CN112991719B

CN112991719B - Traffic congestion prediction method and system based on congestion portrait

Info

Publication number: CN112991719B
Application number: CN202110118624.3A
Authority: CN
Inventors: 马雪峰
Original assignee: Beijing Azel Technology Development Co ltd
Current assignee: Nanoxin Technology (Beijing) Co.,Ltd.
Priority date: 2021-01-28
Filing date: 2021-01-28
Publication date: 2022-05-24
Anticipated expiration: 2041-01-28
Also published as: CN112991719A

Abstract

The invention discloses a traffic jam prediction method and system based on a jam portrait, wherein the method comprises the following steps: acquiring traffic data, and acquiring congestion image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the congestion image data reflects a road condition state of the traffic bottleneck point; according to the congestion image data of the traffic bottleneck point, determining a congestion image characteristic parameter of the traffic bottleneck point; and carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management. The method can predict the traffic data based on the congestion image, the obtained prediction result can be used for active traffic guidance management, and the police output basis can be determined according to the prediction result, so that basis is provided for signal optimization timing and the like, and basis is provided for travel route planning.

Description

Traffic congestion prediction method and system based on congestion portrait

Technical Field

The invention relates to the technical field of traffic congestion prediction, in particular to a traffic congestion prediction method and system based on a congestion portrait.

Background

The existing technologies capable of predicting traffic congestion in real time are mainly divided into three categories, one is simulation prediction carried out by dynamic traffic distribution based on macroscopic traffic demand data; secondly, traffic jam prediction based on big data analysis; and thirdly, microscopic simulation prediction based on real-time traffic data. The current traffic jam prediction mode basically has the defects of low prediction accuracy, insufficient real-time performance of a prediction result, dependence on a large amount of sample data, difficulty in meeting the prediction requirement of the existing traffic data acquisition equipment, single type of acquired data, low data precision and difficulty in ensuring the prediction accuracy.

Thus, there is a need for improvements and enhancements in the art.

Disclosure of Invention

The invention provides a traffic jam prediction method and system based on a jam portrait, and aims to solve the problems that the current traffic jam prediction mode in the prior art is basically low in prediction accuracy, insufficient in real-time performance of prediction results, and needs to rely on a large amount of sample data, the current traffic data acquisition equipment is difficult to meet the prediction requirements, the acquired data type is single, the data accuracy is low, and the prediction accuracy is difficult to ensure.

In order to solve the technical problems, the technical scheme adopted by the invention is as follows:

in a first aspect, the present invention provides a traffic congestion prediction method based on a congestion profile, wherein the method includes:

acquiring traffic data, and acquiring congestion image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the congestion image data reflects a road condition state of the traffic bottleneck point;

according to the congestion image data of the traffic bottleneck point, determining a congestion image characteristic parameter of the traffic bottleneck point;

and carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management.

In one implementation, the obtaining traffic data and obtaining congestion image data of a traffic bottleneck point in the traffic data according to the traffic data includes:

acquiring holographic video data acquired by traffic data detection equipment, and taking the holographic video data as the traffic data;

and inputting the traffic data into a preset congestion image model, and determining a traffic bottleneck point in the traffic data and congestion image data corresponding to the traffic bottleneck point through the processing of the congestion image model.

In one implementation, the congestion representation model is created by:

when the road condition is in an oversaturated traffic state, acquiring traffic jam parameters, wherein the traffic jam parameters comprise traffic bottleneck types, road structures and traffic flow characteristics;

and constructing the congestion portrait model according to the traffic congestion parameters.

In one implementation, the obtaining the traffic congestion parameter includes:

acquiring on-site traffic investigation data, or acquiring traffic real-time data acquired based on a sensor;

and determining the traffic jam parameters according to the traffic research data or the traffic real-time data.

In one implementation, the determining a congestion image characteristic parameter of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point includes:

determining a congestion type corresponding to the congestion image data according to the congestion image data of the traffic bottleneck point;

and determining congestion image characteristic data of the traffic bottleneck point according to the congestion type.

In one implementation manner, if the congestion type is a signal light intersection vehicle saturated congestion, the determining congestion image characteristic data of the traffic bottleneck point according to the congestion type includes:

acquiring vehicle access requirements of the traffic bottleneck point and the congestion delay characteristics of two adjacent times;

and calculating to obtain the congestion image characteristic data of the traffic bottleneck point according to the vehicle access demand or the congestion delay characteristics of the two adjacent times.

In one implementation, the calculating congestion image feature data of the traffic bottleneck point according to the vehicle access demand or the congestion delay characteristics of the two adjacent times includes:

determining a dynamic vehicle entrance flow rate, a dynamic vehicle exit flow rate and a congestion period of the traffic bottleneck point according to the vehicle entrance and exit requirements;

calculating to obtain the congestion image characteristic data according to the dynamic vehicle entrance flow rate, the dynamic vehicle exit flow rate and the congestion period;

alternatively, the first and second electrodes may be,

determining road section length data and road section speed limit data according to the congestion delay characteristics of the two adjacent times;

and calculating to obtain the congestion image characteristic data according to the road section length data and the road section speed limit data.

In one implementation, if the congestion type is a road construction congestion, the determining congestion image feature data of the traffic bottleneck point according to the congestion type includes:

acquiring a construction road section, and acquiring the total dynamic demand time of the construction road section in a peak time period, the running time difference of the construction road section before and after construction and the average flow speed of a traffic bottleneck point in the peak time period;

and calculating to obtain the congestion image characteristic data according to the dynamic demand total time, the running time difference and the average flow speed.

In one implementation, if the congestion type is a traffic event congestion, the determining congestion image feature data of the traffic bottleneck point according to the congestion type includes:

acquiring a congestion road section of a traffic event, and acquiring total congestion time of the congestion road section, running time difference of the congestion road section before and after the congestion event, average flow speed of a traffic bottleneck point and a congestion period;

and calculating to obtain the congestion image characteristic data according to the total congestion time, the running time difference, the average flow speed and the congestion period.

In a second aspect, the present invention further provides a traffic congestion prediction system based on a congestion image, the system comprising:

the traffic jam image data acquisition module is used for acquiring traffic data and acquiring the traffic jam image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the traffic jam image data reflects the road condition state of the traffic bottleneck point;

the congestion image characteristic parameter acquisition module is used for determining congestion image characteristic parameters of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point;

and the congestion prediction module is used for performing congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, and the prediction result is used for traffic management.

In a third aspect, an embodiment of the present invention further provides an intelligent device, where the intelligent device includes a memory, a processor, and a traffic congestion prediction program based on a congestion map, stored in the memory and executable on the processor, and when the processor executes the traffic congestion prediction program based on a congestion map, the method for predicting traffic congestion based on a congestion map according to any one of the above aspects is implemented.

In a fourth aspect, the embodiments of the present invention further provide a computer-readable storage medium, on which a traffic congestion prediction program based on a congestion profile is stored, and when being executed by a processor, the method for predicting traffic congestion based on a congestion profile realizes the steps of the method for predicting traffic congestion based on a congestion profile in any one of the above aspects.

