TWI859031B - System and method for predicting road vehicular traffic and computer program product thereof - Google Patents

Abstract

A system and a method for predicting road vehicular traffic and a computer program product performing the same are provided. A method performed at a server includes: receiving a plurality of mobile signaling data and road upstream-downstream relationship data thereby generating road information corresponding to each of the mobile signaling data, analyzing each of the mobile signaling data and its corresponding road information using a machine learning method to generate a vehicle type corresponding to each of the mobile signaling data, thereby estimating road real-time vehicular traffic and road upstream-downstream real-time vehicular traffic, and inputting the road real-time vehicular traffic and the road upstream-downstream vehicular traffic, road information corresponding to each of the mobile signaling data, real-time weather data, and real-time time data into a neural network model to output future road vehicular traffic through the neural network model.

Description

System, method and computer program product for predicting road traffic volume

This case is about a road traffic flow prediction technology, especially a system, method and computer program product for predicting future traffic flow on a road using mobile signaling data.

In order to improve traffic congestion, existing technologies provide traffic flow information for reference by road users and traffic management personnel.

A common way to estimate traffic flow is to use a vehicle detector (VD) or an eTag Detector, which can be installed on or beside the road to sense passing vehicles and transmit the sensing data back to the traffic control center. Another estimation method is to use artificial intelligence models to identify objects in the image, thereby calculating their quantity or location to analyze the traffic flow and characteristics on the road.

In addition, existing technologies can use estimated traffic flow data to generate predicted traffic flow. However, the cost of building and maintaining vehicle detectors is high and they are easily affected by environmental factors. Although artificial intelligence image recognition can provide diverse traffic flow information, its disadvantage is that it requires a large amount of image data and computing resources, and may also be affected by image quality and angle.

Therefore, how to accurately estimate the current road traffic volume and predict the future road traffic volume is an issue that needs to be solved.

In order to solve the above problems and other problems, this case proposes a system, method and computer program product for predicting road traffic volume.

The system disclosed in this case for predicting road traffic volume includes: a data receiving module, which receives a plurality of mobile signaling data, real-time weather data, real-time time data, and road upstream and downstream relationship data; a signaling data road segment corresponding module, which generates road segment information corresponding to each mobile signaling data according to each mobile signaling data received by the data receiving module and the road upstream and downstream relationship data; a road segment real-time traffic volume estimation module, which uses a machine learning method to analyze each mobile signaling data received by the data receiving module and the road segment information corresponding to each mobile signaling data generated by the signaling data road segment corresponding module, so as to generate a vehicle type corresponding to each mobile signaling data, and then According to the upstream and downstream relationship data of the road section, the plurality of mobile signaling data, the road section information corresponding to each of the mobile signaling data, and the vehicle type corresponding to each of the mobile signaling data, the real-time traffic flow of the road section and the real-time traffic flow of the upstream and downstream sections of at least one road section are estimated; and the road section future traffic flow prediction module inputs the road section information corresponding to each of the mobile signaling data generated by the signaling data road section corresponding module, the real-time traffic flow of the road section and the real-time traffic flow of the upstream and downstream sections estimated by the road section real-time traffic flow estimation module, the real-time weather data, and the real-time time data into the neural network model, so as to output the future traffic flow of the road section of at least one road section through the neural network model.

The method disclosed in this case for predicting the traffic volume of a road section is executed on the server side, and the method includes: receiving a plurality of mobile signaling data, real-time weather data, real-time time data, and upstream and downstream relationship data of the road section; generating road section information corresponding to each mobile signaling data according to each mobile signaling data and the upstream and downstream relationship data of the road section; using a machine learning method to analyze each mobile signaling data and the road section information corresponding to each mobile signaling data to generate a vehicle type corresponding to each mobile signaling data; and generating a vehicle type corresponding to each mobile signaling data according to the upstream and downstream relationship data of the road section. The method comprises: estimating the real-time traffic flow of at least one road segment and the real-time traffic flow of upstream and downstream roads based on the traffic relationship data, the plurality of mobile signaling data, the road segment information corresponding to each of the mobile signaling data, and the vehicle type corresponding to each of the mobile signaling data; and inputting the road segment information corresponding to each of the mobile signaling data, the real-time traffic flow of at least one road segment and the real-time traffic flow of upstream and downstream roads, the real-time weather data, and the real-time time data into a neural network model, so as to output the future traffic flow of at least one road segment through the neural network model.

