Traffic demand prediction method, traffic guidance method and equipment

Abstract

The invention discloses a traffic demand prediction method, a traffic guidance method and equipment, and relates to the technical field of intelligent traffic. Traffic zones are divided for a research area, and mobile phone signaling data in the research area within a preset historical time range are acquired; oD data and road signaling flow data between traffic cells of a preset time aggregation window are calculated according to the mobile phone signaling data; obtaining road network data of a research area, and carrying out spatial association on OD data, road signaling flow data and road network data between traffic cells; key traffic nodes and flow bottlenecks are identified, and the corrected road section flow of each road on the road network is generated in combination with a space correlation result; training an LSTM model according to the corrected road section flow data; and using the trained LSTM model to predict traffic demand data of each road in the whole road network in a future time period. According to the method, the whole-day time is sliced, and the spatial granularity is improved to the road level by using the GIS technology, so that the high temporal-spatial resolution of traffic flow prediction is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation, and in particular to a traffic demand prediction method, a traffic guidance method and a device. Background Art

[0002] Urban traffic demand is characterized by uneven temporal and spatial distribution and rapid dynamic changes. Traditional traffic management methods are difficult to adapt to the increasingly complex travel needs. Accurately predicting traffic demand and taking effective guidance measures are of great significance to alleviating congestion and improving travel efficiency. Existing technologies mainly rely on historical data and simple statistical models, with insufficient prediction accuracy and timeliness, and relatively simple guidance methods, which are difficult to meet the needs of refined management. Summary of the invention

[0003] The present invention aims at at least one disadvantage of the prior art and provides a method for predicting traffic demand, comprising the following steps:

[0004] Divide the study area into traffic zones, obtain the mobile phone signaling data within the study area within a preset historical time range, and calculate the OD data and road signaling flow data between the traffic zones within a preset time aggregation window based on the mobile phone signaling data;

[0005] Obtain the road network data of the study area, and spatially associate the OD data, road signaling flow data, and road network data between traffic communities;

[0006] Use GIS to identify key traffic nodes and flow bottlenecks, and combine spatial association results to generate corrected section flow for each road on the road network;

[0007] According to the corrected road section traffic, the traffic values ​​of each road in different time aggregation windows are arranged in chronological order according to timestamps to form time series data, and the time series data is used to train the LSTM model; the trained LSTM model is used to predict the traffic demand data of each road in the entire road network in the future time period.

[0008] In order to solve the problem of insufficient spatiotemporal granularity of existing prediction models, the present invention slices the whole day and refines the time granularity to 5 minutes to 10 minutes (time aggregation window), and uses GIS technology to increase the spatial granularity to the road level, thus achieving high spatiotemporal resolution of traffic flow prediction and providing data support for refined traffic management. In order to solve the problem of insufficient generalization ability of existing prediction models, the LSTM neural network algorithm is introduced to enable the prediction model to adapt to different traffic scenarios, thereby improving the robustness, adaptability and accuracy of the prediction.

[0009] Preferably, the spatial association method includes spatially associating the OD data and road network data between traffic communities through QGIS software: the road network data includes road network nodes; importing the traffic community vector layer and the road network node layer in the QGIS software, calculating the distance from the geometric center point of each traffic community to the road network node, for each traffic community, finding the nearest road network node, and establishing a correspondence table between the traffic community and the road network node; using the correspondence table, replacing the starting community and the end community in the OD data between the traffic communities with the corresponding starting road network node and the end road network node, and obtaining an OD matrix at the road network node level.

[0010] In response to the data quality problem of the existing traffic flow prediction model, the accuracy of traffic flow prediction is improved by performing error correction and data cleaning on mobile signaling data and introducing the OD matrix estimation algorithm. This overcomes the limitation of traditional methods that rely on roadside equipment to collect data, fully utilizes mobile signaling big data, and comprehensively perceives the dynamic changes in urban traffic demand.

[0011] Preferably, the spatial association method includes spatially associating the road network data and the road signaling flow data through QGIS software:

[0012] Import the road network node layer and road signaling flow data table into QGIS software, convert the road signaling flow data table into a spatial point layer, and use the spatial connection tool of QGIS software to associate the road flow point layer with the road network node layer;

[0013] Use the field calculator to assign the flow value of the flow point to the road signaling flow field of the corresponding road segment network node.