Has the advantages that: compared with the prior art, the invention provides a traffic jam prediction method based on a jam image, which comprises the steps of obtaining traffic data, and obtaining jam image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the jam image data reflects the road condition state of the traffic bottleneck point; according to the congestion image data of the traffic bottleneck point, determining a congestion image characteristic parameter of the traffic bottleneck point; and carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management. Therefore, the traffic data can be predicted based on the congestion image, the obtained prediction result can be used for active traffic guidance management, the basis for the traffic police to give out the police can be determined according to the prediction result, the basis is provided for the optimized timing of signals and the like, and the basis is provided for the planning of the travel route.

Drawings

Fig. 1 is a flowchart of a traffic congestion prediction method based on a congestion image according to an embodiment of the present invention.

Fig. 2 is a flowchart of acquiring congestion image data in a traffic congestion prediction method based on a congestion image according to an embodiment of the present invention.

Fig. 3 is a congestion image curve in a traffic congestion prediction method based on a congestion image according to an embodiment of the present invention.

Fig. 4 is a flowchart of determining a congestion profile characteristic parameter in a traffic congestion prediction method based on a congestion profile according to an embodiment of the present invention.

Fig. 5 is a schematic block diagram of a traffic congestion prediction apparatus based on a congestion image according to an embodiment of the present invention.

Fig. 6 is a schematic block diagram of an internal structure of an intelligent device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and effects of the present invention clearer and clearer, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The existing technologies capable of predicting traffic congestion in real time are mainly divided into three categories, one is simulation prediction carried out by dynamic traffic distribution based on macroscopic traffic demand data; secondly, traffic jam prediction based on big data analysis; thirdly, microscopic simulation prediction based on real-time traffic data is specifically as follows:

the macroscopic traffic demand is based on the historical OD (the space-time distribution relation of traffic travel) of the statistical analysis of the traffic cells, and the historical OD is utilized for simulation. The method comprises the steps of firstly, establishing a road network model, wherein the road network model comprises a plurality of layers of traffic networks which are the same as the actual road network, such as road network nodes (nodes), links (road sections), zones (traffic cells) and the like, and calibrating road network attributes, such as parameters of link directions, types, lane numbers, traffic capacity and the like, so that the method is huge in workload and needs to consume a lot of working hours to complete; secondly, a traffic trip demand matrix, namely an OD matrix, is established based on a traffic cell, the OD matrix is the space-time distribution of traffic demand trips, the data can be obtained by carrying out big data statistical analysis on intersection license plate number snapshot data of a traffic police, the data can also be statistically analyzed through crowdsourcing data of a mobile phone, no matter which mode is adopted, as many data samples as possible need to be obtained, however, no matter which mode is adopted, full sample data cannot be obtained; thirdly, dynamically distributing based on the calculated demand data, putting the demand data into a road network according to space-time distribution, selecting a path based on a model, performing hundreds of iteration attempts, and striving to achieve the balance of road network vehicles, thereby realizing the simulation process of the whole road network; fourthly, when congestion prediction is needed, interference conditions of a road network, such as road construction, traffic accidents and the like, are implanted into the road network, vehicles in the road network are interfered, congestion queuing is formed, and influences of the interference conditions on the road network, such as delay, queuing length and the like, can be further calculated.

But based on the road network prediction technology, the previewing of various scenario schemes and decisions, especially the sudden traffic incident, can be carried out, and not only can the' what is happening? "," what is about to happen immediately? "also know what, if any, what will be the case? How should we do? ", which is the greatest advantage of this approach. Of course, the method has the disadvantages that the workload is huge in the early stage of implementation, the accurate effect is difficult to achieve, and the floor spreading difficulty is high. The simulation deduction is only directed at the calculation of travel demands, is not a real situation on the road, and only explains where the travel demands are to be from at a certain moment, and as for the travel demands, which are a driving state on the road, the simulation and prediction cannot be accurately performed. Therefore, the defects of the method are avoided by utilizing the advantages of the method, and the prediction deduction based on the historical OD data is just suitable for the application in static aspects such as planning design, construction scheme evaluation and the like. Although the method has good application, the analysis models of various software manufacturers are basically dependent on the precision of historical OD, and if the OD travel demand is inaccurate, the simulation prediction basically has no precision, which is the maximum pain point of the method.

And, the technical shortcoming of simulation prediction based on macroscopic traffic demand data:

1) the investment is huge at the early stage: because the design of the road network model is needed in the early stage of the project, and the traffic demand analysis, calibration and the like of the whole road network are needed, the workload required to be invested in the early stage of the project is huge, the realization time is long, the realization is difficult, and the speed of traffic development is difficult to keep up.

2) The calculation speed of the prediction result is slow: because the scale of the related road network is large, hundreds of iterations are needed for dynamic balance calculation, a certain time is needed to output a prediction result even if parallel calculation is adopted, and the prediction result can be output in at least 10 minutes at present, so the prediction real-time performance is poor.

3) The fineness is insufficient: due to the adoption of macroscopic traffic demand calculation, the congestion prediction refinement degree of the congestion bottleneck point is insufficient, including the congestion prediction of each intersection or bottleneck point in a road network, such as the congestion will need to last for a long time, the congestion will generate delay time and the like. The method can only carry out macroscopic prediction on the regional traffic situation, and is difficult to realize microscopic prediction.

The big data content in the existing traffic jam prediction method based on big data analysis mainly comprises the following steps: the mobile phone crowdsourcing and floating car data mainly comprises mobile phone crowdsourcing data (crowdsourcing big data formed by mobile phone GPS positioning data) and the dripping data mainly comprises floating car data (floating car big data formed by GPS positioning data on a dripping car). By the sample big data, the real-time running state of a part of vehicles in the road network can be obtained, and the real-time running state comprises running speed, road section vehicle sample density and other parameters. If the running speed and the running position of each vehicle on the road can be known, the distribution condition of the vehicles in the road network and the real-time running state of the road network can be easily reproduced. The method generally adopts big data analysis, can roughly obtain historical congestion condition data such as congestion time period, congestion time length, queuing length and the like, and can also calculate the time required by congestion when a vehicle reaches a congested road section based on real-time travel time sample data. However, the method is a defect of the method that mobile phone sample data is collected, not all vehicles use APP, and only a part of vehicles can be covered, so that all traffic demands cannot be accurately acquired, and when traffic efficiency of a congested road section changes, such as an inductive signal control bottleneck road section, the 'how long a congestion will last', 'how many vehicles will be affected by congestion' and the like cannot be accurately predicted. It is currently unlikely that every vehicle will have GPS data uploaded, and this is the pain point of this method, even if it is hundredths and grand. The method has the advantages that the large data of the samples are crowdsourced by the mobile phone, real-time road network operation conditions can be simulated, development situations under future normal traffic conditions can be predicted based on large data analysis of historical data, however, the method does not research the total traffic demand amount, only researches the samples, so that the analysis on traffic problems is not fine enough, the complexity of traffic jam is added, and once an accident such as induction control, traffic accidents, construction and the like occurs, the development situations of traffic are unknown, which is the biggest defect of the method.

And the technical shortcomings of traffic congestion prediction based on big data analysis are as follows:

1) the accuracy is not sufficient: since the method adopts partial sample analysis, and since the samples are difficult to be 100% in quantity, the demand cannot be accurately estimated, the time for which the congestion will last cannot be accurately predicted, and how many vehicles are affected by the congestion.

