Summary of the invention
The invention discloses a kind of identifications of highway danger road conditions and alarm suitable for vehicle-mounted short distance communication network
Method, main realize is to select to shed the recognition element of object, fault car and crowded three factors of road as Dangerous Area.Know
Other method is the bayesian network structure and algorithm of section risk identification established first, then establishes and determines to shed object, failure
The bayesian network structure and algorithm of vehicle and crowded three recognition elements of road, recognition methods is by vehicle-mounted short haul connection
It is each to obtain that the trackside node of network receives the heartbeat message comprising vehicle location of vehicle-mounted node periodic broadcasting in coverage area
The position of vehicle, so obtain vehicle speed and driving direction variation, then obtained by Bayesian network shed object,
Then the posterior probability of fault car and crowded three recognition elements of road obtains section risk by Bayesian network again
Posterior probability, when probability is higher than the threshold value of default, vehicle broadcast alarms message into coverage area.
Specifically, the invention discloses a kind of highway danger road conditions identifications suitable for vehicle-mounted short distance communication network
And alarm method, which comprises
It selects to shed the recognition element of object, fault car and crowded three factors of road as Dangerous Area, by vehicle-mounted
The trackside node of short distance communication network receives the heartbeat letter comprising vehicle location of vehicle-mounted node periodic broadcasting in coverage area
Cease to obtain the position of each vehicle, and then obtain the variation of the speed and driving direction of vehicle, then by Bayesian network come
The posterior probability for shedding object, fault car and crowded three recognition elements of road is obtained, is then obtained again by Bayesian network
The posterior probability of section risk, when there is dangerous phenomenon, the vehicle into coverage area is alerted.
Preferably, after trackside node calculates the threshold value that the probability of section risk is more than default, by vehicle-mounted short
Range communication network, the vehicle broadcast alarms information into coverage area.
Preferably, it is identified using Bayesian network, method is the Bayesian network knot of established section risk identification
Structure and algorithm, then establish determine shed object, fault car and crowded three recognition elements of road bayesian network structure and
Algorithm.
Preferably, trackside node is the vehicle region detour obtained by V2X network and vehicle to object recognition element is shed
Two information walk or drive slowly to calculate the probability of generation.
Preferably, trackside node is that the vehicle obtained by V2X network stops and vehicle is slow to fault car recognition element
Two information of row calculate the probability of generation.
Preferably, trackside node is that the vehicle obtained by V2X network stops and vehicle is slow to the crowded recognition element of road
Two information of row calculate the probability of generation.
The embodiment of the present invention is deployed in the received vehicle-mounted node broadcasts of the trackside node along highway using V2X network
Heartbeat message, obtain trackside coverage within the scope of vehicle position, and then obtain vehicle speed and vehicle line variation,
By the change in location of vehicle, cooperate electronic map, the risk of highway is identified, when risk is more than threshold value
When, to vehicle broadcast alarms information, to improve the travel safety of highway.
Specific embodiment
The present embodiment realizes a kind of highway danger road conditions identification suitable for vehicle-mounted short distance communication network and accuses
Alarm method, it is therefore an objective to it is big by the trackside coverage range of vehicle-mounted short haul connection, there is the advantage of link connection between each other,
Receive the message comprising geographical position coordinates that the V2X mobile unit periodic broadcasting within the scope of trackside coverage in vehicle is sent
(hereinafter referred to as heartbeat message), using Bayesian network model, the comprehensive road conditions satisfied the need in side gusset coverage area carry out dangerous
Situation differentiates, if road conditions danger probability is more than the threshold value of default, warning information is passed through this trackside node and neighbour
Close trackside node is broadcasted, and driver is reminded to take appropriate measures early, thus the traffic of effectively Improving Expressway
Level of security.
For highway since speed is very fast, several factors make certain section become Dangerous Area, and three kinds of the present embodiment selection is right
Highway influences big factor, is to shed object or temporary obstructions, fault car or traffic accident vehicle, road are gathered around respectively
It squeezes, comprehensive judgement is carried out by Bayesian network.
Highway sheds object due to its Unpredictability, can bring great danger to the vehicle run at high speed.In addition
Provisional barrier or paroxysmal road abnormal conditions, also belong to this kind of risk factor.Vehicle encounters this kind of situation, and one
As be regular detour.The heartbeat message that trackside node is broadcasted by V2X mobile unit in the vehicle in reception coverage area,
This regular detour can be identified, to judge that road has the probability for shedding object or road exception.
Fault car or traffic accident vehicle are that another to the vehicle run at high speed can bring great risk factor,
This factor trackside node can accurately obtain the location information of vehicle by receiving the heartbeat message of fault car broadcast, still
The case where needing the position according to fault car and occupying road is come hazard recognition severity.
Road congested conditions trackside node can be by receiving speed and the vehicle density progress by the vehicle in the section
Identification.
The degree of danger in section is related to several factors, and each factor has biggish uncertainty, Bayesian network mould again
Type can be used for expressing and analyzing uncertain and probabilistic event, can be from incomplete, inaccurate or uncertain knowledge
Or reasoning is made in information, it is effective theoretical model in uncertain knowledge and reasoning field.The danger in section in highway
Property is an extremely complex system problem.Therefore, there is preferable feasibility using the method for probability theory.