The computer program product disclosed in this case for predicting the traffic volume on a road section is loaded into a computer to execute the method disclosed in this case for predicting the traffic volume on a road section.

The computer-readable recording medium disclosed in this case for predicting the traffic flow of a road section stores instructions, and can use a computing device or a computer to execute the computer-readable recording medium through a processor and/or a memory, so as to execute the method for predicting the traffic flow of a road section disclosed in this case when executing the computer-readable recording medium.

In one embodiment, the neural network model is trained with long short-term memory (LSTM) using historical weather data, historical time data, historical road section lane number data, historical road section type data, historical road section upstream and downstream traffic flow data, and historical road section traffic flow data, wherein the road section information corresponding to each of the mobile signaling data, the road section real-time traffic flow and upstream and downstream road section real-time traffic flow of at least one road section, the real-time weather data, and the real-time time data are input into the trained neural network model, so that the neural network model outputs the future traffic flow of the road section of at least one road section.

In one embodiment, each of the mobile signaling data includes a rate, and the road segment real-time traffic flow estimation module uses a support vector machine method to analyze the road segment information corresponding to each of the mobile signaling data and the rate of each of the mobile signaling data to generate the vehicle type of each of the mobile signaling data, and then estimates the road segment real-time traffic flow and upstream and downstream road segment real-time traffic flow of at least one road segment based on the plurality of mobile signaling data, the road segment information corresponding to each of the mobile signaling data, the vehicle type corresponding to each of the mobile signaling data, and the upstream and downstream relationship data of the road segment, wherein one of the parameters of the road segment real-time traffic flow and upstream and downstream road segment real-time traffic flow of at least one road segment is the sum of the mobile signaling of the vehicle type being a bus or a car.

In one embodiment, each of the mobile signaling data includes a location, and the signaling data segment mapping module maps the location of each of the mobile signaling data to a plurality of grids to confirm the corresponding grids of each of the mobile signaling data, and then calculates the distance between each of the mobile signaling data and each of the road segments in the corresponding grids, so as to sort the calculated distances, and then take the road segment with the smallest distance as the road segment where each of the mobile signaling data is located, so as to obtain the road segment information including the road name, the number of lanes in the road segment, and/or the road segment type of the road segment where each of the mobile signaling data is located, wherein the signaling data stop points in the plurality of mobile signaling data are eliminated, and the signaling data stop points are those where the mobile signaling data of a block reaches a specific amount and does not move within a specific time.

In one embodiment, the system for predicting road traffic volume disclosed in this case further includes a network transmission module, wherein the data receiving module receives the plurality of mobile signaling data to transmit to the signaling data road segment corresponding module, the data receiving module also receives the real-time weather data and the real-time time data, and transmits them from the signaling data road segment corresponding module to the road segment future traffic volume prediction module via the road segment real-time traffic volume estimation module, and the network transmission module transmits the road segment future traffic volume of at least one road segment from the server end to the display end.

By using the system, method, computer program product, and computer-readable recording medium disclosed in this case for predicting road traffic flow, it is possible to use mobile phone signaling to estimate the real-time traffic flow of a road section, the real-time traffic flow of upstream and downstream sections, and then predict the future traffic flow of the road section without setting up hardware such as vehicle detectors or electronic tag detectors. Therefore, it can help road users or traffic control centers understand road conditions, thereby accurately providing diversions or taking corresponding evacuation measures.

1: Display terminal

11: Future traffic flow display module

2: Server side

21: Data receiving module

22: Signaling data segment corresponding module

23: Road section real-time traffic flow estimation module

24: Historical database

25: Road section future traffic flow prediction module

251:Neural network model

26: Network transmission module

S1~S5: Steps

S21~S23: Steps

S31~S32: Steps

S41~S43: Steps

Figure 1 is a schematic diagram of the architecture of an implementation example of the system for predicting road traffic volume in this case.

Figure 2 is a schematic diagram of the implementation process of the method for predicting road traffic volume in this case.

Figure 3 is a schematic diagram of an implementation process of mapping mobile signaling data to road sections in the method for predicting road section traffic volume in this case.

Figure 4 is a schematic diagram of an implementation process of estimating the real-time traffic flow of a road section in the method for predicting the traffic flow of a road section in this case.