2) The prediction result has insufficient real-time property: since the method analyzes real-time big data, the problem of data lag is inevitable, the prediction result is lagged, the prediction result is used for real-time traffic management, and the application of the lag is such as signal control, active guidance, traffic jam dispersion and the like. The effect inevitably deteriorates due to the management of the hysteresis.

3) The prediction results have limited application: the maximum data value of the method is the travel time and the real-time vehicle speed of sample data, so that the travel time prediction effect is good, the congestion prediction aiming at the bottleneck point is not fine enough, for example, the congestion lasts for a long time, the number of vehicles is influenced by the congestion, and the complexity of traffic congestion is added, once an emergent traffic influence event occurs, such as induction control adjustment, traffic accidents, construction and the like, the development situation of traffic is unknown, and the total demand amount cannot be controlled.

The existing microscopic simulation prediction technology based on real-time traffic data utilizes a microscopic simulation model and a driving behavior model to simulate the actual situation according to the real-time traffic data and predict the traffic development situation according to microscopic traffic demands and input simulation interference items. The technology calibrates parameters such as traffic distribution parameters and vehicle running speed in a microscopic simulation process by utilizing real-time data detected by a sensor to calibrate a simulated real-time traffic state, and predicts a traffic development situation through time acceleration running based on traffic demands. Compared with macroscopic demand simulation prediction, the method has the advantages that the microscopic degree is high, the display effect is better, and the method is the greatest advantage. However, the accuracy of the simulation prediction of this technology completely depends on the microscopic driving model and the traffic disturbance parameters input in advance, and the driving behavior of the vehicle is difficult to be accurately modeled, because the simulated vehicle walking way and the real walking way have great difference, so the simulation prediction result in this way is often poor in accuracy, which is the biggest defect of this method. In addition, the micro simulation needs to predict the OD data of a small area, and the accuracy of the OD data also affects the prediction accuracy. In addition, the method is only suitable for simulation of small areas or road sections, and large-scale road network simulation is slow in speed and cannot achieve rapid prediction, so that the method is usually only used for traffic simulation prediction of a single intersection.

And the technical shortcomings of microscopic simulation prediction based on real-time traffic data are as follows:

1) simulation results are difficult to keep consistent with reality: because the method adopts a general vehicle driving model, the difference and complexity of the driving technology are difficult to accurately describe by the model, and because the driving technology is too micro and the micro model is difficult to realize, the simulation prediction result of the method has larger deviation from the actual result.

2) Depending on the OD accuracy of the simulation area: the micro simulation also needs small-area OD data to better predict the traffic situation, the accuracy of the OD data becomes a condition for restricting the micro simulation, and when the accuracy of the OD data is not good, the simulation prediction result is poor.

3) The real-time performance is poor: since the method analyzes real-time traffic flow data and signal control data, the problem of data lag inevitably occurs, so that the prediction result is lagged, the prediction result is used for real-time traffic management, and the method has some lags, such as signal control, active guidance, traffic jam evacuation and the like. The effect inevitably deteriorates due to the management of the hysteresis.

Therefore, the current traffic jam prediction mode basically has the defects of low prediction accuracy, insufficient real-time performance of a prediction result, dependence on a large amount of sample data, difficulty in meeting the prediction requirement of the existing traffic data acquisition equipment, single type of acquired data, low data accuracy and difficulty in ensuring the prediction accuracy.

However, in the prior art, the traffic data acquisition devices can be basically divided into three categories, specifically as follows:

collection devices and sources for macroscopic traffic demand data (OD data). At present, OD data mainly come from two types, one type is license plate number data, and the other type is mobile phone crowdsourcing data. The license plate data can be realized by using a snapshot camera, the snapshot data of a gate electronic police of a public security traffic police is usually utilized, the data only has license plate number and time information, and no positioning and speed data of a single vehicle target, and the data is just the defect that cross section flow data is unavailable due to frequent missed shooting of gate equipment, and the data has the advantages of fully mining the data value of the existing equipment and reducing hardware investment; the method is characterized in that mobile phone crowdsourcing data is obtained by collecting position and time data of each traveler by utilizing APP on a mobile phone, and each traffic participant is identified by ID of the mobile phone, so that the data can accurately position the real-time position and the traveling speed of each target, the track of each target can be obtained by big data analysis, and finally macroscopic demand OD data can be obtained. In addition, the macroscopic traffic demand needs to realize accurate dynamic distribution, and accurate road section traffic flow data including flow, vehicle speed, travel time and the like are needed, while the data generally come from sensors installed on the road, the types of the sensors are more, the video detection technology and the radar detection technology are popular at present, and no matter which technology, the video detection technology and the radar detection technology are traffic data for section detection, the traffic parameters such as section flow, vehicle speed, vehicle type, head time and distance are generally provided, the accuracy can reach 90%, and the data can be used for the calibration of the traffic demand. The drawbacks of this device are mainly: the acquired data is single in type and not good enough in data accuracy, and due to the fact that the data is not fully utilized, maintenance is basically lacked, and the data integrity degree is poor.

The method aims at the collection equipment and source of the crowdsourcing data of the mobile phone. As described above, the mobile phone crowdsourcing data is that the APP on the mobile phone is used to collect the position and time data of each traveler, and each traffic participant is identified by the ID of the mobile phone, so that the data can accurately locate the real-time position and the traveling speed of each target, and the track of each target can be obtained through big data analysis.

Apparatus and source of micro-simulation prediction data for real-time traffic data

The microcosmic simulation predicted data mainly uses OD data of a small area, and intersection steering flow data and section flow data to calibrate dynamically distributed parameters in real time, so as to meet the actual situation. The existing equipment generally adopts a camera and microwave radar detection to detect data, the equipment is generally only suitable for a free flow traffic state, the data type is single, however, due to urban traffic jam, the detection accuracy is greatly reduced, so that the simulation prediction accuracy is limited, and the defect of the equipment is that. Since the microscopic simulation adopts a microscopic analysis method, the requirement on the real-time performance of signal lamp control data is very high, so that the real-time signal lamp control data is also required, and the data acquisition is relatively complex.

It can be seen that the existing data acquisition equipment has poor precision: the microscopic simulation prediction technology mainly utilizes real-time data of a traffic flow sensor, the quality of the data can also influence the prediction result, the existing traffic flow sensor basically has a single data detection function, only can be used as section traffic flow data or steering data, and few sensors with various data fusion are available, particularly road section travel time data, intersection driving section vehicle arrival data, intersection leaving section traffic flow data, queuing data, occupancy data and the like; in addition, most of the existing devices have better detection accuracy under the free flow traffic state, and the detectors basically fail and have poorer accuracy when the traffic is congested or even in a super-saturated state.

In order to solve the problems of the prior art, the embodiment provides a traffic congestion prediction method based on a congestion image, and in specific implementation, the embodiment acquires traffic data, and obtains congestion image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the congestion image data reflects a road condition state of the traffic bottleneck point; according to the congestion image data of the traffic bottleneck point, determining a congestion image characteristic parameter of the traffic bottleneck point; and carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management. Therefore, the traffic data can be predicted based on the congestion image, the obtained prediction result can be used for active traffic guidance management, the basis for the police to give out the police can be determined according to the prediction result, the basis is provided for the optimal timing of signals and the like, and the basis is provided for the travel route planning.

