CN101383090A - Floating vehicle information processing method under parallel road network structure - Google Patents

Abstract

The invention relates to a method for processing the information of float cars under the structure of a parallel road network, which is mainly applied to the service field of the intelligent dynamic traffic information. The method comprises the steps: step 1, an SVM (Support Vector Machine) sorter is adopted to judge the matching relationship between sampling points and roads, and map match is carried out; step 2, according to map match information, a heuristic route deduction algorithm is adopted to deduce possible running routes of a car, and according to the properties of road chains contained in the routes, whether the car runs in the structure of a parallel road network or not is judged; the shortest running route or a main running route and a subsidiary running route are selectively output in different circumstances, and the reliability of average running speed information provided by the car is estimated; step 3, the D-S (Dempster-Shafer) evidence reasoning theory based on classification is adopted to merge road condition information provided by all the float cars passing by a certain road chain in a current period and consider the reliability of the road condition information in a merging process so as to obtain the mathematical expectations of the average speed and the travel time of the float cars passing by the road chain in the current period. On the basis that the data quality of the prior float cars is kept unchanged, the invention realizes the acquisition of the real-time dynamic traffic information under the structure of a parallel road network.

Description

Floating vehicle information processing method under a kind of parallel road network structure

Technical field

The present invention relates to intelligent transportation system (Intelligent Transportation System, ITS) field is based on Floating Car (Float Car Data, FCD) dynamic information service system, the floating vehicle information processing method under particularly a kind of parallel road network structure.

Background technology

In view of based on the dynamic information service system of Floating Car in theory and technical advance, domestic scientific research institution has carried out the research practice activity based on the Floating Car technology in good time, and has obtained a series of scientific payoffss with independent intellectual property right.Chinese patent application numbers 200610112433.1, denomination of invention has proposed a kind of quick map-matching method that is used for complicated road road network organization and management for " a kind of quick map-matching method based on the small lattice road network institutional framework ", by with the gridding of road network mechanism, grid that can first positioned vehicle place when map match, again all roads in the grid are mated, thereby improved the efficient of map match.Chinese patent application 200610112606.X, denomination of invention has proposed a kind of didactic path Floating Car path culculating algorithm for " being used to handle the heuristic path culculating method of large scale floating vehicle data ", on the basis of map match, infer the possible driving trace of vehicle in conjunction with the road network topology structure, and with the average speed mark to the road of vehicle approach, as the foundation of judging this road conditions information.When Floating Car quantity reaches certain scale, have many Floating Car processes on a road in the same time period, and the traffic information that different vehicle provides may be also not quite identical.For this reason, number of patent application: 200610168271, name is called " a kind of traffic information fusion processing method and system " and has proposed a kind of traffic information fusion processing method, its that provides according to vehicle be the traffic information of a certain section road in the travel route alone, merges the comprehensive traffic information that entire road.

Under common road network structure, merge three steps by above-mentioned map match, path culculating and traffic information, can finish of the mapping of Floating Car raw data preferably to real-time road condition information.But for complicated road network structure such as main and side road and viaduct, existing Floating Car system just reveals gradually in the declining trend aspect the information processing accuracy.With Beijing is example, loop and backbone mostly adopt the main and side road parallel road network structure, spaced far between the main and side road is less than the limits of error of Floating Car GPS sampled point, quick map-matching algorithm mentioned above can only be according to the vertical line distance of sampled point to road, and the span of the angle of vehicle heading and road direction judges the matching relationship of sampled point and road simply, so be difficult to sampled point is accurately navigated on main road or the bypass.In this case, often there are many possible candidate's driving paths in vehicle in a period of time, and existing path culculating algorithm can't therefrom distinguish the true driving trace of vehicle.The accuracy of the traffic information that present traffic information blending algorithm does not provide vehicle is judged, so be difficult to guarantee the accuracy of traffic information fusion treatment under parallel road network structure.The vital role in city road network in view of loop and backbone needs a kind of new Floating Car information processing model at the main and side road parallel road network structure of design, and this model can be realized the compatible processing of common road network structure floating car data simultaneously.

Summary of the invention

The technical matters that the present invention solves is: overcome the deficiencies in the prior art, floating vehicle information processing method under a kind of parallel road network structure is provided, on the basis that existing floating car data quality remains unchanged, real-time dynamic information obtains under the realization parallel road network structure.