Figure 5 is a schematic diagram of an implementation process of predicting the future traffic flow of a road section in the method for predicting the traffic flow of a road section in this case.

FIG6 is a schematic diagram of a specific embodiment of the system and method for predicting road traffic volume in this case.

The following specific examples are used to illustrate the implementation of this case. People familiar with this technology can easily understand the other advantages and effects of this case from the content disclosed in this article. The structures, ratios, sizes, etc. shown in the attached figures of this manual are only used to match the content disclosed in the manual for people familiar with this technology to understand and read, and are not used to limit the conditions under which this case can be implemented. Therefore, any modification, change or adjustment should still fall within the scope of the technical content disclosed in this case without affecting the effects and purposes that can be achieved by this case.

As used herein, the terms "include", "comprising", "having", "containing" or any other variations thereof are intended to cover a non-exclusive inclusion. Unless otherwise indicated, singular forms such as "a", "an", "the" may also be used in the plural, and "or", "and/or" and the like may be used interchangeably.

Please refer to Figure 1, which is a schematic diagram of the implementation example of the system for predicting road traffic volume in this case. The system in this case includes a data receiving module 21 on the server side 2, a signaling data road section corresponding module 22, a road section real-time traffic volume estimation module 23, a historical database 24, a road section future traffic volume prediction module 25, and a network transmission module 26.

The data receiving module 21 is used to receive mobile signaling data (such as mobile signaling location, mobile signaling time, mobile signaling moving speed, etc.) from the base station, real-time weather data at the time of mobile signaling (its type is such as sunny day, rainy day), and real-time time data at the time of mobile signaling (its type is such as off-peak or peak), wherein the real-time weather data and real-time time data are collectively referred to as real-time data, which can be obtained from the meteorological bureau data and time zone standard time of the area where the mobile signaling is located preset by the system. In one embodiment, the mobile device can have connection records when making a call, when the base station is transferred, when the network is used, or it can also return signals at fixed times. The data receiving module 21 also receives upstream and downstream relationship data of the road section.

The signaling data segment mapping module 22 is used to generate corresponding segment information according to the mobile signaling data (such as the mobile signaling location) and the upstream and downstream relationship data of the segment. In one embodiment, the signaling data segment mapping module 22 maps the mobile signaling location to multiple grids to confirm which corresponding grid the mobile signaling falls into, and then calculates the straight-line distance between the mobile signaling location and each segment in the corresponding grid. After sorting these distances, the segment with the smallest distance is taken as the segment where the mobile signaling is located, and the segment information such as the road name, the number of lanes in the segment, and the segment type (such as expressway, expressway, urban segment, scenic area segment) of the segment is queried.

The road segment real-time traffic flow estimation module 23 is used to analyze the road segment upstream and downstream relationship data, mobile signaling data (such as mobile signaling moving speed) received by the data receiving module 21, and the road segment information generated/queried/obtained by the signaling data road segment corresponding module 22 using a machine learning method or algorithm to estimate the road segment real-time traffic flow and the road segment upstream and downstream real-time traffic flow. In one embodiment, the road segment real-time traffic flow estimation module 23 adopts a support vector machine (SVM). Machine) method or algorithm is used to analyze the moving speed of the mobile signaling to generate the vehicle type of the mobile signaling, such as rail transportation, bus, car, motorcycle, and walking/bicycle. Then, according to the road section where the mobile signaling data is located, the time of the mobile signaling data, the vehicle type of the mobile signaling data, and the data volume of the mobile signaling, the real-time traffic flow of the road section at the current time is calculated. The real-time traffic flow of the road section can be formed by summing up the number of vehicles whose vehicle type is bus and car. That is to say, excluding rail transportation (such as train or MRT), motorcycle, and walking/bicycle, the mobile signaling whose vehicle type is bus or car will be counted and summed up as the parameter of the real-time traffic flow of the at least one road section and the real-time traffic flow of the upstream and downstream sections. Similarly, for the mobile signaling data on the upstream and downstream sections of the road section, the road section real-time traffic flow estimation module 23 estimates the upstream and downstream real-time traffic flow of the road section according to the amount of mobile signaling data and the type of vehicle.

In another embodiment, other supervised learning models with relevant learning methods or algorithms that can perform classification and regression analysis on data may be used, such as decision trees, random forest, k -NN, linear regression, perceptron, etc.