The embodiment is specially designed for the bottleneck point of road traffic jam, particularly realizes the second-level overall process prediction of the traffic jam development situation in the oversaturated traffic state, and provides data support for various traffic jam management decisions. The embodiment can be used for congestion bottleneck points in various supersaturated traffic states, including traffic congestion bottlenecks formed by facilities or structural features of roads, such as scenes of signal lamp control intersections, road entrance and exit ramps and the like, and traffic congestion bottlenecks caused by various traffic events, such as scenes of road construction, traffic accidents, traffic control and the like; the congestion types can be classified into long-term bottleneck congestion, short-term bottleneck congestion and sudden bottleneck congestion. Aiming at various bottleneck points of different congestion types, under the oversaturated traffic state, prediction of a second-level speed and a congestion overall process is realized based on changes of real-time traffic demands, active traffic control is realized by utilizing edge calculation, and active traffic guidance is realized through a vehicle-road cooperative communication channel. Firstly, the method aims at the prediction of a traffic jam bottleneck point, and is not regional prediction or whole road network prediction; secondly, the research of the embodiment is that the traffic state is oversaturated, not normal, or the traffic jam is common; thirdly, in the embodiment, the second-level rapid congestion prediction is realized, the traffic congestion prediction is realized instantly, and the congestion prediction is not realized through a long simulation or iteration process; fourthly, in the embodiment, to realize the prediction of the whole congestion process, instead of predicting for several minutes, dozens of minutes or more in the future, the prediction with time as a basic axis is realized, and the embodiment only predicts the whole process from the beginning of the traffic congestion to the ending of the congestion and does not predict other traffic state time periods; fifthly, the embodiment predicts the whole process of the traffic jam to be generated, and does not predict when the traffic jam will occur; sixth, this embodiment predicts how long the congestion "will" last based on the change of real-time demand, how many vehicles delay the whole congestion process, how much total delay time is, and how long it will take for vehicles arriving at the end of the line at each moment to pass through the congested road segment.

Exemplary method

The traffic congestion prediction method based on the congestion image in the embodiment may be applied to an intelligent device, and in a specific implementation, as shown in fig. 1, the traffic congestion prediction method based on the congestion image specifically includes the following steps:

step S100, obtaining traffic data, and obtaining congestion image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the congestion image data reflects a road condition state of the traffic bottleneck point.

In this embodiment, first, traffic data is acquired, and then congestion image data corresponding to the traffic data is acquired according to the traffic data, and the congestion image data can be used for predicting the state of a traffic intersection. In order to more accurately determine the traffic jam image data, the embodiment constructs a jam image model, and the jam image model can process the traffic data to obtain the jam image data in the traffic data. In this embodiment, after the traffic data is obtained, the traffic data is input to a preset congestion image model, so as to obtain congestion image data of a traffic bottleneck point in the traffic data. The traffic jam image data reflects the road condition of the traffic bottleneck point, so that traffic prediction can be performed according to the traffic image data in the subsequent steps. In this embodiment, the traffic bottleneck point is a congestion bottleneck point in various oversaturated traffic states, such as a congestion point caused by too many vehicles at an intersection. Generally, traffic congestion is mostly caused by structural bottleneck points in a road network or temporary traffic bottlenecks, such as traffic accidents and road construction. Therefore, the analysis of the congestion mechanism of the traffic bottleneck is more and more meaningful, especially in the super-saturated traffic state.

In one implementation, as shown in fig. 2, the step S100 specifically includes the following steps:

s101, acquiring holographic video data acquired by traffic data detection equipment, and taking the holographic video data as the traffic data;

step S102, inputting the traffic data into a preset congestion image model, and determining a traffic bottleneck point in the traffic data and congestion image data corresponding to the traffic bottleneck point through processing of the congestion image model.

In specific implementation, the congestion image is a very effective analysis method designed for the traffic bottleneck congestion, extracted according to parameters such as different traffic bottleneck types, road structures, traffic flow characteristics and the like, and has a targeted congestion characteristic model, and the advantages of the congestion characteristic model can be more prominent particularly in a super-saturated traffic state. By using the congestion images, the congestion development situation of the traffic bottleneck can be rapidly deduced, the time-varying delay time, the time-varying queuing length and the time-varying travel time are predicted, and the total delay, the average delay and the average travel time of the bottleneck are evaluated. FIG. 3 is a congestion pictureFunction curve of image, where mu is actual traffic capacity at exit of road bottleneck point, t₀、t₂、

Is the solution of the model function and respectively represents the queue generation time, the queue length longest time, the queue ending time, t₃The dissipation time is queued.

The type of the congestion image can be divided into 3 types according to the time length of the traffic bottleneck:

a long-term bottleneck jam image is a jam image which is established aiming at the bottleneck of traffic jam which is generated every day. Such traffic bottlenecks can have a significant impact on traffic congestion over a long period of time. The congestion image can be further divided into a static image and a dynamic image according to whether the traffic parameter is changed. The static congestion image is determined by the road structure, and the traffic congestion condition is basically kept unchanged in a long time, such as the number of the entrance ramp, the exit ramp and the lane change points; dynamic congestion images are due to the fact that certain traffic parameters are constantly changing, such as the split green of a sensory controlled signal, which determines that the congestion images must also be dynamic.

The short-term bottleneck jam image is a jam image established aiming at a planned bottleneck which always generates traffic jam in a short time, such as road construction, and the influence time is short, namely days, and long, namely months. For the traffic bottlenecks, the congestion image can deduce the traffic congestion development situation after the bottleneck appears, and can predict and evaluate the influence degree of the bottleneck on the traffic before the bottleneck appears.

The sudden bottleneck congestion image is a congestion image which is established aiming at the traffic bottleneck caused by the sudden event, such as a traffic accident, traffic control and the like. The congestion image is created according to congestion data generated by historical events, and is classified according to event types and road types to create a congestion image library. Since the location of the occurrence of an abnormal traffic event is not determinable, similar representations from the representation library can be used for traffic impact prediction when events of similar conditions occur.

In order to improve the accuracy of traffic prediction, in this embodiment, a congestion image model is first constructed, and in specific implementation, a single congestion bottleneck point may be selected for the congestion image model in this embodiment, or a congestion image attribute parameter may be established on the basis of a road network model and associated with a corresponding road link (a road segment including at least one entrance). The traffic bottleneck point of the embodiment is a vehicle congestion point in an oversaturated traffic state. Therefore, when the road condition is in an oversaturated traffic state, traffic jam parameters are obtained, wherein the traffic jam parameters comprise traffic bottleneck types, road structures and traffic flow characteristics; and then constructing the congestion representation model according to the traffic congestion parameters. That is, the present embodiment divides the method of creating the congestion image model into two methods: firstly, on-site investigation and research of support data and establishment of a static portrait; secondly, the real-time data of the sensor is utilized to establish a dynamic portrait. For the congestion image caused by the traffic event, a congestion image library needs to be established by using historical event congestion data, certain accumulation is needed, and if no historical event congestion data exists, some parameters need to be estimated empirically for prediction. The support data of the congestion image model is flexible in practice, and may be cross-sectional traffic flow data or travel time data of a road segment. The data can be acquired by portable equipment, on-site investigation, and section traffic flow data or travel time data on the upstream and downstream of a traffic bottleneck; the installed traffic flow detection sensor can also be utilized to obtain dynamic traffic flow data, and a dynamic congestion image can be established. Although the congestion traffic image can be established by various data, the accuracy of the data is important, and the holographic video traffic data detection equipment is developed aiming at the requirement of a congestion image model to ensure the accuracy requirement of the data.

