CN101739824A - Data fusion technology-based traffic condition estimation method - Google Patents

Abstract

The invention discloses a method for estimating traffic conditions based on data fusion technology. According to the principle of "more information contributes to more accurate estimation of traffic conditions", the present invention utilizes sensors such as global positioning system (GPS) and circular vehicle detectors to obtain current traffic condition information, and after preprocessing, it is compared with weather Bayesian network is used for fusion processing of sensor data at different times, time periods, and at different times to obtain accurate information on current traffic conditions. In addition to the current sensor data, this method also considers multiple other factors that affect traffic conditions, and performs fusion processing to avoid the misjudgment problem caused by a single sensor information. The traffic condition estimation method is especially suitable for the traffic environment of a big city.

Description

Traffic Condition Estimation Method Based on Data Fusion Technology

technical field

The invention relates to a method for estimating traffic conditions, in particular to a method for estimating traffic conditions based on data fusion technology which overcomes the disadvantages of inaccurate state estimation and large errors caused by a single data source.

Background technique

With the development of the world's automobile industry and the increasing popularity of family cars, road traffic congestion has become an important problem in urban traffic; on the other hand, the rapid progress of information technology has brought opportunities for comprehensive solutions to traffic problems. It is against this background that advanced traffic information management systems have received widespread attention prior to other systems of intelligent transportation systems, and have been rapidly developed in countries all over the world. They are used in dynamic route planning, dynamic navigation, road network adjustment Coordinate traffic signal system, dynamic traffic dispatching and other aspects. Among them, the estimation of real-time road network traffic status is a key component of the traffic information management system.

The real-time estimation of road network traffic status is related to the traffic information used, and different traffic information determines the estimation method and accuracy. At present, there have been many related studies in the world, among which the representative one is Martin L. Hazclton ("Estimating Vehicle Speed from Count and Occupancy data, Journal of Data Science 2(2004), 231-244"). MartinL.Hazclton effectively considered and modeled the problem of high error rate and low reliability of road detection ring data, and the results are gratifying, but his research was carried out on the highway, which is only applicable to the situation where the traffic flow is a continuous flow. The urban traffic flow is discontinuous, so it is not suitable for the estimation of the traffic flow state of the urban road network. Using the detection ring data to estimate the traffic flow of the urban road network has high requirements on the urban infrastructure. In many cities, enough required data is often not obtained, and the problem of high error rate cannot be effectively solved.

In the patent document "Urban road network traffic flow state estimation method, application number: 200510026214.7", the applicant proposed a state estimation method based on GPS data and traffic control information of an adaptive traffic control system (SCATS). But simply using GPS data, there will be a lot of errors, and it is easy to cause misjudgment. For example, GPS data shows that the vehicle speed is 0, and the system generally judges it as a traffic jam, but the current vehicle may also be due to passengers getting on and off, or the GPS data of the vehicle caused by a red light shows that the vehicle speed is 0, which is not true. Traffic jams can easily cause errors in the estimation of traffic conditions. And the information of the traffic control system is not easy to get.

Contents of the invention

In view of the above problems, the main purpose of the present invention is to address the deficiencies in the prior art, to propose a new method for estimating the state of traffic flow in urban road networks based on data fusion, to overcome the inaccurate state estimation and error caused by a single data source. Big disadvantage.

The present invention solves the above-mentioned technical problems through the following technical solutions: in order to achieve such purpose, the data source of the technical solution of the present invention comes from, and a class comes from and sensor data, comprises the GPS loaded on the ring vehicle detector and the vehicle receiver. The circular vehicle detector can obtain the average speed information of vehicles on the current road; the GPS receiver mounted on the vehicle can also obtain the vehicle speed information on the current road, if the GPS information sent by multiple vehicles on the same road section is received at the same time , the average speed information of vehicles on the road section can be obtained.

Another type of information is non-sensor information closely related to traffic conditions, including current weather conditions and current time information. This information also has a very important impact on whether a road is congested.

The last type of information is sensor history information, including vehicle average speed information obtained by GPS receivers and road vehicle average speed information obtained by ring vehicle detectors within a short time from now.

This method mainly includes the following steps:

(1). Collect traffic condition information on the road surface through sensors;

(2). Preprocessing the sensor data collected by the sensor in step (1);

(3). Use the Bayesian network method to fuse the collected sensor data with the current weather and time;

(4). Estimate traffic conditions through the data obtained after fusion processing.

Described below:

1. Preprocess the GPS data:

The current real-time GPS data transmitted by the vehicle-mounted GPS receiver that can provide information including distance, time, and speed is regarded as the sampling point of traffic flow on each directional road section. The preprocessing of GPS data mainly includes firstly performing road matching according to the latitude and longitude information in the GPS data, and determining that the GPS data comes from the vehicle on that road section; then according to the vehicle speed information in all the GPS data on the road section, calculate Get the average speed of the vehicles on the road.

2. Preprocess the circular vehicle detector data:

Since the circular vehicle detector can directly provide the average speed of vehicles traveling on the road section, no additional processing is required.

3. Delineate the road congestion state by the average speed:

Taking the average speed of each directional road section in the urban road network at time _t0 as the index, the road congestion state is demarcated. The average speed of road sections in the urban road network is divided into five speed grades, which correspond to five road congestion states: smooth, relatively smooth, not smooth, congested, and severely congested. The congestion state of each directional road section is judged according to the speed level of the average speed corresponding to each directional road section.