[0014] Preferably, the method of using GIS to identify key traffic nodes and traffic bottlenecks and combining spatial association results to generate corrected road section traffic for each road on the road network includes:

[0015] Based on the OD data of the traffic zone, the OD flow data of each road on the road network in each time aggregation window is obtained;

[0016] The difference between the signaling flow data and the OD flow data of each road on the road network in each time aggregation window is calculated to obtain the error vector, which is used to correct the section flow and generate a dynamic section flow matrix.

[0017] Preferably, the method further comprises the steps of: performing normalization processing on time series data, comprising the following steps:

[0018] The traffic time series of each road section are placed in their respective sets, and the time series data are normalized; the normalized time series data are used to train the LSTM model.

[0019] (II) Another traffic guidance method of the present invention predicts the traffic demand data of each road in the entire road network in the future time period by using the traffic demand prediction method, and outputs a traffic guidance strategy, including the following steps:

[0020] Regulate and guide travel demand, generate road congestion warning information and detour information, and release detour information through variable traffic signs and intelligent traffic signal systems;

[0021] Evaluate the regulation effect and optimize the LSTM model parameters and guidance strategy based on the evaluation results.

[0022] Among them, the method for generating road congestion warning information includes: determining the traffic capacity threshold of each road based on historical data and traffic management experience, and when the predicted traffic volume exceeds the threshold, determining that there is a probability of congestion in the road section exceeding the threshold during the corresponding time period, and generating road congestion warning information based on the predicted traffic demand data of each road in the entire road network in the future time period.

[0023] A traffic guidance method aims to solve the problems of extensive and inefficient existing traffic induction and guidance methods. By combining the traffic flow prediction results obtained through the traffic demand prediction method and real-time road condition information, and using big data mining technology to analyze the driver's travel preferences, it realizes personalized detours and travel guidance, effectively alleviates local congestion, and improves the traffic efficiency of the road network. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0025] Figure 1 It is a flow chart of the traffic demand forecasting method. DETAILED DESCRIPTION

[0026] The present invention is further described in detail below in conjunction with the embodiments. The following embodiments are explanations of the present invention and the present invention is not limited to the following embodiments. Any changes or substitutions within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

[0027] Glossary:

[0028] A traffic cell refers to the basic spatial unit on a map artificially divided for traffic demand analysis and prediction. The division is based on the rules of each cell being a maximum of 3 square kilometers, ensuring that it does not cross administrative boundaries, and that the boundaries are mainly main roads and expressways, consistent with the coverage of base stations, and combined with the rules of the current land use map. The boundaries of a traffic cell are mainly used to determine whether a signaling base station belongs to the cell.

[0029] Critical Traffic Nodes refer to intersections, bridges, tunnels and other locations that play an important role in the transportation network.

[0030] A traffic bottleneck refers to a place where traffic flow becomes narrow or blocked for some reason at a specific section or location, resulting in a decrease in vehicle speed and increased queuing.

[0031] A network node is a location with a specific geometric or topological meaning in the road network structure. It is included in the road network data, which can be obtained through application to the traffic management department or through other available and public databases.

[0032] OD (Origin-Destination) data refers to the travel volume from the starting community (the starting point Origin is in the community) to the destination community (the destination Destination is in the community) within a period of time, with traffic communities as the unit.

[0033] Road signaling traffic data refers to the reconstruction of the user's travel trajectory through mobile phone signaling data, and then map matching of each signaling trajectory point, with each signaling trajectory point matched to the road closest to it in space.

[0034] Embodiment 1:

[0035] A traffic demand forecasting method is disclosed, comprising the following steps:

[0036] Step 1, divide the study area into traffic zones, obtain the mobile phone signaling data in the study area within a preset historical time range, and calculate the OD data and road signaling flow data between the traffic zones within a preset time aggregation window based on the mobile phone signaling data;

[0037] Step 2, obtain the road network data of the study area, and spatially associate the OD data, road signaling flow data and road network data between traffic cells;

[0038] Use GIS to identify key traffic nodes and flow bottlenecks, and combine spatial association results to generate corrected section flow for each road on the road network;

[0039] Step 3: According to the corrected road section traffic, the traffic values ​​of each road in different time aggregation windows are arranged in chronological order according to timestamps to form time series data, and the time series data is used to train the LSTM model; the trained LSTM model is used to predict the traffic demand data of each road in the entire road network in the future time period.