In Bayesian network model, unavailable stochastic variable is inferred by obtainable stochastic variable, is carried out general
Rate reasoning.Therefore, solving the differentiation of section risk using Bayesian network can state are as follows: in existing road risk bring
In the case where the empirical data that vehicle-state changes, trackside node is under conditions of obtaining real-time vehicle state, road hazard shape
Condition is in the probability of various degree (dangerous, safety).
Bayesian network binary group BN=<G, Θ>statement probability uncertain inference network, G be node with it is discrete
Stochastic variable { X1,X2,...,XnOne-to-one directed acyclic graph, directed edge then indicates to be determined between variable by conditional probability
Dependence;Θ is the set for describing BN local condition probability distribution parametersIndicate nodes XiIn his father
Node collection YiA certain valued combinations state yiUnder conditional probability distribution.BN can be quantified between chance event by conditional probability
Causality is accordingly used in the identification of fatigue driving.
The implementation of the present embodiment includes three parts, and process is as shown in Figure 1.
First part, the bayesian network structure and algorithm for establishing the identification of section risk.
1.1, Bayesian network is constructed with directed acyclic graph, element of the node therein for section dangerous discernment is corresponding
Variable, directed edge represent the condition dependence of variable.
1.2, the Bayesian Network Topology Structures of the present embodiment are as shown in Figure 2.Road hazard identification in the present embodiment
Using three acceleration and deceleration characteristic, deviation characteristic, direction of traffic characteristic characteristics, hereinafter referred to as three characteristics of fatigue identification.Father's section
Point (Rd) indicates section risk, and object factor is shed in child node (Ao) expression, and child node (Fv) indicates fault car factor, sub- section
Point (Rc) indicates the crowded factor of road.Father node (Rd) and child node (Ao), child node (Fv), the line between child node (Rc)
Indicate section risk with shed the conditional probability p (Ao | Rd) of object factor, fault car factor and the crowded factor of road, p (Fv |
) and p (Rc | Rd) Rd.
1.3, the Bayes net algorithm that section risk differentiates.The size of section risk using Probability p (Rd | Ao,
Fv, Rc) it indicates, according to Bayesian formula,
In the present embodiment, the item of line between father node and child node in Bayesian network is determined according to expertise first
After part probability, then obtain 3 nodes prior probability, so that it may obtain section risk posterior probability p (Rd | Ao, Fv,
Rc)。
For simplify operation, each state value be with and without, respectively correspond 0 and 1, then:
Second part establishes the bayesian network structure and algorithm for determining each recognition element.
2.1, bayesian network structure and algorithm that object (Ao) is shed in identification are established.In this fact Example, the identification of object is shed
It is to be identified by vehicle region detour (Aa) and vehicle jogging (Va) two vehicle running characteristics.Shed probability existing for object
It is indicated with p (Ao | Aa, Va), according to Bayesian formula,
Wherein, p (Aa, Va)=p (Aa | Va) p (Va), p (Aa, Va | Ao)=p (Aa | Va, Ao) p (Va | Ao), it is above-mentioned general
Rate can be obtained from the historical data that trackside node is collected.
2.2, the bayesian network structure and algorithm of identification fault car (Fv) are established.In this fact Example, the knowledge of disabled vehicle
It is not to be identified by vehicle stopping (Vs) and vehicle jogging (Va) two vehicle running characteristics.Probability existing for fault car
P (Fv | Vs, Va) indicates, according to Bayesian formula,
Wherein, p (Vs, Va)=p (Vs | Va) p (Va), p (Vs, Va | Fv)=p (Vs | Va, Fv) p (Va | Fv), it is above-mentioned general
Rate can be obtained from the historical data that trackside node is collected.
2.3, the bayesian network structure and algorithm of identification road crowded (Rc) are established.In this fact Example, road is crowded
It is identified by vehicle stopping (Vs) and vehicle jogging (Va) two vehicle running characteristics is identified.It is general existing for fault car
Rate p (Rc | Vs, Va) indicates, according to Bayesian formula,
Wherein, p (Vs, Va)=p (Vs | Va) p (Va), p (Vs, Va | Rc)=p (Vs | Va, Rc) p (Va | Rc), it is above-mentioned general
Rate can be obtained from the historical data that trackside node is collected.
Part III, express highway section degree of danger sail knowledge and alarm.
3.1, the trackside node of highway on the way is deployed in by receiving vehicle on-board V2X device broadcasts in coverage area
The message comprising vehicle geographical location information sent;
3.2, trackside node is according to the location information of the vehicle received, and the form change in location obtained, so as to
Judged with detour, jogging and the probability of stopping to vehicle.Judgement is carried out by the way of counting in set period of time:
Wherein, CAaIt is the vehicle fleet size for occurring to deflect to the same direction in one place, CVsIt is to occur to stop phenomenon
Vehicle fleet size, CVaIt is that the slow vehicle fleet size of speed occurs, C is total vehicle fleet size of identification section.
3.3, when V2X trackside node device initializes, the experience number of setting can be passed through according to the electronic map in this section
According to then by formula (1) to formula (6), the danger classes in this available section;
3.4, after the risk probability in section is more than threshold value, warning information is passed through into vehicle of the V2X network into coverage area
It is broadcasted, it, can be using sound and display at that time to the danger of Dangerous Area after vehicle-mounted V2X equipment receives warning information
Factor and generation position are reminded;
3.5, by prior probability p (Rd) of the posterior probability p (Rd | Ao, Fv, Rc) as next calculating cycle at this time,
Make prediction that can more meet current character of road.
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.