The historical database 24 is used to store historical data, including upstream and downstream road segment relationship data, historical mobile signaling data collected by the data receiving module 21 (which may be less than or more than three months), historical road segment information corresponding to the historical mobile signaling data generated by the signaling data road segment corresponding module 22, historical road segment traffic flow data estimated by the road segment real-time traffic flow estimation module 23 based on the historical mobile signaling data, and historical road segment upstream and downstream traffic flow data. The traffic flow data, that is, the current real-time traffic flow of the road section and the real-time traffic flow of the upstream and downstream sections estimated by the real-time traffic flow estimation module 23 of the road section become the subsequent historical traffic flow data of the road section and the historical upstream and downstream traffic flow data of the road section. In addition, the historical data further includes the historical weather data (sunny or rainy) corresponding to the historical traffic flow data of the road section and the historical time data (peak or off-peak) corresponding to the historical traffic flow data of the road section.

The road section future traffic flow prediction module 25 is used to input the road section information (such as road name, number of road section lanes, road section type) generated by the signaling data road section corresponding module 22, the road section real-time traffic flow estimated by the road section real-time traffic flow estimation module 23 and the real-time traffic flow of the upstream and downstream of the road section, and the real-time data received by the data receiving module 21 (such as real-time weather data, real-time time data), into the neural network model 251 to output the future traffic flow of the road section through the neural network model 251. In one embodiment, the neural network model 251 is first trained with historical data stored in the historical database 24, for example, historical road section information corresponding to historical mobile signaling data, historical road section traffic data and historical road section upstream and downstream traffic data, historical weather data corresponding to historical road section traffic data (sunny or rainy), and historical time data corresponding to historical road section traffic data (peak or off-peak) are used to perform neural network training, such as long short-term memory (LSTM) training, to train the neural network model 251. In another embodiment, a recurrent neural network, for example, may be used to train the neural network model, or, when the amount of information on certain road sections is relatively small, meta-learning may be used as the neural network for training.

In other words, the upstream and downstream relationship data of the road segment and the plurality of mobile signaling data received by the server 2 are first processed by the signaling data road segment corresponding module 22 and the road segment real-time traffic flow estimation module 23, and then these processed data (such as historical road segment information corresponding to historical mobile signaling data, historical road segment traffic flow data and historical road segment upstream and downstream traffic flow data, historical weather data corresponding to historical road segment traffic flow data, and historical time data corresponding to historical road segment traffic flow data) are used as training data for the neural network model 251. As for the input data of the neural network model 251 (such as road section information, real-time traffic flow of the road section and real-time traffic flow of upstream and downstream of the road section, real-time weather data, and real-time time data), some basic missing value, abnormal value, duplicate value, standardization, data encoding, and feature scaling processing are performed. Among them, standardization is to convert the value range of numerical values into between 0 and 1, and data encoding is to encode non-numeric data (such as 0 to 288), and then input it into the embedding layer of the neural network.

In addition, after receiving a large amount of mobile signaling data, distributed processing can be performed first and the information of the user's stay point can be analyzed regularly. The signaling data of the stay point (which is not needed in the future) can be eliminated in real time to reduce the amount of data to be processed later. For example, the stay point is the user's workplace or home, which is the most signaling point in a certain area during the day and night for at least 3 weeks and reaches at least 300 signaling data. Since the static workplace and home in the same area will not carry out driving activities, the signaling data of the stay point can be filtered out first to reduce the amount of data to be processed. In addition, in one embodiment, the analysis time is from 9:00 am to 5:00 pm and from 9:00 pm to 5:00 am on weekdays to analyze the user's workplace and home. Usually, the time that users stay at the workplace and home is much longer than the time when they are stuck in the same area. Therefore, the place with the most signaling data in the same area will not be the traffic jam location.

The network transmission module 26 is used to transmit the future traffic flow of the road segment predicted by the future traffic flow prediction module 25, the real-time traffic flow of the road segment estimated by the real-time traffic flow estimation module 23, and the real-time traffic flow of upstream and downstream of the road segment to the future traffic flow display module 11 of the display terminal 1. In one embodiment, the future traffic flow of the road segment, the real-time traffic flow of the road segment, and the real-time traffic flow of upstream and downstream of the road segment can be presented by the future traffic flow display module 11 (such as an application (APP)) or a web page of the display terminal 1. In other words, a road network map can be presented on the mobile device of the road user, which displays the name of the road section where the road user's mobile device is located, the current time, the real-time traffic volume of the road section, the upstream and downstream traffic volume of the road section, the future traffic volume of the road section (possibly 15 minutes or 30 minutes later), and the future traffic volume of the upstream and downstream of the road section. The display method of traffic volume can be represented by numbers, levels, and colors.

