Traffic dynamic analysis and route recommendation method, system, device and medium based on ARIMA and graph neural network driven by signaling data

Abstract

The present invention discloses a traffic dynamic analysis and route recommendation method, system, device and medium based on ARIMA and graph neural network driven by signaling data, the method includes: collecting historical traffic data and preprocessing, synchronizing the preprocessed historical traffic data in time and space; extracting the characteristics of historical traffic data; establishing an ARIMA model using historical traffic flow sequences; predicting future traffic flow using the ARIMA model; constructing a graph neural network; defining a traffic network as a graph, in which nodes are road intersections and edges are road sections, and obtaining traffic network data; training a graph neural network using traffic network data, predicting traffic flow patterns using the trained graph neural network; forming a traffic status map based on future traffic flow and traffic flow patterns; and providing users with multiple route options according to user preferences and expected destinations. The present invention improves the efficiency and intelligence level of the traffic system and reduces traffic congestion.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and specifically relates to a traffic dynamics analysis and route recommendation method, system, device and medium based on ARIMA and graph neural network driven by signaling data. Background Art

[0002] Mobile Signaling is a series of protocols and signal processes used to control the communication connection between the network and the mobile phone in the mobile communication system. Using the continuous signaling data collected from mobile devices, especially the location updates, the real-time traffic conditions of specific sections or areas can be monitored. This data helps identify areas and times with high vehicle density, thereby assessing traffic flow and congestion levels.

[0003] Autoregressive Integrated Moving Average Model (ARIMA) is a statistical method for time series forecasting, which is used to predict the trend of a series of data points in the future. The ARIMA model combines three main parts: autoregression (AR), difference (I) and moving average (MA) for modeling and forecasting. In the field of transportation, the ARIMA model can be used to predict the traffic flow or congestion of a specific road section at a future time point.

[0004] Graph Neural Networks (GNN) is a type of neural network that specializes in processing graph-structured data. Graph data structures consist of nodes (representing entities) and edges (representing relationships between entities). By propagating information between nodes, GNNs are able to learn comprehensive feature representations of nodes that reflect the properties of the nodes themselves and their contextual relationships in the graph. In a traffic network, nodes can represent intersections, road sections, or specific geographic areas, and edges represent road connections between these points.

[0005] As global urbanization accelerates and urban populations continue to grow, urban traffic congestion is becoming increasingly serious. According to a World Bank report, urban congestion causes billions of dollars in economic losses each year, seriously affecting urban sustainable development and the quality of life of residents. Traffic congestion not only increases travel time but also increases the risk of traffic accidents. Although current traffic management systems use various sensors and cameras to collect traffic data, these systems often face problems such as incomplete data coverage and delayed updates, making it difficult to reflect changes in traffic conditions in real time. In addition, existing traffic forecasting methods often rely on historical data and simple time series analysis, lacking the ability to deeply understand complex traffic patterns. Therefore, they are limited in their effectiveness in dynamically predicting traffic flow and congestion, which makes it impossible to provide accurate personalized travel recommendations.

[0006] The Chinese patent with the announcement number CN115331427B discloses a method for optimizing a dynamic traffic restriction scheme to alleviate urban traffic congestion, including the following steps: for a given multi-period dynamic traffic restriction scheme, analyzing its impact on the user's travel mode, travel period and travel path, and constructing a multi-mode user equilibrium model based on the multi-period dynamic traffic restriction scheme; using the output of the first stage as input, and processing it through the multi-period dynamic traffic restriction scheme optimization model of the second stage, the multi-period dynamic traffic restriction scheme is optimized with the restriction ratio, restriction area and restriction period as decision variables. The present invention studies the user travel selection behavior under the multi-period dynamic traffic restriction scheme, and proposes a two-layer programming model, in which the lower model is a multi-mode multi-user equilibrium model based on the multi-period dynamic traffic restriction scheme, and the upper model optimizes the multi-period dynamic traffic restriction scheme to improve the utilization rate of transportation infrastructure and effectively alleviate the traffic pressure in the city.

