffic data includes at least the speed of at least one vehicle in the target road segment and the corresponding time. Based on the real-time traffic data and historical traffic data, a multi-time-scale feature set is constructed. The multi-time-scale feature set is respectively input into the SVM model and the speed-flow model designed based on the multi-time-scale features to predict the traffic flow of the target road segment at the time to be predicted, so as to obtain the first predicted traffic flow data predicted by the SVM model and the second predicted traffic flow data predicted by the speed-flow model. The first predicted traffic data and the second predicted traffic data are adaptively fused to obtain the target predicted traffic flow of the target road segment at the time to be predicted.

2. The high-resolution traffic flow prediction method based on multi-model fusion according to claim 1, characterized in that, The steps for constructing a multi-timescale feature set include: Based on the real-time traffic data and historical traffic data, obtain traffic feature datasets at multiple time scales corresponding to the time to be predicted; For each time-scale traffic feature dataset, obtain the discrete mutual information between each traffic feature data in the time-scale traffic feature dataset and the target predicted traffic flow; Based on the discrete mutual information, the importance score of the traffic feature data is determined; Based on the number of assigned features corresponding to the traffic feature dataset at the time scale where the traffic feature data is located and the importance score of the traffic feature data, the traffic feature data used to form the multi-time scale feature set is determined.

3. The high-resolution traffic flow prediction method based on multi-model fusion according to claim 2, characterized in that, Also includes: For each of the time-scale traffic feature datasets, determine the average mutual information corresponding to the time-scale traffic feature dataset; The number of assigned features corresponding to each time-scale traffic feature dataset is determined based on the average mutual information and the total number of features in the multiple time-scale traffic feature datasets.

4. The high-resolution traffic flow prediction method based on multi-model fusion according to claim 1, characterized in that, The SVM model designed based on multi-timescale features adopts the RBF kernel function based on scale-corresponding weights, and the scale-corresponding weights are determined by the average mutual information of the time-scale traffic feature dataset.

5. The high-resolution traffic flow prediction method based on multi-model fusion according to claim 1, characterized in that, The steps of the speed-flow model to predict the flow rate of the target road segment at the predicted time include: The predicted probability of each traffic state is calculated based on the multi-timescale feature set. For each traffic state, the predicted state traffic flow is calculated using the multi-scale regression sub-model corresponding to that traffic state. The second predicted flow data is obtained by weighting all the predicted flow values ​​according to their posterior probabilities.

6. The high-resolution traffic flow prediction method based on multi-model fusion according to claim 5, characterized in that, State flow prediction is performed based on the weights corresponding to the scale.

7. A high-resolution traffic flow prediction device based on multi-model fusion, characterized in that, include: The acquisition module is used to acquire real-time traffic data and historical traffic data of the target road segment. The traffic data includes at least the speed of at least one vehicle in the target road segment and the corresponding time. A construction module is used to construct a multi-timescale feature set based on the real-time traffic data and the historical traffic data. The multi-timescale feature set consists of at least some features from multiple timescale traffic feature datasets, and the number of traffic feature data at each time scale in the multi-timescale feature set is related to the target predicted traffic flow at the time to be predicted. Each timescale traffic feature dataset corresponds to one time scale. The traffic feature data in the timescale traffic feature dataset is calculated based on the real-time traffic data and the historical traffic data. The prediction module is used to input the multi-time-scale feature set into the SVM model and the speed-flow model designed based on the multi-time-scale features respectively to predict the traffic flow of the target road segment at the time to be predicted, and to obtain the first predicted traffic flow data predicted by the SVM model and the second predicted traffic flow data predicted by the speed-flow model. The fusion module is used to adaptively fuse the first predicted traffic data and the second predicted traffic data to obtain the target predicted traffic of the target road segment at the time to be predicted.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the high-resolution traffic flow prediction method based on multi-model fusion as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the high-resolution traffic flow prediction method based on multi-model fusion as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the high-resolution traffic flow prediction method based on multi-model fusion as described in any one of claims 1-6.

Images