4. Bayesian network construction for data fusion:

According to the causal relationship between data, a Bayesian network is constructed. Each node in the Bayesian network represents a random variable, and the arrow points from the cause to the effect. The probability distribution table of the random variable represented by each node is obtained through observation statistics and experiments.

5. Data fusion based on variable elimination method

Variable elimination is a commonly used Bayesian network inference method. Gradually eliminate other variables that do not belong to the query variable, that is, the traffic condition variable, and finally obtain the probability distribution of the query variable, which is the traffic condition, based on the known evidence variables. Estimate the traffic situation after fusion.

The positive progress effect of the present invention is: the traffic condition estimation method based on data fusion technology provided by the present invention has the following advantages:

1. In addition to the current sensor data, this method also considers multiple other factors that affect traffic conditions, and performs fusion processing to overcome the shortcomings of inaccurate state estimation and large errors caused by a single data source.

Description of drawings

Fig. 1 is the block diagram of flow chart of urban road network traffic condition estimation method;

Figure 2 is a schematic diagram of a Bayesian network used for data fusion.

Detailed ways

The preferred embodiments of the present invention are given below in conjunction with the accompanying drawings to describe the technical solution of the present invention in detail. The flow chart of Figure 1 urban road network traffic condition estimation method, as shown in Figure 1, the traffic condition estimation method based on data fusion technology provided by the present invention comprises the following steps:

(1). Collect traffic condition information on the road surface through sensors;

(2). Preprocessing the sensor data collected by the sensor in step (1);

(3). Use the Bayesian network method to fuse the collected sensor data with the current weather and time;

(4). Estimate traffic conditions through the data obtained after fusion processing.

Described below:

1. Data preprocessing

The required input data of the present invention is as shown in Figure 2, comprises GPS to measure current average speed, the average speed of GPS measurement in last moment, ring vehicle detector to measure vehicle average speed and last moment ring vehicle detector to measure vehicle average speed, and weather and time period data. The average speed v can be divided into 5 speed grades according to the actual traffic conditions. The severe traffic congestion corresponds to v≤10 km/h, the congestion corresponds to 10<v≤20, the unsmooth corresponds to 20<v≤30, and the relatively smooth corresponds to When 30<v≤50, unobstructed corresponds to v>50. The weather data is divided into severe, poor, fair and good according to the degree of impact on traffic. The time period is divided into peak period, congested period, general period and smooth period according to the degree of impact on traffic.

2. Bayesian network construction

The constructed Bayesian network structure is shown in Figure 2. The arrows indicate that there is a causal relationship between the two. If you want to improve the accuracy of data fusion, you can add more data, such as adding more historical sensor data, or adding more Sensor data types, for example, video data, etc. For the determination of the probability distribution of each node in the Bayesian network, the method of actual measurement and expert verification is adopted. Through long-term observation of a certain road section, it can be determined that under different weather conditions, traffic conditions, and thus the conditional probability distribution from weather to traffic conditions can be determined.

3. Data fusion based on variable elimination method

The input fusion process is as follows:

①Initialization

Construct an initial intermediate function queue, put the functions corresponding to all nodes in the Bayesian network into it; then set the initial value of the initial function queue according to the obtained sensor data, weather and other variables; then construct an The element variable sequence to be eliminated, the sequence does not include the known preset value variable and the traffic condition variable to be obtained.

②If there are still variables to be eliminated in the sequence of variables to be eliminated, repeat this step

Construct a new function queue, move all the functions related to the variables to be eliminated in the original function queue into the new function queue; construct a new function, which is the calculation result of the new function queue; All variables involved in the function queue are assigned sequentially, that is, from (0, 0, ..., 0) to (m, n, ..., z) (where m, n, ..., z, etc. are the values of each variable maximum value classification number), calculate the multiplication value of the new function queue for each assignment, and accumulate the multiplication value according to the value of the element variable to be eliminated. The value obtained is the function value of the new function under the current assignment condition ; Add the calculated new function to the original function queue;

③Merge all the functions in the current function queue into one function and perform normalization processing. The finally obtained function is the probability distribution of the traffic conditions to be requested under the current known conditions, that is, the current traffic conditions with the highest probability estimate.

The basic principles and main features of the present invention and the advantages of the present invention have been shown and described above. Those skilled in the industry should understand that the present invention is not limited by the above-mentioned embodiments. What are described in the above-mentioned embodiments and the description only illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will also have Various changes and improvements fall within the scope of the claimed invention, which is defined by the appended claims and their equivalents.

Claims (4)

1. traffic condition estimation method based on the data fusion technology, it is characterized in that: described method comprises the steps:

(1). by the traffic related information on sensor acquisition road surface;

(2). the sensing data to sensor acquisition in the step (1) carries out pre-service;

(3). sensing data and current weather, the time of gathering are adopted Bayesian network method, data are carried out fusion treatment;

(4). by the data estimation traffic that obtains after the fusion treatment.

2. the traffic condition estimation method based on the data fusion technology according to claim 1, it is characterized in that: described sensor comprises two classes: GPS (GPS) receiver that a class is on the automobile to be installed, another kind of is to be embedded in the underground annular wagon detector of road.

3. the traffic condition estimation method based on the data fusion technology according to claim 1 is characterized in that: described sensing data had both comprised the sensing data of current time also comprising the sensor historical data.

4. the traffic condition estimation method based on the data fusion technology according to claim 2, it is characterized in that: described data of carrying out fusion treatment comprise that GPS (GPS) measures current average velocity, last average velocity, annular wagon detector measuring vehicle average velocity and the last moment annular wagon detector measuring vehicle average velocity measured of GPS (GPS) constantly, and weather and time period data.

Images