[0007] The Chinese patent application with publication number CN114020993A discloses a personalized travel route recommendation method based on feature mining algorithm, which includes the following steps: Step 1, collect basic information of users and establish a user data information database; Step 2, analyze user information and establish a user feature model; Step 3, combine data analysis technology with deep learning technology, and use data mining algorithm to mine user interest features as much as possible to improve the personalization and satisfaction of recommendation results; add the route that the user has traveled to the search results, and conduct deep learning through user feedback to obtain a path planning result that better meets the user's needs. In addition, the method can also record the travel routes saved by the user for direct use in the next trip, analyze the personalized outbound route finally recommended by the system, and directly push out the return route so that the user can use it directly when returning.

[0008] In the article "Research on Traffic Congestion Prediction and Mitigation Strategy Based on Intelligent Transportation System", the key role of intelligent transportation system in traffic congestion management is discussed in depth, focusing on the analysis of data collection, analysis methods and mitigation strategies. By adopting advanced data collection and processing technology, data analysis algorithms and real-time information transmission systems, intelligent transportation systems can accurately monitor and control traffic flow. Based on the spatiotemporal characteristics of urban traffic data flow, "Urban Area Short-term Traffic Congestion Prediction Algorithm Based on Deep Learning" proposes a deep learning-based urban traffic congestion prediction model (CS-Transformer). The model uses convolutional neural network (CNN) to extract the spatial characteristics of urban area traffic data based on grid division, enhances the expression ability of the model through fully connected neural network, and integrates location information into traffic data using similarity position encoding mechanism (SPEM). Finally, the Transformer network is used to capture the time-dependent characteristics of traffic data, and the model is verified by Chengdu taxi global positioning system (GPS) data. In the article "Urban Vehicle Travel Route Recommendation Method Based on Travel Time Prediction", the author constructed a dynamic graph convolutional neural network (DGCN) to predict the multi-step travel time of each section of the road network, and recommended vehicle travel routes based on travel demand, avoiding congestion areas and balancing road network factors. The article "Visual Recommendation of Urban Mixed Traffic Based on Multi-layer Complex Network" extracted the traffic feature network from real taxi and shared bicycle trajectory data, and fused it to form a more comprehensive urban mixed traffic feature network, and then used the improved genetic algorithm based on the network to recommend low-carbon mixed traffic routes.

[0009] In summary, most of the current academic research on traffic dynamics analysis focuses on the use of GPS data or sensor data for analysis, while the use of signaling data is relatively small. Traditional traffic congestion prediction methods and personalized travel route recommendation systems are often limited by the limitations of data analysis and recommendation algorithms, and cannot achieve accurate personalized recommendations. Based on this, the present invention proposes a novel traffic congestion prediction and personalized travel recommendation system, which significantly improves the monitoring and prediction capabilities of dynamic changes in traffic conditions by integrating mobile signaling data and using ARIMA models and graph neural networks. Through this integration, the system uses signaling data to update the traffic flow information of the road network in real time, while the graph neural network can analyze the complex interactive relationships in the road network to improve the accuracy and timeliness of the prediction. Based on real-time and predicted traffic information, the system can recommend the best travel time and route for users, thereby optimizing the user's travel experience and alleviating road congestion. Specifically, first, the historical traffic data is analyzed by the ARIMA model to predict the future traffic flow and congestion of a specific road section. This step includes data collection, model parameter identification, model fitting and verification. Next, we use graph neural networks to process and analyze the complex structure of the traffic network, where nodes represent intersections or road sections and edges represent the connections between them, and train the GNN model to identify and predict congestion points in the network. Finally, we combine the prediction results of these two technologies to provide users with travel routes optimized according to real-time and predicted traffic conditions, effectively improving traffic management efficiency and travel experience. Summary of the invention