Please refer to Figure 2, which is a schematic diagram of an implementation process of the method for predicting road traffic volume in this case. The method includes steps S1 to S5, which can be executed by the server 2 shown in Figure 1.

Step S1, receiving mobile signaling data and real-time data. In one embodiment, the mobile signaling data includes location, time, speed, etc. In detail, the location of the signal user or the service range of the base station can be defined through the signal connection between the mobile device and the base station. Generally speaking, the mobile device mainly has connection records when making calls, base station transfers, or network use, or returns signals at fixed times. In one embodiment, the real-time data includes real-time weather data corresponding to the time of the mobile signaling data (its type is such as sunny or rainy) and real-time time data (its type is such as peak or off-peak). In one embodiment, the upstream and downstream relationship data of the road section can also be received, specifically a road network data.

Step S2, generate road segment information corresponding to the mobile signaling data. The detailed method of corresponding the mobile signaling data to the road segment is shown in steps S21 to S23 of Figure 3.

Step S21, mapping the location of the mobile signaling data to a plurality of grids to confirm the corresponding grid of the mobile signaling data; Step S22, calculating the distance between the mobile signaling data and each road section in the corresponding grid to sort the calculated distances; Step S23, taking the road section with the smallest distance as the road section where the mobile signaling data is located to obtain the road section information including the road name, the number of lanes in the road section and/or the road section type of the road section where the mobile signaling data is located. Specifically, a region can be divided into multiple grids, and the corresponding grid to which each mobile signaling data belongs is confirmed based on its location. Then, the straight-line distance between each road in the corresponding grid and the mobile signaling data is calculated. After sorting by distance, the road section with the smallest distance is taken as the road section where the mobile signaling data is located, so as to obtain the road name, the number of lanes in the road section (e.g. 1, 2) and the road section type (e.g. national highway, urban area, scenic area section).

In one embodiment, signaling data retention points can be eliminated according to the location of each mobile signaling data, and the signaling data retention point is when the mobile signaling data of a block reaches a specific amount and does not move within a specific time. For example, a block has the most mobile signaling data and the number reaches about 300, and these mobile signaling data are almost motionless during working hours (such as: 9:00~17:00) or rest time (such as 21:00~05:00) for about three weeks, which should be for residence or office work. Then these signaling data retention points can be eliminated first to gradually reduce the amount of data to be processed later.

Step S3, based on the mobile signaling data, the road section information, and the upstream and downstream relationship data of the road section, the real-time traffic flow of the road section and the real-time traffic flow of the upstream and downstream of the road section are estimated. The detailed method for estimating the real-time traffic flow of the road section is shown in steps S31~S32 of Figure 4.

Step S31, using machine learning methods or algorithms, such as support vector machines (SVMs) The mobile signaling data is analyzed by a support vector machine (SVM) method or algorithm to generate the vehicle type of the mobile signaling data; step S32, according to the vehicle type of the mobile signaling data (such as rail transportation, bus, car, motorcycle, and pedestrian/bicycle) and the road section information (such as road name, number of lanes in the road section, and road section type), the real-time traffic flow of the road section is estimated, wherein the real-time traffic flow of the road section is the sum of the number of vehicles with the vehicle type of bus and car, that is, rail transportation (such as train or MRT), motorcycle, and pedestrian/bicycle, or residents/office workers are excluded, and the mobile signaling with the vehicle type of bus or car will be counted and summed up as the parameters of the real-time traffic flow of the at least one road section and the real-time traffic flow of the upstream and downstream sections.

In one embodiment, support vector machines can handle small sample, nonlinear, high dimensional and local minimum problems. In addition, SVM is a linear classifier and can also perform effective nonlinear classification. In another embodiment, other supervised learning models with related learning methods or algorithms that can classify and regress data can also be used, such as decision trees, random forests, k -NN, linear regression, perceptrons, etc.