[0040] Exemplarily, the time aggregation window is 5 minutes, and the implementer can adjust it according to needs, such as 10 minutes, 8 minutes, etc.

[0041] In step 1, the OD data is calculated as follows:

[0042] The preset historical time range in this embodiment uses 2 months or more than two months, that is, obtaining mobile phone signaling data in the research area with a time span of more than 2 months. It can be known that the longer the time duration, the more accurate the research on historical data will be, which will be more conducive to future road traffic prediction.

[0043] For each user, we first calculate the starting point and end point of the mobile phone signaling data. If the timestamp of the starting point is in the corresponding 5-minute time aggregation window, the starting point of the user is attributed to the corresponding time aggregation window. Then, we spatially associate the longitude and latitude coordinates of the starting point and end point of the user with the boundary of the traffic cell to determine the traffic cell where the starting point and end point of the user are located. Finally, we count the total number of OD people in each time aggregation window as the OD data of the traffic cell in the time aggregation window.

[0044] Then, based on the time aggregation window, the number of mobile phone users on each road is counted as the road signaling traffic data within the time aggregation window.

[0045] In step 2, the detailed operations of spatial association using QGIS software are as follows:

[0046] 1) Use QGIS software to spatially associate the OD data and road network data between traffic communities:

[0047] The road network data includes road network nodes;

[0048] Import the traffic zone vector layer and road network node layer into QGIS software.

[0049] Calculate the distance from the geometric center point of each traffic zone to the road network node. For each traffic zone, find the nearest road network node and establish a corresponding relationship table between the traffic zone and the road network node.

[0050] The corresponding relationship table is used to replace the starting and ending areas in the OD data between traffic areas with the corresponding starting and ending road network nodes to obtain the OD matrix at the road network node level.

[0051] Specifically, the "Vector"-"Data Management Tool"-"Distance Matrix" tool in the QGIS software is used to calculate the distance from the geometric center point of each traffic zone to the road network node.

[0052] 2) Use QGIS software to spatially associate road network data with road signaling flow data:

[0053] Import the road network node layer and road signaling flow data table into QGIS software, convert the road signaling flow data table into a spatial point layer, and use the spatial connection tool of QGIS software to associate the road flow point layer with the road network node layer;

[0054] Use the field calculator to assign the flow value of the flow point to the road signaling flow field of the corresponding road segment network node.

[0055] After correlating the three data, GIS is used to identify key traffic nodes and flow bottlenecks, and the method of generating the corrected section flow of each road on the road network combined with the spatial correlation results includes:

[0056] Based on the OD data of the traffic zone, the OD flow data of each road on the road network in each time aggregation window is obtained;

[0057] The difference between the signaling flow data and the OD flow data of each road on the road network in each time aggregation window is calculated to obtain the error vector, which is used to correct the section flow and generate a dynamic section flow matrix.

[0058] (I) Based on the OD data of traffic cells, the OD flow data of each road in each time aggregation window on the road network is obtained. The specific implementation is as follows: First, the road network data is converted into a directed graph, where road intersections represent nodes, road segments represent directed or undirected edges connecting nodes, and each edge is attached with length and travel time attributes. For each pair of traffic cells (such as cell A and cell B), the Dijkstra algorithm is used to calculate their shortest path on the constructed directed graph. The steps of the Dijkstra algorithm are:

[0059] Step 1), initialize the distance from the starting point to each node, the distance from the starting point to itself is 0, and the distance from the starting point to other nodes is infinite.

[0060] Step 2) Select the node with the shortest current distance and mark it as visited.

[0061] Step 3) Update the distance from the starting point to all unvisited nodes. If the distance to a certain node through the current node is shorter, update the distance of the node.

[0062] Step 4), repeat steps 2) and 3) until the target node is visited or all nodes are visited.

[0063] After calculating the shortest paths between all cells, the OD volume between cells is allocated to the corresponding shortest paths. After the allocation, it is aggregated according to the time aggregation window (5 minutes) to obtain the OD flow data of each road on the road network within the time aggregation window.