[0010] In view of the deficiencies in the prior art, the present invention provides a method, system, device and medium for traffic dynamic analysis and route recommendation based on ARIMA and graph neural network driven by signaling data, which realizes real-time monitoring of traffic flow, congestion prediction and route recommendation by combining advanced technologies such as traffic dynamic analysis driven by signaling data, ARIMA model and graph neural network. It improves the efficiency and intelligence level of the transportation system, reduces the impact of traffic congestion on the lives and economic activities of urban residents, improves travel experience and promotes sustainable urban development.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] A traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data includes the following steps:

[0013] Step 1: Collect historical traffic data from mobile signaling, traffic cameras, and GPS;

[0014] Step 2: Preprocess the historical traffic data and synchronize the preprocessed historical traffic data in time and space to ensure that the time and space labels of all historical traffic data are aligned;

[0015] Step 3: Extract the characteristics of historical traffic data, including traffic density, average travel time and traffic flow;

[0016] Step 4: Use the historical traffic flow series to establish an ARIMA model; use the ARIMA model to predict future traffic flow;

[0017] Step 5: Construct a graph neural network; define the traffic network as a graph, where nodes are road intersections, edges are road sections, node features include traffic flow, and edge weights include the length of the road section, average travel time, and traffic density, to obtain traffic network data; use the traffic network data to train the graph neural network, and use the trained graph neural network to predict traffic flow patterns;

[0018] Step 6: Based on the future traffic flow predicted by the ARIMA model and the traffic flow pattern predicted by the graph neural network, a traffic status map is formed; multiple route options are provided to users based on their preferences and intended destinations, each route including estimated travel time and traffic conditions.

[0019] To optimize the above technical solutions, the specific measures taken also include:

[0020] Furthermore, in step 2, the preprocessing of the historical traffic data specifically includes:

[0021] Remove abnormal speed data from historical traffic data, satisfying the condition |v n -μ v |>3σ v The speed data is abnormal speed data, v n Indicates the nth speed data, μ v represents the mean of the velocity data, σ v represents the standard deviation of the velocity data;

[0022] Traffic flow normalization:

[0023]

[0024] In the formula, F norm (t,l) represents the normalized traffic flow, F(t,l) represents the traffic flow of the lth road section at time t, max(F) and min(F) represent the maximum and minimum traffic flow values, respectively.

[0025] Furthermore, in step 2, the time-space synchronization of the pre-processed historical traffic data is specifically as follows:

[0026] Time and space synchronization is achieved using synchronization functions:

[0027]

[0028] Where D sync It is a collection of synchronized historical traffic data. is the ith historical traffic data after synchronization; f sync (d″ i ,T,S) represents the synchronization function, d″ i represents the i-th historical traffic data before synchronization, T and S represent the standard reference systems of time and space respectively; D″ represents the set of preprocessed historical traffic data.

[0029] Furthermore, in step 4, the use of historical traffic flow series to establish an ARIMA model specifically includes:

[0030] Use ADF to test whether the historical traffic flow sequence is stationary. If the sequence is not stationary, perform difference until the sequence is stationary:

[0031] Δx t =x t -x t-1

[0032] Where Δx t represents the first-order difference sequence, x t represents the traffic flow at time t, x t-1 represents the traffic flow at time t-1;

[0033] By observing the ACF and PACF graphs, the number of autoregressive terms p and the number of moving average terms q are determined, and the best parameter combination (p, d, q) is selected, where d represents the number of differences made when the historical traffic flow series becomes a stationary series;

[0034] Fit the ARIMA model using the best parameter combination.

[0035] Furthermore, in step 5, the graph neural network is GCN or GAT; and constructing the graph neural network includes designing the number of network layers, the number of hidden units in each layer, and the activation function.

[0036] Furthermore, in step 5, the use of traffic network data to train the graph neural network is specifically as follows:

[0037] Divide the traffic network data into a training set and a test set, use the training set to train the graph neural network, perform forward propagation and backward propagation, and update the weights of the graph neural network; select mean square error or cross entropy loss as the loss function; select Adam or SGD as the optimizer to optimize the graph neural network parameters;

[0038] Use the test set to evaluate the performance of the graph neural network model, with performance indicators including accuracy, recall, and F1 score; adjust hyperparameters such as the graph neural network structure, learning rate, and regularization term, and use grid search or random search methods to find the optimal graph neural network model; analyze the contribution of the features of different nodes and edges to the prediction of the graph neural network model, and adjust the input training set to improve the explanatory power and prediction accuracy of the graph neural network model.