Technical solution of the present invention: the Floating Car information processing model under the parallel road network structure, realize by following steps:

The first step, employing svm classifier device is judged the matching relationship between sampled point and the road, carries out map match, and map matching result is submitted to follow-up heuristic path culculating algorithm;

Second step, match information according to the map, adopt the heuristic path culculating algorithm to infer the driving path that vehicle is possible, whether travel in parallel road network structure according to the road chain determined property vehicle that comprises the road chain in the path, the situation of branch is selected the shortest driving trace of output or two driving traces of main and side road, and the average overall travel speed information that vehicle provides is carried out the assessment of confidence level;

The 3rd step, employing is based on the D-S evidential reasoning theory of classification, the traffic information that all Floating Car of passing through a certain road chain in the current period are provided merges, and in the process that merges, consider the confidence level of these traffic informations, to obtain in the current period by the average velocity of this road chain and the mathematical expectation of hourage.

In the described first step, employing svm classifier device is accurately judged the matching relationship between sampled point and the road, and the process of carrying out map match is as follows:

(1) grid at positioned vehicle place, and obtain all interior highway sections of this grid;

(2) to each bar highway section, at first calculating sampling is put the angle angle apart from distance and sampled point direction and highway section direction in this highway section, if distance≤30 meter and angle≤30 degree are then concentrated the candidate matches highway section of adding sampled point to the highway section to;

(3) each bar highway section of concentrating for the candidate matches highway section, extract a five-tuple＜distance, angle, speed, state, link-type 〉, wherein the implication of distance and angle is above providing, and speed is the instantaneous velocity information of vehicle, and state is the carrying status information of vehicle, link-type has identified the grade of affiliated road, this highway section chain, and above-mentioned five-tuple is corresponding to the data characteristics vector of a four-dimension $\stackrel{\→}{s}=\⟨p_1,p_2,p_3,p_4\⟩,$ Wherein each dimension of vector all is the posterior probability of judging sampled point and highway section matching relationship, i.e. a p ₁=p (match|distance), p ₂=p (match|angle), p ₃=p (match|link-type, speed), p ₄=p (match|link-type, state), its basic meaning is for when certain characteristic condition is satisfied in sampled point and highway section, and sampled point matches the probable value on the chain of road;

(4), obtain the optimal classification normal vector and be according to the training result of svm classifier device

Threshold value is b, if $\stackrel{\→}{\mathrm{\ω}}\·\stackrel{\→}{s}+b\≥0,$ Think that then sampled point successfully matches on the highway section, otherwise think that it fails to match;

(5) highway section that will it fails to match is concentrated from candidate road section and is rejected, and will mate the highway section result and submit to follow-up path culculating algorithm.

In described second step, the process of heuristic path culculating algorithm is as follows:

(1) the path candidate estimation result of at first judging vehicle is concentrated several paths is arranged, if having only one, then exports this path, otherwise changes step (2);

(2) choose shortest path, main road path and bypass path, do not comprise bypass road chain in main road road chain or the bypass path, the output shortest path if do not comprise in the main road path; Otherwise change step (3);

(3) similarity in shortest path and main road path and bypass path is relatively respectively replaced the path similar to it with shortest path, and exports with the driving trace that another paths constitutes vehicle jointly;

(4) the driving trace Reliability of Information of output is assessed, the parameter of assessment comprises: the probability of the respective stretch map match that comprises in bar number, sampled point and the path of outgoing route, vehicle are issued the number percent that distance that time interval of being and vehicle travel accounts for the road chain length through highway section and traffic information on the chain of road.

Implementation procedure in described the 3rd step is as follows:

(1) according to D-S evidential reasoning theory, each bar record of floating vehicle travelling trace information data file all will be finished evidence to the effectively support of proposition distribution in the identification space as an evidence;

(2) for each bar road chain, the degree of support of each evidence to the primitive proposition calculated on evidence by the institute that is collected in this road chain road conditions information of sign in the current period;

(3), determine the primitive proposition collection of its support according to the degree of support of evidence to each primitive proposition;

(4), determine the supporting evidence collection of each primitive proposition of its correspondence for each bar road chain;

(5) evidence in each classification results is carried out the D-S evidence and synthesize, obtain one group of combination;

(6) the synthetic evidence of all classification is done weighted mean and obtain final synthetic card;

(7) calculate the support of final synthetic evidence to each primitive proposition;

(8) those are supported the velocity information of the evidence correspondence of final primitive proposition average, thus obtain in the current period by this road chain average velocity and hourage mathematical expectation.