In specific implementation, the holographic video data acquired by the traffic data detection device is acquired, and the holographic video data is traffic jam parameters, where the traffic jam parameters include traffic bottleneck types, road structures, and traffic characteristics. In the prior art, when acquiring traffic data or traffic movement data, a single video detection technology, a section microwave detection technology, a geomagnetic detection technology and a multi-target radar detection technology are basically adopted. However, in any of the above detection techniques, the detection data is single except its own technical defects, and the data requirements of various congestion images cannot be completely satisfied, as follows:

single video detection techniques: because the video detection technology depends on the light condition, the detection data precision is good when the light condition is good, once the light condition is poor, for example, the detection result can be influenced under the conditions of dusk time period and low visibility, such as rainstorm, sand storm and the like, so the precision of the video detection technology can be seriously influenced by the severe weather condition.

Section microwave detection technology: because the microwave detection technology relies on the reflection of electromagnetic waves to detect vehicles, when traffic jams, higher big vehicles such as buses can easily shield nearby cars, and the electromagnetic waves cannot bypass the shielding, so that the shielded cars can be missed, and the shielding is the biggest defect of the section microwave detection technology.

Geomagnetism detection technology: as the geomagnetism is installed in a fixed-point ground mode and is drilled in the middle of a lane, the geomagnetism can work in a severe environment, wheels are rolled, exposed to high temperature, soaked in rainwater and the like, so that the geomagnetism is very easy to damage. Therefore, the geomagnetic detection has drawbacks in both the life span and the kind of detection data information.

The multi-target radar detection technology comprises the following steps: because the single multi-target radar detection technology can only provide tracking data of the vehicle, the data output is single, and the license plate number, the vehicle type and even the high-precision track can not be identified like a video. The detected vehicle data is almost perfect because the method of combining radar positioning scanning and Doppler velocity measurement is adopted, the vehicle running process can be tracked, and therefore the vehicle can not be influenced by vehicle shielding and weather conditions.

Therefore, the embodiment acquires the holographic video data acquired by the traffic data detection equipment, and uses the holographic video data as the traffic data or traffic jam data. The traffic data detection device in the embodiment integrates the structured video technology and the multi-target radar technology, and is designed in an integrated structure, wherein the video cameras are arranged on the upper portion, and the radars are arranged on the lower portion. The structural design can ensure that the horizontal X, Y coordinate origin of the video camera and the radar are at the same point, which is beneficial to the effective fusion of the two data; because the visual angles of the video camera and the radar are slightly different, the pitching angle adjustment of the two parts is independently completed; the whole holder structure can rotate horizontally and shake head left and right, so that the applicability of various installation conditions is ensured. The device can thus collect the following data: license plate number data, vehicle type flow data, steering flow data and section average speed data at the exit of the traffic bottleneck, and vehicle arrival data, queuing length data and vehicle type flow data at the upstream of the bottleneck. In addition, the shell of the camera adopts a double-layer sheet metal structure, the inner layer is made of 304 stainless steel, and the outer layer is made of aluminum alloy, so that the camera can resist rain and heat, and the stable work of the camera is ensured; the shell structure of the radar radiation cover adopts a compact design, and the consistency of radar signals is ensured by adopting mould processing.

Therefore, when the road condition is in the oversaturated traffic state, the traffic congestion parameters are acquired based on the traffic data detection equipment, and then the congestion image model can be constructed according to the traffic congestion parameters. After the congestion image model is constructed, the traffic data may be input into a preset congestion image model, and a traffic bottleneck point in the traffic data and congestion image data corresponding to the traffic bottleneck point are determined through processing of the congestion image model, so that traffic congestion prediction may be performed according to the congestion image data in subsequent steps.

And S200, determining a congestion image characteristic parameter of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point.

In this embodiment, when the congestion image data of the traffic bottleneck point is obtained, the congestion image characteristic parameter of the traffic bottleneck point can be determined according to the traffic congestion image data. The traffic jam image characteristic parameter is a key for predicting traffic jam, the road condition state of the traffic bottleneck point can be reflected according to the traffic jam characteristic parameter, and the traffic jam characteristic parameter can more accurately represent key information in the jam image data, such as the length of a jammed road section, the jam period and the like. Therefore, the traffic prediction can be more accurately carried out according to the traffic jam characteristic parameter.

In one implementation, as shown in fig. 4, the step S200 specifically includes the following steps:

step S201, determining a congestion type corresponding to congestion image data according to the congestion image data of the traffic bottleneck point;

and step S203, determining congestion image characteristic data of the traffic bottleneck point according to the congestion type.

In specific implementation, because the congestion image data of different congestion types are applied differently in congestion prediction, after the traffic image data is obtained, the congestion type corresponding to the congestion image data can be determined according to the congestion image data of the traffic bottleneck point; and then determining congestion image characteristic data of the traffic bottleneck point according to the congestion type.

In an implementation manner, if the congestion type is a signal lamp intersection vehicle saturated congestion, the embodiment can obtain the vehicle access demand of the traffic bottleneck point and the congestion delay characteristics of two adjacent times; and then calculating to obtain the congestion image characteristic data of the traffic bottleneck point according to the vehicle access demand or the congestion delay characteristics of the two adjacent times. Specifically, the embodiment determines a vehicle dynamic entrance flow rate, a vehicle dynamic exit flow rate and a congestion period of the traffic bottleneck point according to the vehicle entrance and exit demand; and calculating to obtain the congestion image characteristic data according to the dynamic vehicle entrance flow rate, the dynamic vehicle exit flow rate and the congestion period. Or, in this embodiment, according to the congestion delay characteristics of the two adjacent times, determining the length data of the road section and the speed limit data of the road section; and then calculating to obtain the congestion image characteristic data according to the road section length data and the road section speed limit data.

In particular, in a super-saturated state, signal control is usually embodied as overlong queuing length, excessive parking times and longer vehicle delay time, and in such a state, the purpose of optimizing traffic efficiency is the most important purpose. The application is used for predicting each entrance lane of the intersection and belongs to fine evaluation.

