Traffic flow prediction method fusing air pollution factors

Abstract

The invention provides a traffic flow prediction method fusing air pollution factors, and the method comprises the steps: obtaining a traffic topological structure, historical traffic flow and air pollutant concentration data, constructing a traffic feature matrix, an adjacent matrix, a related pollutant concentration attribute matrix, and a mapping function, the air pollutant concentration data are fused through an attribute enhancement unit; adopting K-means clustering to screen pollutants with the highest correlation with the traffic flow, and utilizing an attribute enhancement unit to integrate the pollutants into traffic flow features; and building and training a deep learning model for traffic flow prediction. According to the method, by combining the relevance between the spatial-temporal characteristics of the traffic flow and the air pollutants, the limitation that an existing method only depends on traffic flow data is overcome, and the prediction accuracy is effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of traffic flow prediction, and in particular to a traffic flow prediction method integrating air pollution factors. Background Art

[0002] Traffic flow forecasting refers to the technology of using historical and / or real-time traffic data (such as vehicle volume, speed, event information, etc.) to estimate the traffic operation status (such as traffic volume, congestion level) of specific time periods, road sections or regions in the future through mathematical models or algorithms, aiming to provide support for traffic management and travel decision-making.

[0003] Currently common prediction methods include time series models based on historical statistical laws (such as ARIMA), traditional machine learning methods that rely on manual feature extraction (such as support vector machines and random forests), deep learning methods that can automatically learn complex spatiotemporal dependencies (such as RNN, LSTM, GNN and its variants STGCN, DCRNN), and hybrid methods that integrate multi-source data (such as weather and events).

[0004] However, existing methods generally have the following shortcomings: first, they fail to fully utilize the spatiotemporal correlation characteristics of traffic flow data; second, they ignore the impact mechanism of air pollutant concentration data on traffic flow. Summary of the Invention

[0005] The purpose of the present invention is to provide a traffic flow prediction method that integrates air pollution factors, and to solve at least one problem existing in the prior art by integrating the spatiotemporal correlation characteristics of road traffic flow with ambient air quality information.

[0006] The present invention provides a traffic flow prediction method integrating air pollution factors, comprising the following steps:

[0007] Obtain road network topology information, historical traffic flow time series datasets, and air pollutant concentration datasets for the target area;

[0008] An adjacency matrix A representing road network connectivity is established based on the road network topology information, a traffic characteristic matrix X is established based on the historical traffic flow time series dataset, and a concentration matrix Y of multiple target pollutants is established based on the air pollutant concentration dataset. The concentration matrix Y of the target pollutant with the highest correlation with the traffic characteristic matrix X is expressed as a related pollutant concentration attribute matrix K;

[0009] Building a neural network model, and training the neural network model using the adjacency matrix A, the traffic characteristic matrix X, and the relevant pollutant concentration attribute matrix K; wherein, an augmented matrix P is established based on the traffic characteristic matrix X and the relevant pollutant concentration attribute matrix K to fuse the air pollutant concentration data;

[0010] According to the loss function, the parameters of the neural network model are optimized by a back propagation algorithm to obtain a traffic flow prediction model;

[0011] Traffic flow prediction is performed using the traffic flow prediction model.

[0012] This method uses air pollutant concentration data for traffic flow forecasting, improving the reliability of traffic flow forecasting. At the same time, a related pollutant concentration attribute matrix is ​​established based on the historical traffic flow time series dataset and the air pollutant concentration dataset, and historical traffic flow is associated with air pollutant concentrations, thereby improving the accuracy of traffic flow forecasting.

[0013] Based on the traffic flow prediction method that integrates air pollution factors, the present invention provides a traffic flow prediction device, which includes one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to implement the traffic flow prediction method that integrates air pollution factors.

[0014] In addition, the present invention also provides a computer-readable storage medium storing instructions, which, when executed by a processor, implement the traffic flow prediction method integrating air pollution factors.

[0015] The beneficial technical effect of the present invention is that by constructing a multimodal fusion traffic flow prediction model of traffic flow and air pollutants, an innovative dual-dimensional feature extraction mechanism of time, space and environment is established, which not only fully explores the temporal and spatial evolution laws of traffic flow data, but also accurately captures the impact of air quality fluctuations on traffic patterns, effectively solving the problem of insufficient modeling of traffic environment interactions in traditional models. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 Flowchart of a traffic flow prediction method integrating air pollution factors in an embodiment of the present invention.