[0039] The present invention also proposes a traffic dynamic analysis and route recommendation system based on ARIMA and graph neural network driven by signaling data, including:

[0040] Data collection module, which collects historical traffic data from mobile signaling, traffic cameras and GPS;

[0041] The preprocessing and synchronization module is used to preprocess the historical traffic data and synchronize the preprocessed historical traffic data in time and space to ensure that the time and space labels of all historical traffic data are aligned;

[0042] Feature extraction module, used to extract features of historical traffic data, including traffic density, average travel time and traffic flow;

[0043] Traffic flow prediction module, used to establish ARIMA model using historical traffic flow series; using ARIMA model to predict future traffic flow;

[0044] Traffic flow pattern prediction module is used to build a graph neural network; define the traffic network as a graph, where nodes are road intersections, edges are road sections, node features include traffic flow, and edge weights include the length of the road section, average travel time, and traffic density, to obtain traffic network data; use the traffic network data to train the graph neural network, and use the trained graph neural network to predict traffic flow patterns;

[0045] The route recommendation module is used to form a traffic status map based on the future traffic flow predicted by the ARIMA model and the traffic flow pattern predicted by the graph neural network; it provides users with multiple route options based on user preferences and intended destinations, and each route includes estimated travel time and traffic conditions.

[0046] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as described above is implemented.

[0047] The present invention also proposes a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as described above.

[0048] The beneficial effects of the present invention are:

[0049] (1) Mobile signaling data analysis: Using advanced data acquisition technology, we can achieve real-time capture and processing of large-scale mobile signaling data. Through this method, the system can accurately monitor and analyze the dynamic changes of urban traffic flow and provide scientific data support for traffic management.

[0050] (2) Time series forecasting optimization: The enhanced ARIMA model is used to perform time series analysis of traffic flow dynamics. The model is specially optimized to adapt to complex urban traffic patterns, effectively predict short-term and long-term traffic trends, and provide accurate forecasting support for traffic scheduling.

[0051] (3) Graph Neural Network Application: We implement an innovative graph neural network architecture specifically designed to analyze complex traffic network structures. This network is able to deeply learn the dependencies between traffic nodes. Through this deep network analysis, the accuracy and response speed of traffic flow prediction can be significantly improved.

[0052] (4) Intelligent route customization system: Develop a highly personalized route recommendation system that combines user behavior analysis and real-time traffic data and uses advanced algorithms to dynamically generate the most optimized route. This personalized service not only improves user satisfaction, but also optimizes the overall allocation of traffic resources.

[0053] (5) Improve traffic management efficiency:

[0054] By monitoring and predicting traffic flow in real time, management departments can conduct traffic scheduling and planning more effectively, thereby reducing traffic congestion and improving road network utilization.

[0055] (6) Optimizing user travel experience: Providing users with personalized route suggestions based on real-time data can significantly reduce users’ travel time and avoid traffic congestion, thereby improving users’ travel satisfaction.

[0056] (7) Enhance traffic safety: Reduce traffic risks caused by congestion or accidents through accurate traffic flow prediction and timely route adjustment.

[0057] (8) Data-driven decision support: Provide decision support based on data analysis to help urban planners and policy makers make transportation planning and policy adjustments based on scientific evidence. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is an overall flow chart of the traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data proposed in the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0060] Embodiment 1

[0061] The present invention proposes a traffic dynamic analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data. The flowchart of the method is as follows: Figure 1 As shown, the following steps are included:

[0062] Step 1: Collect historical traffic data from mobile signaling, traffic cameras, and GPS;

[0063] Step 2: preprocess the historical traffic data, and synchronize the preprocessed historical traffic data in time and space to ensure that the time and space labels of all historical traffic data are aligned; in step 2, the preprocessing of the historical traffic data specifically includes:

[0064] Remove abnormal speed data from historical traffic data, satisfying the condition |v n -μ v |>3σ v The speed data is abnormal speed data, v n Indicates the nth speed data, μ v represents the mean of the velocity data, σ v represents the standard deviation of the velocity data;

[0065] Traffic flow normalization:

[0066]

[0067] In the formula, F norm (t,l) represents the normalized traffic flow, F(t,l) represents the traffic flow of the lth road section at time t, max(F) and min(F) represent the maximum and minimum traffic flow values, respectively.