In step S4, the road segment information obtained in step S2, the real-time traffic flow of the road segment and the real-time traffic flow of the upstream and downstream of the road segment estimated in step S3, the real-time weather data and real-time time data obtained in step S1 are input into the neural network model to output the future traffic flow of the road segment through the neural network. The detailed method for predicting the future traffic flow of the road segment is shown in steps S41 to S43 of Figure 5.

Step S41, receiving historical weather data, historical time data, historical road section lane number data, historical road section type data, historical road section upstream and downstream traffic flow data, and historical road section traffic flow data; Step S42, using historical weather data, historical time data, historical road section lane number data, historical road section type data, historical road section upstream and downstream traffic flow data, and historical road section traffic flow data to perform neural network training, such as long short-term memory (LSTM) training, to train a neural network model; Step S43, inputting road section information, real-time traffic flow of the road section, real-time weather data, time data, and upstream and downstream relationship data of the road section into the trained neural network model, so that the neural network model outputs the future traffic flow of the road section. In one embodiment, a recurrent neural network, for example, can be used to train the neural network model, or, when the amount of information on certain road sections is relatively small, meta-learning can be used as the neural network for training.

Step S5, transmit the future traffic flow of the road section to the display end. More specifically, transmit the real-time traffic flow of the road section, the real-time traffic flow of the upstream and downstream of the road section, and the future traffic flow of the upstream and downstream of the road section to the display end.

In addition to one or more of the above embodiments, the present invention provides a computer program product that executes one or more of the above methods after the program is loaded into a computer. In addition, the computer program (product) can be stored in a recording medium or directly transmitted and provided on the Internet, that is, the computer program (product) is a thing that carries a computer-readable program and is not limited to an external form. The computer includes but is not limited to an electronic device with a processor, such as a server, etc. In addition, the present case also provides a computer-readable recording medium, which is applied to a computing device or a computer having a processor and/or a memory, and the computer-readable recording medium stores instructions, and the computing device or the computer can execute the computer-readable recording medium through the processor and/or the memory, so as to execute the above-mentioned method and/or content when executing the computer-readable recording medium. The computer-readable recording medium (such as a hard disk, a floppy disk, an optical disk, a USB flash drive) stores the computer program (product). In one embodiment, the computer-readable recording medium is a non-transitory computer-readable recording storage medium.

Please refer to FIG. 6 , which is a screen presented by the future traffic flow display module 11 (e.g., an application (APP)) of the display terminal 1 (e.g., a mobile device of a passerby) of FIG. 1 , i.e., a road network map, which displays the road name of the road section where the mobile device of the passerby is located (e.g., Zhongzheng 2nd Road), the current time (e.g., 2023/7/27/16:05:00), the real-time traffic flow of the road section (e.g., Zhongzheng 2nd Road traffic flow 50), and the future traffic flow of the road section (e.g., estimated 70 after 15 minutes), where the numbers can represent but are not limited to the number of cars.

In other words, this case uses the mobile signaling data and the upstream and downstream relationship data of the road section to analyze the real-time traffic flow of each road section. Of course, the real-time traffic flow of the upstream and downstream sections of each road section is the real-time traffic flow of each other's upstream and downstream sections, so this case does not require vehicle detectors or electronic tag detectors. In addition, with real-time data such as real-time weather data (weather type is sunny or rainy), real-time time data (time type is off-peak or peak), as well as the road section information such as the number of lanes and road section type where the mobile signaling data is located, the trained neural network model is used to predict the future traffic flow of the road section where the mobile signaling data is located. In addition, the estimated real-time traffic flow of each road section, along with the weather, time, road section and other data at the time, is also used as training data for the neural network model.

In another specific embodiment, if the traffic flow data of Xinyi Road Section 4 at 15:30 on 2023/11/22 is to be predicted at 16:00 on 2023/11/22, the current real-time traffic flow of Xinyi Road Section 4, the real-time traffic flow of the upstream section Xinyi Road Section 3, the real-time traffic flow of the downstream section Xinyi Road Section 5, the current weather information of Taipei City obtained from the Meteorological Bureau, the number of lanes of Xinyi Road Section 4 2, the road type of Xinyi Road Section 4 is an urban road, and the predicted time 16:00 can be used as parameters of the neural network model, and the neural network model can be used to predict the traffic flow data of Xinyi Road Section 4 at 16:00 on 2023/11/22.