The method is mainly used for making a timing scheme of the single-point signal machine, improving the single-point signal control efficiency and providing prediction data support for regional optimization. The method is characterized in that the development situation of congestion is predicted by using a congestion image, a signal timing scheme in a super-saturated state is adjusted, and in the application, two methods can be adopted to calculate the characteristic parameters of the congestion image according to the data type, wherein one method adopts a dynamic inflow and outflow equation to calculate, and the other method adopts a time-varying delay equation to calculate. The specific implementation steps are as follows:

in the first implementation mode, assuming that the requirement of driving in and out of a bottleneck point of each traffic jam is not changed, when calculating the characteristic data of the jam portrayal, the embodiment can collect the traffic flow of a section at a stop line (intersection radar data can be used or portable equipment can be used) in a peak period, the diversion flow is best, the dynamic flow rate mu in the score period is obtained, the average travel time tt of an inlet road section corresponding to the flow rate time period is calculated by using number plate data and floating car data (the travel time is calculated according to the stop line of a vehicle passing through the stop line, a holographic video traffic data collecting device is arranged on an electric police rod, the number plate at the stop line is captured or the portable equipment is used for investigation), and simultaneously the queuing length of each time point is investigated and used for determining t₃-t₀Determination of the congestion period t at two points in time, or by a functional relationship between the incoming flow rate and the outgoing flow rate₀、t₂、t₃。

Then, using historical congestion period data, according to the formula: λ (t) - μ (t) ═ γ (t-t0) (t-t2) (t-t3), a γ coefficient can be obtained. λ (t) is the dynamic entry flow rate and μ (t) is the dynamic exit flow rate. When the method is used for calculating gamma, the data acquisition period is longer than the signal period, and 5 minutes is recommended to form a data set. The method is suitable for the bottleneck points of the closed road sections, which are inconvenient to obtain travel time data and are calculated. Since a time-varying exit flow rate μ (t) is used, rather than a congestion period average exit flow rate, the calculated γ may be more accurate.

In a second implementation manner, assuming that delay characteristics of two adjacent traffic jams are the same, when calculating the jam image feature data, the method includes the following steps:

first, the length L of the road section and the coordinated speed V of the speed limit or line of the road section are utilized_fCalculating the free stream travel time TT_f＝L/V_f。

And secondly, calculating gamma by using a delay time-varying equation:

w(t)＝tt-TT_fand substituting the multi-period mu, t3 and t0 into a formula to obtain a gamma coefficient, wherein the gamma coefficient is a road congestion characteristic coefficient and is related to the road structure, and the gamma coefficient is kept unchanged in a normal condition. The method is suitable for calculating the bottleneck point with the flow interference of the entrance and the exit of the road section, and the delay characteristic of each congestion changes due to the existence of traffic optimization, so when the method is used, the dynamic congestion image characteristic parameters are required to be adopted.

With the congestion image characteristic parameter gamma, a congestion image evolution curve can be drawn according to a queuing length time-varying function, a delayed time-varying function and a time-varying function of travel time in the congestion image, and time-varying parameters such as time-varying queuing length, time-varying delay and time-varying travel time can be predicted; meanwhile, according to the total delay, the average delay and the average travel time function of the system, the evaluation parameters of the affected road section relative to the free flow can be calculated, wherein the evaluation parameters comprise the total delay parameter, the average travel time parameter and the like.

And the evolution process of congestion is predicted in real time by using the congestion image, the split green ratio of each phase is adjusted in real time, the control is optimized, and the traffic efficiency is improved. Since the flow rate μ changes all the time during congestion due to adaptive control, the average flow rate from t0 to t is counted every five minutes, and the real-time delay is calculated every 5 minutes by substituting the off-line calculated γ and the time t3 into the time-varying delay equation:

and predicting the delayed evolution process, adjusting the split green ratio of each phase based on the real-time delayed development situation every 5 minutes, and considering that the split green ratio is adjusted every 5-15 minutes to realize single-point control optimization.

Naturally, the dynamic prediction result of the congestion image is usually used for optimizing single-point adaptive control, determining the split green ratio of each phase, realizing finer active control, and also can be used for actively inducing traffic congestion, and the congestion information is sent to the vehicle to be arrived through a vehicle-road cooperative communication channel, so as to realize the purpose of relieving the congestion; the method has the advantages that important data support is certainly achieved for prediction information of each intersection provided by the coordination optimization and the congestion images. For line coordination control, while optimizing the green signal ratio by using the congestion portrait, the green wave bandwidth is considered, the minimum green wave bandwidth is ensured, meanwhile, the traffic efficiency of single-point signals is fully ensured, and the delay is minimum.

In another implementation manner, if the congestion type is road construction congestion, the embodiment first obtains a construction road segment, and obtains a total dynamic demand time of the construction road segment in a peak time period, a travel time difference before and after construction of the construction road segment, and an average flow speed of the traffic bottleneck point in the peak time period. And then calculating to obtain the congestion image characteristic data according to the dynamic demand total time, the running time difference and the average flow speed.

In specific implementation, the evaluation of the influence of the road construction is basically carried out before the road construction, so that some parameters are assumed. In the first few days of the construction, since the drivers who often pass by are not aware of the construction, the passing requirement is assumed to be unchanged, so that traffic jam is only caused to the construction road section, the influence is not spread to other road sections, and the traffic influence parameters such as delay, queue and the like can be calculated by using the jam images at the stage. The specific implementation steps are as follows:

firstly, surveying the section traffic flow of the whole peak time period including the peak in the morning, the middle and the evening at the position to be constructed, namely the position of the future traffic bottleneck, using a portable roadside laser device, collecting the section flow and the average speed V_fThe parameters are equal, the period is 5 minutes, and the total D of the dynamic demands in each peak period is counted; the length L of the affected link is measured, and the travel time TTf of the link before the construction is not performed is calculated to be L/Vf, i.e. the travel time is considered to be the travel time of the free stream.

And a second step of estimating an average flow rate mu and an average speed VL during a peak period of a bottleneck position after starting construction, based on the detected traffic flow data, the number of lanes of the road section, and the construction influence type, and assuming that the average speed is an average speed of the whole influenced road section, the average travel time tt of the road section after construction is L/VL. The accuracy of the two parameter estimation directly affects the accuracy of the evaluation result, and from experience, the accuracy of the two parameter estimation needs to be improved continuously.

Thirdly, according to an average delay equation of the congestion image:

w ═ tt-TTf, D and μ described above are substituted into the equation to obtain a γ coefficient, which is a link congestion feature coefficient and is related to the road structure, and this coefficient is normally kept constant.

The delay time-varying equation is not used for solving the gamma, mainly because in the application, the traffic jam bottleneck is virtualized, only the average flow rate mu and the average speed VL can be estimated, and the section flow and the average speed of each time interval cannot be obtained, so that a simplified method is used, the average delay equation is used for calculating the gamma, the method generates deviation, and the gamma needs to be calibrated after construction.

And fourthly, calculating the congestion ending time t3 according to the congestion period formula P, namely t3-t0, namely D/mu. Assuming here that the total demand D during peak hours is constant, μmust be, as an estimate, much smaller than the flow rate of the free stream, so that the congestion period P will necessarily be lengthened, t₃The value becomes large.

With the above-mentioned congestion image characteristic parameters gamma and t₃After the value is obtained, a congestion image evolution curve can be drawn according to a time-varying function of the queuing length of the congestion image, a time-varying function of delay and a time-varying function of travel time, and time-varying parameters such as time-varying queuing length, time-varying delay and time-varying travel time are predicted; meanwhile, according to the system total delay equation, a total delay parameter, an average delay parameter and the like can be calculated.