[0068] In step 2, the time and space synchronization of the pre-processed historical traffic data is specifically as follows:

[0069] Time and space synchronization is achieved using synchronization functions:

[0070]

[0071] Where D sync It is a collection of synchronized historical traffic data. is the ith historical traffic data after synchronization; f sync (d″ i ,T,S) represents the synchronization function, d″ i represents the i-th historical traffic data before synchronization, T and S represent the standard reference systems of time and space respectively; D″ represents the set of preprocessed historical traffic data.

[0072] Step 3: Extract the characteristics of historical traffic data, including traffic density, average travel time and traffic flow;

[0073] Traffic density D is an important feature that represents the number of vehicles on a unit road section and can be used to predict congestion. Assume that the length of the road section l is L,

[0074]

[0075] Where D(t,l) is the traffic density on road section l at time t, and N(t,l) is the number of vehicles observed on road section l at time t.

[0076] Average travel time T avg Indicates the average time required to travel a specific road segment;

[0077]

[0078] Where L is the length of the road section, V avg is the average speed.

[0079] Step 4: Establishing an ARIMA model using the historical traffic flow sequence; and predicting future traffic flow using the ARIMA model. In step 4, establishing an ARIMA model using the historical traffic flow sequence specifically includes:

[0080] Use ADF to test whether the historical traffic flow sequence is stationary. If the sequence is not stationary, perform difference until the sequence is stationary:

[0081] Δx t =x t -x t-1

[0082] Where Δx t represents the first-order difference sequence, x t represents the traffic flow at time t, x t-1 represents the traffic flow at time t-1;

[0083] By observing the ACF and PACF graphs, the number of autoregressive terms p and the number of moving average terms q are determined, and the best parameter combination (p, d, q) is selected, where d represents the number of differences made when the historical traffic flow series becomes a stationary series;

[0084] Use the best parameter combination to fit the ARIMA model. The fitted ARIMA model is as follows:

[0085] x t =c+φ1x t-1 +…+φ p x t-p +θ1∈ t-1 +…+θ q ∈ t-q +∈ t

[0086] In the formula, x t represents the current traffic flow, x t-1 represents the traffic flow at the previous moment, x t-p represents the traffic flow in the previous p moments, c is a constant term, ∈ t is the error term, φ and θ are the autoregressive and moving average coefficients of the model respectively, φ1 represents the autoregressive coefficient of the first autoregressive term, φ p represents the autoregressive coefficient of the pth autoregressive term, θ1 represents the moving average coefficient of the first moving average term, θ q represents the moving average coefficient of the qth moving average term, ∈ t-1 represents the first moving average term, ∈ t-q represents the qth moving average term.

[0087] Step 5: Construct a graph neural network; define the traffic network as a graph, where nodes are road intersections, nodes can also be bus stops or any key locations; edges are road sections, node features include traffic flow, edge weights include the length of the road section, average travel time and traffic density, and obtain traffic network data; use the traffic network data to train the graph neural network, and use the trained graph neural network to predict traffic flow patterns; in step 5, the graph neural network is GCN (Graph Convolutional Network) or GAT (Graph Attention Network); the construction of the graph neural network includes designing the number of network layers, the number of hidden units in each layer, and the activation function. Multi-layer GCN can help capture high-order dependencies between nodes.