[0018] Figure 2 Schematic diagram of the road network (streets) of the target area in an embodiment of the present invention.

[0019] Figure 3Schematic diagram of a road network topology graph and an adjacency matrix in an embodiment of the present invention, wherein (a) is a road network topology graph and (b) is a schematic diagram of an adjacency matrix.

[0020] Figure 4 Schematic diagram of integrating air pollutant concentration data into a traffic characteristic matrix in an embodiment of the present invention.

[0021] Figure 5 Schematic diagram of the structure of the neural network model in an embodiment of the present invention.

[0022] Figure 6 This is a comparison chart of the prediction results of the traffic flow prediction method that integrates air pollution factors in an embodiment of the present invention, where (a) is the prediction result chart and (b) is the actual result chart. DETAILED DESCRIPTION

[0023] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Traffic flow forecasting plays a vital role in urban planning. Integrating traffic flow forecasting with air quality data not only helps planners more comprehensively assess the environmental impact of traffic, thereby developing more sustainable traffic management strategies, but also improves forecast accuracy and optimizes traffic control measures by linking air quality and traffic flow. However, due to the complex interactions between urban transportation systems and air pollution, accurately modeling the relationship between the two remains a challenging task.

[0025] Traffic flow prediction and air quality exhibit complex temporal dependencies. Both traffic volume and pollutant concentrations vary dynamically over time and exhibit cyclical characteristics. For example, traffic congestion during peak hours is often accompanied by increases in pollutant concentrations, while air quality may improve at night when traffic is low. While existing traffic flow prediction methods (such as Kalman filtering, support vector regression, and recurrent neural networks) can capture temporal dependencies, most fail to fully integrate the mechanisms by which time-series air quality data influences traffic flow.

[0026] Furthermore, traffic flow and air quality exhibit complex spatial dependencies. From a spatial distribution perspective, the diffusion of traffic pollutants is influenced by road network structure and meteorological conditions, and the feedback effects of air quality on traffic flow vary across regions. For example, higher pollutant concentrations around major roads may suppress travel demand, while lower pollution levels in green areas may attract more traffic. While existing research has attempted to model spatial correlation using convolutional neural networks or graph neural networks, few have simultaneously considered the coupled relationship between road network topology, traffic flow, and pollutant diffusion, making it difficult to accurately characterize the heterogeneous associations between traffic patterns and air quality across regions. Therefore, effectively integrating multi-source spatiotemporal data (such as traffic flow, road network structure, and air quality monitoring data) to develop more accurate prediction models remains a challenge.

[0027] In order to solve at least one of the above technical problems, the present invention provides a traffic flow prediction method that integrates air pollution factors. This method innovatively introduces air quality monitoring data as an auxiliary feature into the prediction model. By establishing a traffic-environment coupling analysis framework, it significantly improves the accuracy of the prediction results and the robustness of the model in complex environments.

[0028] like Figure 1 As shown in FIG, a traffic flow prediction method integrating air pollution factors includes the following steps:

[0029] S1: Obtain the road network topology information, historical traffic flow time series dataset, and air pollutant concentration dataset of the target area;

[0030] S2: establishing an adjacency matrix A representing road network connectivity based on the road network topology information, establishing a traffic characteristic matrix X based on the historical traffic flow time series dataset, establishing a concentration matrix Y of multiple target pollutants based on the air pollutant concentration dataset, and representing the concentration matrix Y of the target pollutant with the highest correlation with the traffic characteristic matrix X as a related pollutant concentration attribute matrix K;

[0031] S3: Building a neural network model, and using the adjacency matrix A, the traffic characteristic matrix X, and the relevant pollutant concentration attribute matrix K to train the neural network model; wherein, an augmented matrix P is established based on the traffic characteristic matrix X and the relevant pollutant concentration attribute matrix K to integrate the air pollutant concentration data;

[0032] S4: Optimizing the parameters of the neural network model through a back propagation algorithm according to the loss function to obtain a traffic flow prediction model;

[0033] S5: Use the traffic flow prediction model to perform traffic flow prediction.