The present invention's advantage compared with prior art is: the present invention has carried out profound data mining to existing Floating Car raw data, employing svm classifier device is judged the matching relationship between sampled point and the road, carry out map match, improved the precision of map match greatly; Simultaneously, routing strategy to Floating Car is adjusted, the selectivity the shortest driving trace of output or two driving traces of main and side road efficiently solve the blindness of original algorithm on routing and the risk of having avoided this blindness to bring under parallel road network structure; At last, application is carried out data fusion based on the D-S evidential reasoning theory of the classification of evidence, solved the insincere and information collision problem of information that the routing strategy adjustment brings, finally made the more original algorithm of Floating Car information processing accuracy under the parallel road network structure that significantly raising is arranged.

Description of drawings

Fig. 1 is the overall flow figure of the inventive method;

Fig. 2 is the road net data structural representation;

Fig. 3 carries out the map match process flow diagram for employing svm classifier device of the present invention;

Fig. 4 is the process flow diagram of heuristic path culculating algorithm of the present invention;

Fig. 5 is that the traffic information of the D-S evidential reasoning theory based on the classification of evidence of the present invention merges synoptic diagram.

Embodiment

As shown in Figure 1, step of the present invention is: (1) expands the data characteristics that is used to judge sampled point and road matching relation, by statistical method the span of data feature is carried out refinement, and the abundant data association between digging vehicle travelling characteristic and the map topology attribute, introduce the svm classifier device simultaneously and accurately judge matching relationship between sampled point and the road.(2) on the basis that obtains matching relationship between sampled point and the road, use heuristic path culculating algorithm mentioned above, obtain all possible candidate's driving path of vehicle in a period of time.For parallel road network structure, often more than one of candidate's driving path of vehicle, and the very difficult true driving trace of confirming vehicle.In this case, the shortest driving trace of selectivity output vehicle, or with main road path and bypass path simultaneously as the driving trace of vehicle, and coupling and path culculating result carry out the assessment of confidence level to driving trace information according to the map.(3) result of path culculating has reflected that vehicle is at its traffic information in travel route alone, at the traffic information fusing stage, with the road chain is unit, application is based on the D-S evidential reasoning theory of the classification of evidence, the traffic information that all Floating Car of passing through a certain road chain in the current period are provided merges, and in the process that merges, consider the confidence level of these traffic informations, the final mathematical expectation that obtains by the hourage of this road chain.

Concrete enforcement is as follows:

(1) data association between abundant digging vehicle travelling characteristic and the map topology attribute, design svm classifier device improves the precision of map match.Fig. 3 has provided improved map match flow process, to vehicle last time each sampled point in the sampling period all repeat following process.

The first step, the grid at positioned vehicle place, and obtain all interior highway sections of this grid.

(a kind of three layers of road net model that the present invention adopts are described at first here.As shown in Figure 2, road network structure is made up of node, highway section and road chain three floor notion.Simultaneously, the needs in order to locate have fast carried out gridding to road net model.

Node is a notion the most basic in the road network structure, is made up of latitude and longitude coordinates.According to role's difference, node can be divided into connective node and ordinary node.The general sign 1 of connective node) level crossing, 2) the road starting point, 3) the road terminal point, 4) category of roads change point etc., other nodes in the road network are referred to as ordinary node.3 ordinary nodes and 2 connective nodes of having listed among Fig. 2.

If there is a directed path in adjacent two nodes, be called a highway section, the highway section is represented is unidirectional straight line road in the road network, has attributes such as road section length, direction.4 highway sections dividing on the one-way road have been listed among Fig. 1.

If there is a directed path between two adjacent connective nodes, this directed path is defined as the road chain so.The highway section is represented is one-way road in the road network, may comprise some highway sections in the chain of road, identical with the highway section, and the road that expression can two way should be divided into two opposite road chains of direction according to the difference of direction.Listed the road chain of dividing on the one-way road 1 among Fig. 2, it comprises highway section 1, highway section 2, highway section 3 and highway section 4, and other road chains that link to each other with road chain 1, comprise road chain 2, road chain 3, road chain 4, road chain 5 and road chain 6.The road chain includes a lot of attributes, and wherein chain grade in road can identify this road chain and belongs to main road road chain, bypass road chain or other road chains)

Second step, to each bar highway section, at first calculating sampling is put the angle angle apart from distance and sampled point direction and highway section direction in this highway section, if distance≤30 meter and angle≤30 degree are then concentrated the candidate matches highway section of adding sampled point to the highway section to.

