Background technique
Vehicle-mounted short haul connection (Vehicle to X:V2X) network be by wireless communication, the short distances such as GPS/GIS, sensing
Car (CAN-Controller Area Network), bus or train route (Vehicle-2-RSU), workshop from communication technology realization
(Vehicle-2-Vehicle), between vehicle outer (vehicle-2-Infrastructure), people's vehicle (Vehicle-2-Person)
Communication.
For vehicle-mounted node in V2X equipped with equipment such as GPS or Beidous, vehicle-mounted node is regular at set time intervals
The information (referred to as heartbeat message) such as the geographical location to surrounding broadcast this node, while also receiving the heartbeat letter that surroundings nodes are sent
Breath, so that the distance of around vehicle-mounted node with this vehicle is obtained, to calculate the relative distance with this vehicle, speed, acceleration etc.
Information.
It is identical with vehicle-mounted node in trackside node device structure in V2X network, but since its allocating antenna position is high,
Therefore coverage area is big, can receive the heartbeat message of vehicle broadcast in larger scope.Meanwhile trackside node is with powerful
Operational capability powers unrestricted, therefore, can execute more complicated operation.
Effectively the driving behavior of driver is evaluated and tested, finds high risk vehicle, and remind in time driver,
It can effectively improve traffic safety, but in traditional road management, the real-time row of vehicle since some region can not be obtained
Situation is sailed, this function is difficult to carry out.
Using the advantage of the trackside node in V2X network, the information of vehicle in larger range can be obtained, but to high wind
The identification of dangerous vehicle needs to face the comprehensive identification to Multiple factors, and is a random probability event, therefore in algorithm
On face bigger difficulty.Uncertainty, relevance between multiple variables can be effectively treated in Bayesian network, and supports to pass through
A large amount of historical datas are trained, to obtain significantly more efficient network parameter.
The embodiment of the present invention receives the heartbeat message of vehicle in coverage area by the Road V2X side gusset, to obtain each
The real-time position information of a vehicle, and then the information such as the distance between the acceleration of each vehicle, vehicle and acceleration are acquired, pass through
Bayesian network identifies the dangerous driving of vehicle, and using V2X network by the dangerous driving risk of high risk vehicle because
Element is broadcast to the vehicle and other surrounding vehicles, to achieve the purpose that improve road safety.
Specific embodiment
The present embodiment realizes a kind of trackside node high risk vehicle alarm suitable for vehicle-mounted short distance communication network and mentions
Show method, coverage area big advantage high using trackside node antennas deployed position, to driving behavior risk in coverage area compared with
High vehicle is identified, and driver's risk factors that thresholding is set more than risk is reminded by alarm, while also reminding week
Vehicle is enclosed note that improve the safety of road.
High risk driving behavior identification is related to several factors, and the present embodiment mainly considers two aspects: (1) longitudinal drive.
Car speed is too fast, too fast acceleration and deceleration are the Major Risk Factors for causing longitudinal impact, and caused rear-end impact is touched in vehicle
The accident of hitting occupies sizable part.Therefore, in the present embodiment, the running speed of driver, acceleration and deceleration characteristic is selected to know
Not.(2) transverse driving.It is also the main reason for causing traffic accident that vehicle, which sails out of former lane, and transverse driving is mainly to driving
Member's changing Lane, the driving behavior overtaken other vehicles, turned round are identified.In the present embodiment, select side front-and-rear vehicle distance from, to rear car
Influence is evaluated and tested.
The core of the present embodiment is to carry out automatic identification to high risk driving behavior and risk information is notified driver
With the vehicle of surrounding, in V2X network, trackside node by receive vehicle periodic broadcasting in coverage area comprising geographical location
Heartbeat message can obtain the position of vehicle, the speed and change in location of each vehicle can accordingly be calculated, to obtain height
Risk drives the essential information needed.
Driving behavior risk identification needs comprehensive Multiple factors, and Bayesian network is a kind of in artificial intelligence field carries out generally
The method of rate reasoning can carry out two-way uncertain inference using local condition's dependence in model, be applied to analysis
With judge etc., the present embodiment using Bayesian network realize multi-variables analysis.
The implementation of the present embodiment includes five parts.
First part determines driving behavior risk identification factor and its value.
1.1, driving behavior risk identification variable factors are determined.In the present embodiment, driving behavior risk D drives from longitudinal respectively
Two aspects of behavior A and transverse driving behavior B are sailed to be identified.Longitudinal drive A is again special from car speed A1 and acceleration and deceleration respectively
Property A2 is identified.Car speed A1 and acceleration and deceleration characteristic A2 is again related to road conditions A11 and leading vehicle distance A12 respectively.It is horizontal
It is identified respectively from preceding rear car distance B1 and rear car influence B2 to B is driven.
1.2, each variable-value is determined.The present embodiment to simplify the calculation, using identical value, i.e., D, A, A1, A11,
A12, B1, B2 } variable to take range be { danger 0, safety 1 }.
Second part, the framework for determining identification Bayesian network
Bayesian network includes two parts: (1) the bayesian network structure figure of directed acyclic, and node on behalf variable is oriented
Arc represents interrelated;(2) conditional probability table between node and node indicates the relationship of relationship between node.