[0034] Among them, road network topology information can be obtained by collecting traffic road network information in the target area, historical traffic flow time series datasets can be obtained by measuring in the target area or based on traffic monitoring data, and air pollutant concentration datasets can be obtained by measuring in the target area or based on meteorological monitoring data. The data can be preprocessed, for example, using the MinMax scaler to normalize the historical traffic flow time series dataset and using linear interpolation to fill in the missing data. The dataset is divided into a training set and a test set for training the neural network model. The target pollutants can include one of carbon monoxide, particulate matter, sulfur dioxide, nitrogen dioxide, and ozone, thereby establishing the concentration matrix Y of the five target pollutants.

[0035] Furthermore, in step S2, establishing an adjacency matrix A representing the connectivity of the road network according to the road network topology structure information includes: modeling the road network topology structure information to obtain a road network topology graph G = (V, E) representing the connection relationship between roads, an adjacency matrix A representing the connectivity of the road network, and a mapping function ; Where V represents the set of road segments, and E represents the set of edges that connect two road segments.

[0036] Specifically, if Figure 2 and Figure 3 As shown, the road segment set is represented as V, V = v1, v2, ..., v n , where n represents the number of road sections, also known as the number of nodes; the set of edges is represented by E, E = e1, e2, …, e m , m represents the number of edges; if two road segments are connected, the connection relationship is marked as an undirected edge; if two road segments are not connected, there is no edge relationship; the road segment set V and the edge set E are drawn as a point-line graph, which is recorded as the road network topology graph G;

[0037] Ignoring the situation where a road segment is self-connected, create an adjacency matrix A that is not self-connected. The adjacency matrix A is an n×n matrix. Each element of the adjacency matrix is ​​initialized to 0. An element of 0 in the adjacency matrix A indicates that there is no edge connecting the two road segments.

[0038] Traverse all edges in the road network topology graph. For an undirected edge connecting road segment i and road segment j, set the corresponding elements in row i, column j and row j, column i in the adjacency matrix A to 1. After the traversal is completed, the adjacency matrix A corresponding to the road network topology graph G can be obtained, and the mapping function can be obtained. .

[0039] Furthermore, in step S2, establishing a traffic characteristic matrix X based on the historical traffic flow time series data set includes: establishing a traffic characteristic matrix X based on the historical traffic flow time series data from time tT to time t , , ;in, express The traffic flow of n road sections at the moment, Indicates that the i-th road segment is Traffic flow at all times.

[0040] Furthermore, in step S2, establishing a concentration matrix Y of multiple target pollutants according to the air pollutant concentration data set includes: establishing a concentration matrix Y of target pollutants according to the target pollutant concentration data from time tT to time t , , ;in, express The pollutant concentration of n road sections at the moment, Indicates that the i-th road segment is Pollutant concentration at a given moment.

[0041] Furthermore, the concentration matrix Y of the target pollutant with the highest correlation with the traffic characteristic matrix X is expressed as the related pollutant concentration attribute matrix K, which includes: and the concentration matrix of a certain type of target pollutant The traffic flow data and pollutant concentration data in the dataset are clustered separately to obtain two data with cluster labels. The two data with cluster labels are matched and calculated to obtain the cluster similarity index. ; Cluster similarity index The concentration matrix Y of the target pollutant with the highest value is represented as the related pollutant concentration attribute matrix K.

[0042] Specifically, the clustering method can be a two-step K-means clustering method. Clustering similarity index The larger the value of , the higher the correlation. By adjusting the threshold, the pollutant concentration data with the highest correlation can be used as the data for constructing the augmented matrix P.

[0043] For the traffic flow dataset , a pollutant concentration dataset First, cluster the traffic flow data and air pollutant concentration data to obtain data with cluster labels. Then define a matching matrix ,in, represents the number of identical sample points in the i-th cluster of the traffic flow dataset X and the j-th cluster of the pollutant concentration dataset Y, k and q represent the maximum number of rows and columns of the matrix, respectively. Finally, a clustering similarity index is defined: , the calculation formula is:

[0044]

[0045] in, 、 、 The specific definition is as follows:

[0046]

[0047]

[0048]

[0049] pass The calculation formula shows that when the class number changes in two clusters, it does not affect The value of The size of has no relation to the cluster number. The calculation formula and The calculation formula can be known, ,so ;when When , it means that the results of the two clusterings are completely similar; when , it means that the results of the two clusterings are completely different.