In other embodiments, if the congestion type is a traffic event congestion, the embodiment first obtains a congested road segment of a traffic event, and obtains a total congestion time of the congested road segment, a travel time difference between the congested road segment before the occurrence of the congestion event and the congested road segment after the occurrence of the congestion event, an average flow speed of the traffic bottleneck point, and a congestion period. And then calculating to obtain the congestion image characteristic data according to the total congestion time, the running time difference, the average flow speed and the congestion period. In specific implementation, various emergencies are various, such as construction, accidents, large-scale activity management and control, hospital periphery, school periphery and the like. The various emergencies in this embodiment are classified into two categories: firstly, events are influenced in a normal state, such as long-term construction, periphery of a hospital, periphery of a school and the like; and secondly, emergencies affect events, such as accidents, large-scale activity management and control and the like.

The method is mainly used for predicting the congestion development situation of the known traffic bottleneck in real time, making a congestion response plan, reducing the congestion degree of the bottleneck of the road network through various information distribution and improving the traffic efficiency; the embodiment is mainly used for rapidly predicting suddenly appearing trafficThe congestion development situation of the bottleneck is established, a congestion response plan is formulated, the congestion degree of the bottleneck of the road network is reduced through various information issues, and the traffic efficiency is improved. Specifically, in predicting normal event traffic, normal impact events include long term construction, hospital surroundings, school surroundings, etc., since such events have occurred, the average flow rate μ at the bottleneck, and the average speed V_LCan be actually detected. The specific implementation steps are as follows:

firstly, surveying the section traffic flow of the whole jam period at the position of a traffic jam bottleneck, and recording t₀And t₃Temporarily installing a portable roadside laser device (making a survey) at the position of the congestion bottleneck, and collecting section flow rate mu in each time interval and average speed V in each time interval_LThe parameters are equal, the period is 5 minutes, the length L of the affected road section is measured, and the average travel time tt of the road section in each detection period is calculated to be L/V_LCounting the total demand D of the whole congestion period; calculating the free flow travel time TT of the road section_f＝L/V_fVelocity of free flow V_fAnd limiting speed by adopting a road section.

And secondly, obtaining a gamma coefficient by using a delay time-varying equation of the congestion image:

w(t)＝tt-TT_fthe above-mentioned multi-period μ and t₃、t₀And substituting the coefficient into a formula to obtain a gamma coefficient, wherein the gamma coefficient is a road congestion characteristic coefficient and is related to the road structure, and the gamma coefficient is kept unchanged in a normal condition.

With the congestion image characteristic parameter gamma, a congestion image evolution curve can be drawn according to a time-varying queuing length function, a time-varying delay function and a time-varying travel time function of the congestion image, and time-varying parameters such as time-varying queuing length, time-varying delay, time-varying travel time and the like can be predicted; meanwhile, according to the total delay, the average delay and the average travel time function of the system, the evaluation parameters of the affected road section relative to the free flow can be calculated, wherein the evaluation parameters comprise the total delay parameter, the average travel time parameter and the like.

For the prediction of the traffic influence of the emergency, because the emergency has short influence time and belongs to a local road section, the prediction method and the prediction steps are basically the same as the method for predicting the traffic influence of the normal event, but the actual data acquisition cannot be carried out, and the rapid prediction is carried out by adopting the historical events with various attributes similar to the road section of the emergency as data bases, and the specific method is as follows:

for different types of emergencies, data acquisition can be carried out by using the RLU5 at bottleneck section immediately when the emergencies occur, and the data acquisition is used for analyzing the influence time t3-t0 after different types of emergencies are generated, the traffic flow rate mu at the bottleneck and the average speed V_LAnd the length L of the affected road section, calculating a bottleneck congestion characteristic coefficient gamma, establishing a congestion parameter library of the affected events aiming at different road structures such as the number of lanes, the road grade, the traffic flow and the number of the affected lanes of the event, directly using parameters of the parameter library to predict the congestion when similar affected events occur again,

after that, the influence of the emergency on the traffic jam can be predicted quickly and accurately. The significance of the prediction result is that the prediction result can provide a timing implementation basis for intersection control or entrance and exit control at the upstream of a bottleneck, so that the occurrence of congestion is reduced.

And step S300, carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management.

In this embodiment, after obtaining the congestion image characteristic parameter, the present embodiment may perform congestion prediction on the traffic data according to the congestion image characteristic parameter to obtain a prediction result. Therefore, the embodiment can predict time-varying parameters such as traffic time-varying queuing length, time-varying delay, time-varying travel time and the like according to the congestion image characteristic data; meanwhile, according to the total delay, the average delay and the average travel time function of the system, the evaluation parameters of the affected road section relative to the free flow, including the total delay parameter, the average travel time parameter and the like, can be calculated, so that the traffic can be better managed. After the prediction result is obtained, the prediction result can be sent to other equipment, such as a signal lamp controller, to optimize a signal lamp, or sent to a vehicle road for active induction in cooperation.

In summary, in the embodiment, traffic data is acquired, and congestion image data of a traffic bottleneck point in the traffic data is acquired according to the traffic data, where the congestion image data reflects a road condition state of the traffic bottleneck point; according to the congestion image data of the traffic bottleneck point, determining a congestion image characteristic parameter of the traffic bottleneck point; and carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management. Therefore, the traffic data can be predicted based on the congestion image, the obtained prediction result can be used for active traffic guidance management, the basis for the police to give out the police can be determined according to the prediction result, the basis is provided for the optimal timing of signals and the like, and the basis is provided for the travel route planning.

Therefore, the traffic jam image model and the traffic data detection equipment supporting efficient operation of the model are organically combined together, and dynamic congestion image is realized. By utilizing edge calculation, the local traffic jam is quickly and accurately predicted, and the prediction information is sent to a signal controller, so that active signal control can be realized; through the cooperation of the vehicle and the road, the traffic prediction information can be issued to the vehicle for active traffic guidance management. The method can realize multiple application functions, for example, the basis of police dispatch of traffic police is determined according to the congestion degree, the basis is provided for signal optimization timing, the influence evaluation is carried out on the emergency traffic incident, the traffic demand trip is actively induced through the cooperation of the vehicle and the road according to the congestion development situation, and the basis can also be provided for the trip route planning. Because the embodiment specially develops holographic video traffic data detection equipment and edge calculation equipment for the congestion image model, provides high-precision real-time traffic data and travel time data for the model, and lays a foundation for high-precision prediction, the prediction result of the method is more accurate.

Exemplary device

As shown in fig. 5, an embodiment of the present invention provides a traffic congestion prediction system based on a congestion profile, the system including: a congestion image data acquisition module 10, a congestion image characteristic parameter acquisition module 20 and a congestion prediction module 30. In this embodiment, the congestion image data obtaining module 10 is configured to obtain traffic data, and obtain congestion image data of a traffic bottleneck point in the traffic data according to the traffic data, where the congestion image data reflects a road condition of the traffic bottleneck point. The congestion image characteristic parameter obtaining module 20 is configured to determine a congestion image characteristic parameter of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point. The congestion prediction module 30 is configured to perform congestion prediction on the traffic data according to the congestion image characteristic parameter to obtain a prediction result, where the prediction result is used for traffic management.