[0064] (ii) Calculate the difference between the signaling traffic data and the OD traffic data of each road on the road network in each time aggregation window to obtain the error vector, use the error vector to correct the road section traffic, and generate a dynamic road section traffic matrix. After correction, a dynamic road section traffic matrix is ​​generated that can reflect the actual road conditions and take into account the accuracy of demand-supply matching. This matrix integrates the traffic characteristics of both time and space dimensions, and includes both the complete OD demand distribution between cells and the endogenous path selection probability calculated based on mobile phone signaling data. The specific process of correction is as follows:

[0065] (1) For each road, calculate the difference between its signaling traffic data a and OD traffic data b. The main difference between a and b is that the signaling traffic data a is obtained by reconstructing the user's travel trajectory through mobile phone signaling data, matching each signaling trajectory point to the nearest road, and then counting the number of mobile phone users on each road based on the time aggregation window. The OD traffic data b is obtained by first calculating the starting and ending points of each user's mobile phone signaling data, determining the traffic area in which it is located, and then counting the travel volume between the areas within each time aggregation window. Subtracting the two gives the error term , where i, j represent the starting and ending nodes of the road, and t represents the time aggregation window.

[0066] (2) Establish error terms based on historical data Linear regression model with signaling traffic data a: ,in, and are the regression coefficients to be estimated, respectively representing the correlation degree of the error term to the signaling traffic and the bias term. The values ​​of these two parameters can be calculated by linear regression based on historical data. Here, historical data refers to the signaling traffic data a and OD traffic data b of each road in each time aggregation window with a time span of more than 2 months calculated in step (1), as well as the difference between the two (error term).

[0067] (3) Using the linear relationship model between the error term and the signaling traffic data in step (2), the section traffic data is corrected to obtain the corrected section traffic c: .

[0068] (4) Repeat steps (1) to (3) for all road segments and all time aggregation windows to obtain the corrected traffic data C(t) of all dynamic road segments, where t represents the time aggregation window.

[0069] In step 3, the method of training the LSTM model is specifically implemented as follows: modeling and predicting the road traffic data C(t) corrected by historical data.

[0070] (1) Arrange the traffic values ​​of each road in different time aggregation windows obtained in the previous step in the order of timestamps to form a time series. Here, the step length of the time series is the same as the time aggregation window, such as 5 minutes disclosed in this embodiment.

[0071] (2) Normalize the traffic data and map the values ​​to the interval [0, 1]. The purpose of normalization is to map data of different dimensions and scales to a common interval, thereby improving the numerical stability and convergence speed of model training. The normalization method adopts the Min-Max method. The specific steps are: put the traffic time series of each section into their respective corresponding sets, and the traffic data of different sections cannot be mixed together. Then calculate the minimum value of the traffic set of each section. and maximum value , for each flow value , calculate the normalized flow value , the formula is: .

[0072] The normalized time series data is divided into training set, validation set and test set according to the chronological order, with the first 2 months of data as the training set, the subsequent 15 days of data as the validation set, and the last 15 days of data as the test set. The total is 3 months of data, and this rule can also be followed in other feasible solutions. For example, for 6 months of data, 4 months can be used as the training set, the last month as the validation set, and the last month as the test set. Among them, the time step is set to 5 minutes, the input step is 6, that is, each input sample contains the flow value of the past 30 minutes (6 time steps), and the prediction step is 1, that is, each output sample corresponds to the flow value of the next 5 minutes (1 time step).

[0073] (3) Use the Keras deep learning framework with Python to build an LSTM model, select the Adam loss function, train the data prepared above, set the batch size (batch_size) to 32, and the number of training rounds (epochs) to 100.

[0074] (4) After the model is trained, the trained LSTM model is used to predict the traffic demand data of each road in the entire road network in the future time period. The specific implementation is as follows:

[0075] Obtain data with the same input step length as the model from the most recent historical data as the input of the model. Perform the same normalization processing on the input data as the training data, convert it into the dimension and scale used in model training, input the normalized data into the trained model, obtain the flow prediction value for the next time slice, denormalize the prediction value, and restore it to the dimension and scale of the original data. The denormalization adopts the following formula: Original flow value = Prediction value (max - min) + min.