[0088] In step 5, the method of training the graph neural network using traffic network data is specifically as follows:

[0089] The traffic network data is divided into a training set and a test set, and the graph neural network is trained using the training set. Forward propagation and backward propagation are performed to update the weights of the graph neural network. The loss function selects mean square error or cross entropy loss. The optimizer selects Adam or SGD to optimize the graph neural network parameters. The loss and verification indicators during training are monitored, and early stopping is performed to prevent overfitting.

[0090] Use the test set to evaluate the performance of the graph neural network model, and the performance indicators include accuracy, recall rate and F1 score; adjust the hyperparameters such as the graph neural network structure (such as the number of layers, the number of units per layer), learning rate and regularization term, and use grid search or random search methods to find the optimal graph neural network model; analyze the contribution of the characteristics of different nodes and edges to the prediction of the graph neural network model, and adjust the input training set to improve the explanatory power and prediction accuracy of the graph neural network model.

[0091] Step 6: Based on the future traffic flow predicted by the ARIMA model and the traffic flow pattern predicted by the graph neural network, a traffic status map is formed; multiple route options are provided to users based on their preferences and intended destinations, each route including estimated travel time and traffic conditions.

[0092] Embodiment 2

[0093] The present invention proposes a traffic dynamics analysis and route recommendation system based on ARIMA and graph neural network driven by signaling data corresponding to the first embodiment, including:

[0094] Data collection module, which collects historical traffic data from mobile signaling, traffic cameras and GPS;

[0095] The preprocessing and synchronization module is used to preprocess the historical traffic data and synchronize the preprocessed historical traffic data in time and space to ensure that the time and space labels of all historical traffic data are aligned;

[0096] Feature extraction module, used to extract features of historical traffic data, including traffic density, average travel time and traffic flow;

[0097] Traffic flow prediction module, used to establish ARIMA model using historical traffic flow series; using ARIMA model to predict future traffic flow;

[0098] Traffic flow pattern prediction module is used to build a graph neural network; define the traffic network as a graph, where nodes are road intersections, edges are road sections, node features include traffic flow, and edge weights include the length of the road section, average travel time, and traffic density, to obtain traffic network data; use the traffic network data to train the graph neural network, and use the trained graph neural network to predict traffic flow patterns;

[0099] The route recommendation module is used to form a traffic status map based on the future traffic flow predicted by the ARIMA model and the traffic flow pattern predicted by the graph neural network; it provides users with multiple route options based on user preferences and intended destinations, and each route includes estimated travel time and traffic conditions.

[0100] The implementation methods of each module and module function in the system are completely consistent with the steps of the method in Example 1, so they will not be repeated here.

[0101] Embodiment 3

[0102] The present invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as described in Example 1 is implemented.

[0103] Embodiment 4

[0104] The present invention proposes a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as described in Example 1.

[0105] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0106] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0107] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data, characterized in that: The following steps are involved: Step 1: Collect historical traffic data from mobile signaling, traffic cameras, and GPS; Step 2: Preprocess the historical traffic data and synchronize the preprocessed historical traffic data in time and space to ensure that the time and space labels of all historical traffic data are aligned; Step 3: Extract the characteristics of historical traffic data, including traffic density, average travel time and traffic flow; Step 4: Use the historical traffic flow series to establish an ARIMA model; use the ARIMA model to predict future traffic flow; Step 5: Construct a graph neural network; define the traffic network as a graph, where nodes are road intersections, edges are road sections, node features include traffic flow, and edge weights include the length of the road section, average travel time, and traffic density, to obtain traffic network data; use the traffic network data to train the graph neural network, and use the trained graph neural network to predict traffic flow patterns; Step 6: Based on the future traffic flow predicted by the ARIMA model and the traffic flow pattern predicted by the graph neural network, a traffic status map is formed; multiple route options are provided to users based on their preferences and intended destinations, each route including estimated travel time and traffic conditions.