[0050] represents the total number of samples whose traffic flow is clustered into the i-th category, that is, the sum of the distribution of this category in all pollutant concentration cluster categories; represents the total number of samples whose pollutant concentration is clustered into the jth class, that is, the sum of the distribution of this class in all traffic flow clustering classes; It represents the sum of all elements in the matching matrix M, that is, the total number of intersections of sample points of all category combinations in the two clustering results. By comparing the similarity results of the traffic flow and air pollutant concentration clustering results of each grid, the degree of correlation between traffic flow and air pollutant concentration is analyzed. The calculation can be found that It is to define a matrix with the same number of sample points in a certain class of the first clustering and each class of the second clustering, similar to finding the intersection of the two clustering results, so that a certain class of the first clustering and each class of the second clustering are associated. The value of n is related to the matching matrix. 、 、 The specific definition is as follows:

[0051]

[0052]

[0053]

[0054] like Figure 4 As shown, the relevant pollutant concentration attribute matrix K can be expressed as , , ;in, express The pollutant concentration of n road sections at the moment, Indicates that the i-th road segment is The pollutant concentration at the moment. Integrate into the traffic feature matrix in the form of augmented matrix In the example, we get the augmented matrix P; the augmented matrix part is called the attribute enhancement unit, and the mapping function is obtained. The augmented matrix containing the air pollutant concentration data and traffic characteristics information at time t is , the prediction result is expressed as .

[0055] Furthermore, in step S3, a neural network model is built using a deep learning method to enhance the mapping function Modeling was performed; the neural network model consisted of an input layer, a two-layer GCN model, and a one-layer GRU model. The input layer was used to input the traffic feature matrix X, the relevant pollutant concentration attribute matrix K, and the adjacency matrix A. The GCN model was used to capture the spatial characteristics of traffic flow data between road segment nodes, and the GRU model was used to capture the temporal characteristics of traffic flow data and air pollutant concentrations between road segment nodes. The constructed neural network model was trained using the training set data, and the accuracy was tested using the test set data.

[0056] Specifically, if Figure 5 As shown, each layer of the two-layer GCN model includes T GCN processing units. The GCN processing units of the two-layer GCN model correspond one to one. The enhanced traffic feature matrix at each sampling moment corresponds to two GCN processing units. The output of the two-layer GCN processing unit is expressed as , that is, H=h t-T ,…,h t-1 , h t ; The output of the second layer of the GCN model is used as the input of the GRU model. The GRU model captures the temporal features and outputs a new node, which is expressed as H'. That is, the output of the GRU model is H'; the output of the GRU processing unit is expressed as , then the overall output of the GRU model is expressed as .

[0057] Furthermore, in step S4, the loss function between the predicted value and the actual value is calculated, the parameters of the neural network model are optimized by the back propagation algorithm, and the Adam optimizer is used to adaptively adjust the learning rate to optimize the parameters of the neural network model to make the model more accurate.

[0058] Specifically, when calculating the loss function between the predicted value and the actual value, the mean square error is used as the loss function Loss. The calculation formula of the loss function Loss is:

[0059] ;

[0060] in, represents the true value at time t, represents the predicted value at time t, represents L2 regularization, is a hyperparameter that controls the regularization rate.

[0061] Furthermore, in step S5, the trained traffic flow prediction model is used to predict the traffic flow data obtained in real time to obtain a prediction result.

[0062] In addition, the root mean square error RMSE is used as the performance evaluation indicator. The calculation formula of the root mean square error RMSE is:

[0063] ;

[0064] in, express The true value of the moment, express The predicted value at time.

[0065] The embodiment of the present invention uses a dataset from http: / / iot.ee.surrey.ac.uk:8080 / datasets.html to conduct experiments to verify the effectiveness of the above-mentioned neural network model. The dataset of Aarhus, Denmark is used, in which traffic flow data records the average vehicle speed every 5 minutes. Data of 100 road segments were selected from the period from 0:05 on August 2, 2014 to 24:00 on August 16, 2014. The rows of the matrix represent road segments, the columns represent timestamps, and the traffic feature matrix dimension is 100×4320. The time interval of air pollution data is consistent with that of traffic flow data, recorded every 5 minutes; the main monitored air pollutants include ozone, particulate matter, carbon monoxide, sulfur dioxide, and nitrogen dioxide, with units of micrograms per cubic meter, and their dimensions are the same as those of the traffic feature matrix.