In summary, the system, method, computer program product, and computer-readable recording medium disclosed in this case for predicting road traffic flow are combined with traffic information (speed, travel time) web pages, display billboards, and APPs. In addition to obtaining necessary traffic information or routes, it can also help road users know which road sections will have increased traffic flow in the future and can be diverted in advance. In addition, the personnel of the traffic control center can also make corresponding traffic control diversion and evacuation measures after knowing the future traffic flow. Therefore, this case can use mobile phone mobile signaling data to predict future traffic flow without setting up hardware such as vehicle detectors VD (vehicle detectors) or eTags to calculate traffic flow, achieving low cost, not easily affected by environmental factors, no need for a large amount of image data and computing resources, and not affected by image quality or angle.

The above embodiments are only illustrative of the effects of this case, and are not intended to limit this case. Anyone familiar with this technology can modify and change the above embodiments without violating the spirit and scope of this case. Therefore, the scope of protection of this case should be as listed in the scope of the patent application described below.

S1~S5: Steps

Claims (11)

A system for predicting road traffic volume, comprising: Data receiving module, receiving multiple mobile signaling data, real-time weather data, real-time time data, and upstream and downstream relationship data of road sections; The signaling data segment corresponding module generates segment information corresponding to each mobile signaling data according to each mobile signaling data received by the data receiving module and the upstream and downstream relationship data of the segment; The road segment real-time traffic flow estimation module uses a machine learning method to analyze each of the mobile signaling data received by the data receiving module and the road segment information corresponding to each of the mobile signaling data generated by the signaling data road segment corresponding module to generate a vehicle type corresponding to each of the mobile signaling data, and then estimates the road segment real-time traffic flow and upstream and downstream road segment real-time traffic flow of at least one road segment based on the upstream and downstream relationship data of the road segment, the plurality of mobile signaling data, the road segment information corresponding to each of the mobile signaling data, and the vehicle type corresponding to each of the mobile signaling data; and The road section future traffic flow prediction module inputs the road section information corresponding to each mobile signaling data generated by the signaling data road section corresponding module, the road section real-time traffic flow and upstream and downstream road section real-time traffic flow estimated by the road section real-time traffic flow estimation module, the real-time weather data, and the real-time time data into the neural network model, so as to output the road section future traffic flow of the at least one road section through the neural network model. The system as described in claim 1, wherein the neural network model is trained with historical weather data, historical time data, historical road section lane number data, historical road section type data, historical road section upstream and downstream traffic flow data, and historical road section traffic flow data for long-term and short-term memory, wherein the road section information corresponding to each of the mobile signaling data, the road section real-time traffic flow and upstream and downstream road section real-time traffic flow of at least one road section, the real-time weather data, and the real-time time data are input into the trained neural network model, so that the neural network model outputs the future traffic flow of the road section of at least one road section. The system as described in claim 1, wherein each of the mobile signaling data includes a rate, and the road segment real-time traffic flow estimation module uses the support vector machine method to analyze the road segment information corresponding to each of the mobile signaling data and the rate of each of the mobile signaling data to generate the vehicle type of each of the mobile signaling data, and then estimates the road segment real-time traffic flow and upstream and downstream road segment real-time traffic flow of at least one road segment based on the plurality of mobile signaling data, the road segment information corresponding to each of the mobile signaling data, the vehicle type corresponding to each of the mobile signaling data, and the upstream and downstream relationship data of the road segment, wherein one of the parameters of the road segment real-time traffic flow and upstream and downstream road segment real-time traffic flow of at least one road segment is the sum of the mobile signaling of the vehicle type being a bus or a car. The system as described in claim 1, wherein each of the mobile signaling data includes a location, and the signaling data segment mapping module maps the location of each of the mobile signaling data to a plurality of grids to confirm the corresponding grids of each of the mobile signaling data, and then calculates the distance between each of the mobile signaling data and each of the road segments in the corresponding grids, sorts the calculated distances, and then takes the road segment with the smallest distance as the road segment where each of the mobile signaling data is located, so as to obtain the road segment information including the road name, the number of lanes in the road segment, and/or the road segment type of the road segment where each of the mobile signaling data is located, wherein the signaling data stop points in the plurality of mobile signaling data are eliminated, and the signaling data stop points are those where the mobile signaling data of a block reaches a specific amount and does not move within a specific time. The system as described in claim 1 further includes a network transmission module, wherein the data receiving module receives the plurality of mobile