Bayesian network is with N={ < V, E >, P }, wherein D indicates the acyclic figure (DAG) of Bayes, X=(X1,
X2,...,Xn) it is node set, joint probability distributionWherein, Parent
(xi) indicate xiThe father node of node.If obtaining the prior probability and node of each interdependent node from the statistical result of existing information
Between conditional probability, using p (X) can reasoning other nodes probability.Bayesian network (BN) inference method is, given
When to the actual value of variables collection E, the posteriority conditional probability distribution P (Q | E) of variables collection Q is calculated, each is driven according to above-mentioned
The causality of risk identification factor is sailed, the Bayesian network framework of the present embodiment is as shown in Figure 2.
Part III determines conditional probability between Bayesian network node
3.1, it is directed to each node, each node is obtained with the condition between father node by expertise and historical data
Probability.By taking the conditional probability of node A and node A1 and A2 as an example, according to expertise or historical data determine each node it
Between probability it is as shown in table 1.
Table 1
Wherein, { A1-1, A1-2 ... ..., A1-m } indicates m different conditions of father node variables A 1, { A2-1, A2-
2 ... ..., A2-n } indicate n different conditions of father node variables A 2, { A1, A2 } indicates a different groups of m × n of two variables
It closes.{ s-1, s-2 ... ..., s-k } indicates the k different states of node A, and Pi-j indicates the A section of corresponding certain { A1, A2 } combination
Point is (s-j) shape probability of state, is had
Further, with one group of specific data explanation.By taking the node A1 of Fig. 2 and two father node A11, A12 as an example, A11
Value range with A12 is { dangerous, safety }, then according to expertise or historical data, can determine A11, various points of A12
The conditional probability of the three state of cloth corresponding A 1, as shown in table 2.
Table 2
3.2, the conditional probability of other each nodes of Bayesian network and its father node can be successively obtained using the method.
Part IV, Bayesian Network Inference and training
4.1, Bayesian Network Inference.
In Bayesian network, it is known that the conditional probability between the probability and node of father node, it can obtain final
Conditional probability, equal to the product of all conditions probability distribution, i.e. joint probability distribution.Have
Wherein, Parent (xi) indicate xiThe father node of node.By taking node A1, A11, the A12 node of Fig. 2 as an example, it is known that certain
The probability of vehicle driving state A11 corresponding { dangerous, safety } is { A11-1, A11-2 } respectively, such as one group of specific number is such as
The probability { A12-1, A12-2 } of { 0.30,0.70 }, A12 corresponding { dangerous, safety } indicates, utilizes the determining node A of Part III
With father node A1、A2Conditional probability, then calculate:
P1=P (A1A11-1A12-1)=P (A11-1)P(A12-1)P(A1|A11-1A12-1)
Same method calculates
P2=P (A1A11-1A12-2),
P3=P (A1A11-2A12-1),
P4=P (A1A11-2A12-2),
In turn, it further obtains
Using same method, it can shifted onto step by step from the top to Bayesian network, finally obtain the lowermost
The identification probability of driving behavior identification D.
4.2, using expertise or the historical data of traffic accident, as prior probability, in conjunction between each node
Conditional probability, obtain the joint probability of each variable, matched with actual result, adjust each conditional probability, carry out
The training of Bayesian network, so that Bayesian network is more efficient.
Part V, driving behavior risk identification and alarm.
5.1, the parameter of the Bayesian network of driving behavior risk identification determines.In V2X network, trackside node has with backstage
Bayes's parameter after training is published in the storage unit of trackside node by the link of system connection, system.
5.2, the location information of vehicle in coverage area is obtained.Trackside node is fixed by receiving vehicle-mounted node in coverage area
The heartbeat message that phase sends obtains the position of vehicle, and then obtains each vehicle and front truck, rear car, the adjacent vehicle of the vehicle
The front truck in road and distance, the speed, acceleration information of rear car.
5.3, according to the threshold value of the data of acquisition and default, it is converted into the top father node of Bayesian network
Then { dangerous, safety } shape probability of state utilizes the software comprising Bayes net algorithm being installed in trackside node device
Derivation calculating is carried out, to finally obtain the probability of { dangerous, safety } evaluation result of each vehicle drive behavior.
5.4, certain vehicle drive behavior { dangerous } probability is more than that the thresholding of default believes risk then by V2X network
Breath, is broadcast to the vehicle and neighbouring vehicle including risk class and risk factors, the vehicle-mounted V2X terminal of the vehicle receives
After the risk information, acousto-optic hint is carried out, and show risk factors.Meanwhile the vehicle of the vehicle periphery, it risky can also mention
Show, driver is reminded to pay attention to the vehicle.
Through this embodiment, can be in vehicle-mounted short-range communication network, trackside node is fixed by vehicle-mounted node around receiving
The heartbeat message of phase broadcast, identifies the driving behavior of the vehicle in coverage area, so that discovery dangerous driving in time is practised
It is used, driver is reminded and prompts risk factor type, to remind driver that bad steering is overcome to be accustomed to, while also being mentioned
Wake up the vehicle periphery vehicle note that improving road traffic safety.
The above is presently preferred embodiments of the present invention and its technical principle used, for those skilled in the art
For, without departing from the spirit and scope of the present invention, any equivalent change based on the basis of technical solution of the present invention
Change, simple replacement etc. is obvious changes, all fall within the protection scope of the present invention.