[0066] Figure 6 This is a comparison chart of the results of predicting street traffic flow every 5 minutes. Figure 6(a) in the figure is the prediction result. Figure 6 (b) in the figure is the actual result. This prediction reflects the traffic flow through traffic speed. The lower the traffic speed, the higher the traffic flow. Figure 6 It can be seen that the error between the predicted result and the actual result is very small, indicating that the accuracy of traffic flow prediction using the above-mentioned neural network model in the embodiment of the present invention is high.

[0067] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A traffic flow prediction method integrating air pollution factors, characterized in that: The steps include: Obtain road network topology information, historical traffic flow time series datasets, and air pollutant concentration datasets for the target area; An adjacency matrix A representing the connectivity of the road network is established based on the road network topology information, a traffic feature matrix X is established based on the historical traffic flow time series data set, and a concentration matrix Y of multiple target pollutants is established based on the air pollutant concentration data set, and the concentration matrix Y of the target pollutant with the highest correlation with the traffic feature matrix X is expressed as a related pollutant concentration attribute matrix K; wherein, expressing the concentration matrix Y of the target pollutant with the highest correlation with the traffic feature matrix X as the related pollutant concentration attribute matrix K includes: clustering the traffic feature matrix X with the traffic flow data and the target pollutant concentration data in the concentration matrix Y of a target pollutant, respectively, to obtain two data with cluster labels, and performing matching calculation on the two data with cluster labels to obtain a cluster similarity index ; Cluster similarity index The concentration matrix Y of the target pollutant with the highest value is expressed as the related pollutant concentration attribute matrix K; Building a neural network model, and training the neural network model using the adjacency matrix A, the traffic characteristic matrix X, and the relevant pollutant concentration attribute matrix K; wherein, an augmented matrix P is established based on the traffic characteristic matrix X and the relevant pollutant concentration attribute matrix K to fuse the air pollutant concentration data; According to the loss function, the parameters of the neural network model are optimized by a back propagation algorithm to obtain a traffic flow prediction model; Traffic flow prediction is performed using the traffic flow prediction model.

2. The method according to claim 1, characterized in that Establishing an adjacency matrix A representing the connectivity of the road network according to the road network topology structure information includes: modeling the road network topology structure information to obtain a road network topology graph G = (V, E) representing the connection relationship between roads, an adjacency matrix A representing the connectivity of the road network, and a mapping function ; Where V represents the set of road segments, and E represents the set of edges that connect two road segments.

3. The method according to claim 2, characterized in that Establishing a traffic feature matrix X based on the historical traffic flow time series data set includes: establishing a traffic feature matrix X based on the historical traffic flow time series data from time tT to time t , , ;in, express The traffic flow of n road sections at the moment, Indicates that the i-th road segment is Traffic flow at all times.

4. The method according to claim 2, characterized in that Establishing a concentration matrix Y of multiple target pollutants according to the air pollutant concentration data set includes: establishing a concentration matrix of target pollutants from the target pollutant concentration data from time tT to time t , , ;in, express The pollutant concentration of n road sections at the moment, Indicates that the i-th road segment is Pollutant concentration at a given moment.

5. The method according to claim 1, wherein The target pollutants include one of carbon monoxide, particulate matter, sulfur dioxide, nitrogen dioxide, and ozone.

6. The method according to claim 1, characterized in that The clustering is K-means clustering.

7. The method according to claim 1, characterized in that The neural network model includes an input layer, a two-layer GCN model and a one-layer GRU model; wherein, the input layer is used to input the traffic feature matrix X, the relevant pollutant concentration attribute matrix K and the adjacency matrix A, the GCN model is used to capture the spatial characteristics of the traffic flow data between the road section nodes, and the GRU model is used to capture the temporal characteristics of the traffic flow data and the air pollutant concentration between the road section nodes.

8. The method according to claim 1, characterized in that The calculation formula of the loss function is: ； in, represents the true value at time t, represents the predicted value at time t, represents L2 regularization, represents a hyperparameter that controls the regularization rate.

9. A traffic flow prediction device, characterized in that: The method comprises one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to implement the traffic flow prediction method integrating air pollution factors as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that Instructions are stored, and when the instructions are executed by the processor, the traffic flow prediction method integrating air pollution factors as described in any one of claims 1-8 is implemented.

Images