In an implementation manner, the traffic data of this embodiment is collected based on traffic data detection equipment, the traffic data detection equipment in this embodiment integrates a structured video technology and a multi-target radar technology, and performs an integrated structural design, and the video cameras are arranged in an upper structure and the radar is arranged in a lower structure. The structural design can ensure that the horizontal X, Y coordinate origin of the video camera and the radar are at the same point, which is beneficial to the effective fusion of the two data; because the visual angles of the video camera and the radar are slightly different, the pitching angle adjustment of the two parts is independently completed; the whole holder structure can rotate horizontally and shake head left and right, so that the applicability of various installation conditions is ensured. The device can thus collect the following data: license plate number data, vehicle type flow data, steering flow data and section average speed data at the exit of the traffic bottleneck, and vehicle arrival data, queuing length data and vehicle type flow data at the upstream of the bottleneck. In addition, the shell of the camera adopts a double-layer sheet metal structure, the inner layer is made of 304 stainless steel, and the outer layer is made of aluminum alloy, so that the camera can resist rain and heat, and the stable work of the camera is ensured; the shell structure of the radar radiation cover adopts a compact design, and the consistency of radar signals is ensured by adopting mould processing.

In a specific prediction process, the edge calculation unit in this embodiment collects data of the holographic video traffic data detection device, can simultaneously receive data of 4-channel devices, performs fusion processing on the data, and packages and sends the data to the central service program, where the data content includes data required by all congestion image models. And then after the data of the edge calculation reaches the management center, the data is split by a center service program, the congestion image characteristic parameter gamma of each bottleneck point is calculated, and the congestion image characteristic parameter gamma is returned to the edge calculation unit. And then after receiving the latest congestion image characteristic parameters, the edge computing unit performs second-level congestion prediction according to data acquired by the whole-system video equipment in real time, and sends the prediction result to other equipment, such as a signal controller, for signal optimization or road coordination for active guidance. The edge computing module in the embodiment has a compact design structure, is convenient to be placed in the existing facilities, such as a signal cabinet, an electronic police distribution box and the like, has rich interfaces, CAN be connected with a center or other equipment in various modes, and comprises a network port, RS232, RS485, CAN, an optical coupler and power supply output. Because the embodiment specially develops holographic video traffic data detection equipment and edge calculation equipment for the congestion image model, provides high-precision real-time traffic data and travel time data for the model, and lays a foundation for high-precision prediction, the prediction result of the method is more accurate.

In one implementation, the congestion image data acquisition module 10 includes:

the traffic data acquisition unit is used for acquiring holographic video data acquired by traffic data detection equipment and taking the holographic video data as the traffic data;

and the congestion image data acquisition unit is used for inputting the traffic data into a preset congestion image model, and determining a traffic bottleneck point in the traffic data and congestion image data corresponding to the traffic bottleneck point through the processing of the congestion image model.

In one implementation, the congestion image characteristic parameter obtaining module 20 includes:

the congestion type determining unit is used for determining a congestion type corresponding to the congestion image data according to the congestion image data of the traffic bottleneck point;

and the congestion image characteristic data determining unit is used for determining congestion image characteristic data of the traffic bottleneck point according to the congestion type.

Based on the above embodiments, the present invention further provides an intelligent device, and a schematic block diagram thereof may be as shown in fig. 6. The intelligent device comprises a processor, a memory, a network interface, a display screen and a temperature sensor which are connected through a system bus. Wherein the processor of the smart device is configured to provide computing and control capabilities. The memory of the intelligent device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the intelligent device is used for connecting and communicating with an external terminal through a network. The computer program is executed by a processor to implement a traffic congestion prediction method based on a congestion profile. The display screen of the intelligent device can be a liquid crystal display screen or an electronic ink display screen, and the temperature sensor of the intelligent device is arranged inside the intelligent device in advance and used for detecting the operating temperature of the internal device.

It will be understood by those skilled in the art that the block diagram of fig. 6 is a block diagram of only a portion of the structure associated with the inventive arrangements and is not intended to limit the smart devices to which the inventive arrangements may be applied, and a particular smart device may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a smart device is provided, where the smart device includes a memory, a processor, and a batch distribution network program of the smart device stored in the memory and executable on the processor, and when the processor executes the batch distribution network program of the smart device, the following operation instructions are implemented:

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, the computer program can include the processes of the embodiments of the methods described above. Any reference to memory, storage, databases, or other media used in embodiments provided herein may include non-volatile and/or volatile memory. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

In summary, the invention discloses a traffic congestion prediction method and system based on a congestion image, wherein the method comprises the following steps: acquiring traffic data, and acquiring congestion image data of a traffic bottleneck point in the traffic data according to the traffic data, wherein the congestion image data reflects a road condition state of the traffic bottleneck point; according to the congestion image data of the traffic bottleneck point, determining a congestion image characteristic parameter of the traffic bottleneck point; and carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management. The method can predict the traffic data based on the congestion image, the obtained prediction result can be used for active traffic guidance management, and the police output basis can be determined according to the prediction result, so that basis is provided for signal optimization timing and the like, and basis is provided for travel route planning.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. A traffic congestion prediction method based on a congestion profile, the method comprising:

carrying out congestion prediction on the traffic data according to the congestion image characteristic parameters to obtain a prediction result, wherein the prediction result is used for traffic management;

the determining the congestion image characteristic parameters of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point comprises the following steps:

determining congestion image characteristic data of the traffic bottleneck point according to the congestion type;

if the congestion type is signal lamp intersection vehicle saturated congestion, determining congestion image characteristic data of the traffic bottleneck point according to the congestion type, wherein the congestion image characteristic data comprises the following steps:

2. The method for predicting traffic congestion based on a congestion image as claimed in claim 1, wherein the obtaining traffic data and obtaining the congestion image data of a traffic bottleneck point in the traffic data according to the traffic data comprises:

and inputting the traffic data into a preset congestion image model, and determining a traffic bottleneck point in the traffic data and congestion image data corresponding to the traffic bottleneck point through processing of the congestion image model.

3. The method of claim 2, wherein the congestion map model is created in a manner comprising:

4. The congestion profile-based traffic congestion prediction method according to claim 3, wherein the obtaining traffic congestion parameters comprises:

acquiring on-site traffic investigation data or acquiring traffic real-time data acquired based on a sensor;

5. The method for predicting traffic congestion based on a congestion map as claimed in claim 1, wherein the calculating congestion map feature data of the traffic bottleneck point according to the vehicle entrance and exit demand or the characteristics of congestion delay between two adjacent times comprises:

alternatively, the first and second electrodes may be,

6. The method for predicting traffic congestion based on a congestion image as claimed in claim 1, wherein if the congestion type is a road construction congestion, the determining the congestion image feature data of the traffic bottleneck point according to the congestion type comprises:

7. The method for traffic congestion prediction based on a congestion profile as claimed in claim 1, wherein if the congestion type is a traffic event congestion, the determining the congestion profile feature data of the traffic bottleneck point according to the congestion type comprises:

8. A traffic congestion prediction system based on a congestion profile, the system comprising:

a congestion image characteristic parameter acquisition module, configured to determine a congestion image characteristic parameter of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point; the determining the congestion image characteristic parameters of the traffic bottleneck point according to the congestion image data of the traffic bottleneck point comprises the following steps:

calculating to obtain congestion image characteristic data of the traffic bottleneck point according to the vehicle access demand or the congestion delay characteristics of the two adjacent times;