[0076] Then, the predicted flow value of the next time slice is added to the end of the input data, and the earliest time step data in the input data is removed to form new input data. The previous prediction process is repeated, and the updated input data is used to predict the flow of the next time step. Continue to iterate until the flow prediction values ​​of all time steps of the next day after the current time are predicted. If the flow prediction value of a more distant future time period needs to be predicted, continue to iterate and calculate according to the above method.

[0077] Embodiment 2:

[0078] A traffic guidance method is disclosed, which predicts the traffic demand data of each road in the entire road network in the future time period through a traffic demand prediction method, and outputs a traffic guidance strategy, including the following steps:

[0079] Regulate and guide travel demand, generate road congestion warning information and detour information, and release detour information through variable traffic signs and intelligent traffic signal systems;

[0080] Evaluate the regulation effect and optimize the LSTM model parameters and guidance strategy based on the evaluation results.

[0081] Among them, the traffic capacity threshold of each road is determined based on historical data and traffic management experience. When the predicted traffic volume exceeds the threshold, it is determined that there is a probability of congestion in the road section exceeding the threshold during the corresponding time period. Road congestion warning information is generated based on the predicted traffic demand data of each road in the entire road network in the future time period.

[0082] The method for issuing diversion information includes: issuing congestion warning information and detour diversion information to drivers on designated sections through mobile phone push channels according to mobile phone signaling data, so as to guide drivers to arrange travel reasonably and avoid congested sections.

[0083] For some smart driving vehicles, messages can also be pushed to the vehicle's smart terminal.

[0084] As a preferred solution, the calculation method of the detour information includes: combining the road network topology and real-time traffic status, selecting the upstream and downstream nodes of the predicted congested section as the starting and ending points of the detour, searching for alternative detour paths, and giving priority to paths with sufficient capacity and shorter travel time during the search process. The upstream is the road that the vehicle passes before entering the congested section, and the downstream is the road that the vehicle will pass after leaving the congested section. The upstream and downstream nodes are the starting and ending points of the congested section.

[0085] The congestion warning information includes warning sections, warning periods, and congestion levels. Warning sections are the names of roads where congestion is predicted to occur; warning periods are the start and end times of predicted congestion; and congestion levels are based on the degree to which the predicted traffic exceeds the threshold, dividing congestion into different levels, including mild congestion, moderate congestion, severe congestion, etc.

[0086] Evaluate the control effect, optimize the LSTM model parameters and guidance strategy based on the evaluation results, continuously improve the system response capability, and improve traffic management efficiency. The specific process is as follows:

[0087] Effect evaluation: Compare the measured road condition data with the predicted results, calculate the errors of various prediction indicators (such as flow rate and speed), statistically analyze the actual flow changes of detour routes during the diversion period, analyze the implementation effect of diversion measures, select typical areas, and compare and analyze the traffic operation conditions before and after the implementation of regulation.

[0088] Iterative optimization: Based on the evaluation results, identify the shortcomings of the prediction model, such as insufficient applicability in special weather conditions and holidays, and make targeted algorithm improvements and parameter adjustments. Optimize the path search strategy, introduce factors such as driver preferences and congestion risks based on travel time and distance, and improve the acceptability and stability of detour routes. Improve the way information is released to improve the timeliness and coverage of warning information.

[0089] Embodiment 3:

[0090] A computer device is disclosed, comprising a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement a traffic demand forecasting method as described in Example 1, or a traffic guidance method as described in Example 2.

[0091] Exemplarily, the computer program can be divided into one or more modules / units, one or more modules / units are stored in the memory and executed by the processor, and the I / O interface transmission of data is completed by the input interface and the output interface to complete the present invention. One or more modules / units can be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device. The computer device can be a computing device such as a desktop computer, a notebook, a handheld computer, and a cloud server. The computer device may include, but is not limited to, a memory, a processor. Those skilled in the art can understand that this embodiment is only an example of a computer device and does not constitute a limitation on the computer device. It can include more or fewer components, or combine certain components, or different components. For example, the computer device may also include an input, a network access device, a bus, etc. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor, or the processor may be any conventional processor or the like.

[0092] The memory can be an internal storage unit of a computer device, such as a hard disk or memory of a computer device. The memory can also be an external storage device of a computer device, such as a plug-in hard disk equipped on a computer device, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Furthermore, the memory can also include both the internal storage unit of a computer device and an external storage device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data in an output device. The aforementioned storage media include various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories ROM, random access memories RAM, disks or optical disks.