signaling data to transmit to the signaling data segment corresponding module, and the data receiving module also receives the real-time weather data and the real-time time data, so as to be transmitted from the signaling data segment corresponding module to the segment future traffic flow prediction module via the segment real-time traffic flow estimation module, and the network transmission module transmits the segment future traffic flow of at least one segment to the display terminal. A method for predicting traffic flow on a road section includes: Receive multiple mobile signaling data, real-time weather data, real-time time data, and upstream and downstream relationship data of road sections; According to each action signaling data and the upstream and downstream relationship data of the road section, the road section information corresponding to each action signaling data is generated; Analyze each mobile signaling data and the road segment information corresponding to each mobile signaling data using a machine learning method to generate a vehicle type corresponding to each mobile signaling data; Estimate the real-time traffic volume of at least one road section and the real-time traffic volume of upstream and downstream sections based on the upstream and downstream relationship data of the road section, the plurality of mobile signaling data, the road section information corresponding to each of the mobile signaling data, and the vehicle type corresponding to each of the mobile signaling data; and The road segment information corresponding to each of the mobile signaling data, the real-time traffic flow of the at least one road segment and the real-time traffic flow of the upstream and downstream road segments, the real-time weather data, and the real-time time data are input into the neural network model, so as to output the future traffic flow of the at least one road segment through the neural network model. The method as described in claim 6, wherein the step of inputting the road segment information corresponding to each of the mobile signaling data, the real-time traffic flow of the at least one road segment and the real-time traffic flow of the upstream and downstream road segments, the real-time weather data, and the real-time time data into the neural network model to output the future traffic flow of the at least one road segment through the neural network model includes: Performing long-term and short-term memory training with historical weather data, historical time data, historical road section lane number data, historical road section type data, historical road section upstream and downstream traffic flow data, and the historical road section traffic flow data to train the neural network model; and Input the road segment information corresponding to each of the mobile signaling data, the real-time traffic flow of the at least one road segment and the real-time traffic flow of the upstream and downstream road segments, the real-time weather data, and the real-time time data into the trained neural network model, so as to output the future traffic flow of the at least one road segment through the neural network model. The real-time traffic flow of the at least one road section and the real-time traffic flow of the upstream and downstream sections are used as the historical traffic flow data of the road section and the historical upstream and downstream traffic flow data of the road section to train the neural network model. The method as described in claim 6, wherein each of the mobile signaling data includes a rate, and the step of using a machine learning method to analyze each of the mobile signaling data and the road segment information corresponding to each of the mobile signaling data to generate a vehicle type corresponding to each of the mobile signaling data includes: Analyze the road segment information corresponding to each mobile signaling data and the speed of each mobile signaling data using a support vector machine method to generate the vehicle type of each mobile signaling data; and According to the plurality of mobile signaling data, the road segment information corresponding to each of the mobile signaling data, the vehicle type corresponding to each of the mobile signaling data, and the upstream and downstream relationship data of the road segment, the real-time traffic volume of the road segment and the real-time traffic volume of the upstream and downstream sections of the at least one road segment are estimated, Among them, one of the parameters of the real-time traffic flow of the at least one road section and the real-time traffic flow of the upstream and downstream sections is the sum of the mobile signaling of the vehicle type being a bus or a car. The method as described in claim 6, wherein each of the mobile signaling data includes a location, and the step of generating the road segment information corresponding to each of the mobile signaling data based on each of the mobile signaling data includes: Mapping the location of each mobile signaling data to a plurality of grids to confirm the corresponding grid of each mobile signaling data; Calculate the distance between each mobile signaling data and each road segment in the corresponding grid to sort the calculated distances; and The road section with the shortest distance is taken as the road section where each mobile signaling data is located, so as to obtain the road section information including the road name, the number of lanes in the road section, and/or the road section type of the road section where each mobile signaling data is located. Among them, the signaling data retention point in the plurality of mobile signaling data is eliminated, and the signaling data retention point is the mobile signaling data of a block reaching a specific amount and no movement within a specific time. The method as described in claim 6 further includes: receiving the plurality of mobile signaling data, the real-time weather data, the real-time time data, and the upstream and downstream relationship data of the road segment from the server end; and transmitting the future traffic flow of the at least one road segment from the server end to the display end. A computer program product that executes the method described in any one of claim 6-10 after being loaded into a computer.

Images