2. The traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as claimed in claim 1, characterized in that: In step 2, the preprocessing of historical traffic data specifically includes: Remove abnormal speed data from historical traffic data, satisfying the condition |v n -μ v |>3σ v The speed data is abnormal speed data, v n Indicates the nth speed data, μ v represents the mean of the velocity data, σ v represents the standard deviation of the velocity data; Traffic flow normalization: In the formula, F norm (t,l) represents the normalized traffic flow, F(t,l) represents the traffic flow of the lth road section at time t, max(F) and min(F) represent the maximum and minimum traffic flow values, respectively.

3. The traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as claimed in claim 1, characterized in that: In step 2, the time and space synchronization of the pre-processed historical traffic data is specifically as follows: Time and space synchronization is achieved using synchronization functions: Where D sync It is a collection of synchronized historical traffic data. is the ith historical traffic data after synchronization; f sync (d″ i ,T,S) represents the synchronization function, d″ i represents the i-th historical traffic data before synchronization, T and S represent the standard reference systems of time and space respectively; D″ represents the set of preprocessed historical traffic data.

4. The traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as claimed in claim 1, characterized in that: In step 4, the use of the historical traffic flow sequence to establish the ARIMA model specifically includes: using ADF to test whether the historical traffic flow sequence is stationary, and if the sequence is not stationary, performing a difference until the sequence is stationary: Δx t =x t -x t-1 Where Δx t represents the first-order difference sequence, x t represents the traffic flow at time t, x t-1 represents the traffic flow at time t-1; By observing the ACF and PACF graphs, the number of autoregressive terms p and the number of moving average terms q are determined, and the best parameter combination (o, d, q) is selected, where d represents the number of differences made when the historical traffic flow series becomes a stationary series; Fit the ARIMA model using the best parameter combination.

5. The traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as claimed in claim 1, characterized in that: In step 5, the graph neural network is GCN or GAT; the construction of the graph neural network includes designing the number of network layers, the number of hidden units in each layer, and the activation function.

6. The traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as claimed in claim 1, characterized in that: In step 5, the method of training the graph neural network using traffic network data is specifically as follows: Divide the traffic network data into a training set and a test set, use the training set to train the graph neural network, perform forward propagation and backward propagation, and update the weights of the graph neural network; select mean square error or cross entropy loss as the loss function; select Adam or SGD as the optimizer to optimize the graph neural network parameters; Use the test set to evaluate the performance of the graph neural network model, with performance indicators including accuracy, recall, and F1 score; adjust hyperparameters such as the graph neural network structure, learning rate, and regularization term, and use grid search or random search methods to find the optimal graph neural network model; analyze the contribution of the features of different nodes and edges to the prediction of the graph neural network model, and adjust the input training set to improve the explanatory power and prediction accuracy of the graph neural network model.

7. A traffic dynamics analysis and route recommendation system based on ARIMA and graph neural network driven by signaling data, characterized in that: include: Data collection module, which collects historical traffic data from mobile signaling, traffic cameras and GPS; The preprocessing and synchronization module is used to preprocess the historical traffic data and synchronize the preprocessed historical traffic data in time and space to ensure that the time and space labels of all historical traffic data are aligned; Feature extraction module, used to extract features of historical traffic data, including traffic density, average travel time and traffic flow; Traffic flow prediction module, used to establish ARIMA model using historical traffic flow series; using ARIMA model to predict future traffic flow; Traffic flow pattern prediction module, used to build graph neural network; Define the traffic network as a graph, where nodes are road intersections, edges are road sections, node features include traffic flow, and edge weights include the length of the road section, average travel time, and traffic density, to obtain traffic network data; use the traffic network data to train a graph neural network, and use the trained graph neural network to predict traffic flow patterns; The route recommendation module is used to form a traffic status map based on the future traffic flow predicted by the ARIMA model and the traffic flow pattern predicted by the graph neural network; it provides users with multiple route options based on user preferences and intended destinations, and each route includes estimated travel time and traffic conditions.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for traffic dynamics analysis and route recommendation based on ARIMA and graph neural network driven by signaling data as described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: A computer program is stored, and the computer program enables the computer to execute the traffic dynamics analysis and route recommendation method based on ARIMA and graph neural network driven by signaling data as described in any one of claims 1 to 6.

Images