Claims

1. A traffic demand forecasting method, characterized in that: The steps include: Divide the study area into traffic zones, obtain the mobile phone signaling data within the study area within a preset historical time range, and calculate the OD data and road signaling flow data between the traffic zones within a preset time aggregation window based on the mobile phone signaling data; Obtain the road network data of the study area, and spatially associate the OD data, road signaling flow data, and road network data between traffic cells; Use GIS to identify key traffic nodes and flow bottlenecks, and combine spatial association results to generate corrected section flow for each road on the road network; According to the corrected road section traffic, the traffic values ​​of each road in different time aggregation windows are arranged in chronological order according to timestamps to form time series data, and the time series data is used to train the LSTM model; the trained LSTM model is used to predict the traffic demand data of each road in the entire road network in the future time period.

2. A traffic demand prediction method as claimed in claim 1, characterized in that: The spatial association method includes spatially associating the OD data and road network data between traffic communities through QGIS software: The road network data includes road network nodes; Import the traffic zone vector layer and road network node layer into QGIS software. Calculate the distance from the geometric center point of each traffic zone to the road network node. For each traffic zone, find the nearest road network node and establish a corresponding relationship table between the traffic zone and the road network node. The corresponding relationship table is used to replace the starting and ending areas in the OD data between traffic areas with the corresponding starting and ending road network nodes to obtain the OD matrix at the road network node level.

3. A traffic demand prediction method as claimed in claim 2, characterized in that: The spatial association method includes spatially associating the road network data and the road signaling flow data through QGIS software: Import the road network node layer and road signaling flow data table into QGIS software, convert the road signaling flow data table into a spatial point layer, and use the spatial connection tool of QGIS software to associate the road flow point layer with the road network node layer; Use the field calculator to assign the flow value of the flow point to the road signaling flow field of the corresponding road segment network node.

4. A traffic demand forecasting method as claimed in claim 3, characterized in that: The method of using GIS to identify key traffic nodes and traffic bottlenecks and combining spatial association results to generate corrected road section traffic for each road on the road network includes: Based on the OD data of the traffic zone, the OD flow data of each road on the road network in each time aggregation window is obtained; The difference between the signaling flow data and the OD flow data of each road on the road network in each time aggregation window is calculated to obtain the error vector, which is used to correct the section flow and generate a dynamic section flow matrix.

5. A traffic demand prediction method as claimed in claim 1, characterized in that: The method also includes the following steps: a method for normalizing time series data, comprising the following steps: Put the traffic time series of each road section into their respective sets and normalize the time series data; The LSTM model is trained using the normalized time series data.

6. A traffic guidance method, characterized in that: Predicting the traffic demand data of each road in the entire road network in the future time period by the traffic demand prediction method described in any one of claims 1 to 5, and outputting a traffic guidance strategy, comprises the following steps: Regulate and guide travel demand, generate road congestion warning information and detour information, and release detour information through variable traffic signs and intelligent traffic signal systems; Evaluate the regulation effect and optimize the LSTM model parameters and guidance strategy based on the evaluation results.

7. A traffic guidance method as claimed in claim 6, characterized in that: The method for generating road congestion warning information comprises: Based on historical data and traffic management experience, the capacity threshold of each road is determined. When the predicted traffic volume exceeds the threshold, the probability of congestion occurring on the road section exceeding the threshold in the corresponding time period is determined. Road congestion warning information is generated based on the predicted traffic demand data of each road in the entire road network in the future time period.

8. A traffic guidance method as claimed in claim 6, characterized in that: The method for issuing diversion information through an intelligent traffic signal system comprises: Through mobile phone push channels, road congestion warning information and detour information are issued to drivers on designated roads based on mobile phone signaling data.

9. A traffic guidance method as claimed in claim 8, characterized in that: The method for calculating the detour diversion information includes: combining the road network topology and real-time traffic status, selecting the upstream and downstream nodes of the predicted congested road section as the starting and ending points of the diversion, searching for alternative detour paths, and giving priority to paths with surplus capacity and / or shorter travel time during the search process.

10. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the traffic demand prediction method as described in any one of claims 1 to 5, or the traffic guidance method as described in any one of claims 6 to 9.

Images