The 3rd step, for each concentrated bar highway section of candidate matches highway section, extract a five-tuple＜distance, angle, speed, state, link-type 〉, wherein the implication of distance and angle is above providing, and speed is the instantaneous velocity information of vehicle, state is the carrying status information of vehicle, and link-type has identified the grade of affiliated road, this highway section chain.Above-mentioned five-tuple is corresponding to the data characteristics vector of a four-dimension $\stackrel{\→}{s}=\⟨p_1,p_2,p_3,p_4\⟩,$ Wherein each dimension of vector all is the posterior probability of judging sampled point and highway section matching relationship, i.e. a p ₁=p (match|distance), p ₂=p (match|angle), p ₃=p (match|link-type, speed), p ₄=p (match|link-type, state).Its basic meaning is for when certain characteristic condition is satisfied in sampled point and highway section, and sampled point matches the probable value on the chain of road.The corresponding relation of eigenwert and posterior probability all is that the data mining work by off-line obtains.

In the 4th step,, obtain the optimal classification normal vector and be according to the training result of svm classifier device , threshold value is b.If $\stackrel{\→}{\mathrm{\ω}}\·\stackrel{\→}{s}+b\≥0$ , think that then sampled point successfully matches on the highway section, otherwise think that it fails to match.

In the 5th step, the highway section that it fails to match concentrated from candidate road section reject, and will mate the highway section result and submit to follow-up path culculating algorithm.

Certainly, be not that each bar record all needs the repeated grid location and obtains the process that all highway sections compare in the grid.Only need carry out grid location to article one record of Floating Car in each cycle, the map matching process of every record all only limits to some the highway sections that heuristic search is come out on last record matching result basis thereafter.

(2) according to the road chain attribute that comprises the road chain among the path culculating result, judge whether vehicle travels in parallel road network structure, export the possible driving trace of vehicle, and the average overall travel speed information that vehicle provides is carried out the assessment of confidence level, main flow process is shown in Figure 4.

Through didactic path culculating process, the discrete GPS sampled point of vehicle can combine with road network structure, forms some possible vehicle driving traces.The purpose of path selection module is to concentrate the possible driving trace of picking out vehicle from numerous path candidate.Under common road network structure, the shortest track of module output cumulative length; And under the main and side road parallel road network structure, the selectivity the shortest driving trace of output or output main road path and two driving traces in bypass path.

The road chain rank that provides of manufacturer can be divided into the road chain three kinds: main road road chain, bypass road chain and other road chains according to the map.In numerous path candidates, select the true driving trace of vehicle, not only need to consider the attribute of road chain, also will consider the length relation of vehicle driving trace.

The process of routing is as follows:

The first step, the path candidate estimation result of at first judging vehicle is concentrated several paths, if having only one, then exports this path, otherwise changes for second step;

Second step, choose shortest path, main road path (main road road chain number accounts for the path that total road chain is counted the ratio maximum) and bypass path, do not comprise bypass road chain in main road road chain or the bypass path if do not comprise in the main road path, the output shortest path; Otherwise changeed for the 3rd step;

The 3rd step, the similarity (the tolerance rule of similarity is the road chain number during two paths occur simultaneously) that compares shortest path and main road path and bypass path respectively, replace the path similar with shortest path, and export with the driving trace that another paths constitutes vehicle jointly to it.

In the 4th step, the driving trace Reliability of Information of output is assessed.The parameter of assessment comprises: the bar number of outgoing route (have only a paths export then confidence level is bigger), the probability of the respective stretch map match that comprises in sampled point and the path (is in the map matching process

), the vehicle time interval that issue is through highway section and traffic information (traffic information ageing), and the vehicle distance of travelling on the chain of road accounts for the number percent (coverage rate of traffic information) of road chain length.

(3) be unit with the road chain, the mean velocity information that provides through all Floating Car of this road chain in the current period is merged, the final mathematical expectation that obtains by the hourage of this road chain, main flow process is shown in Figure 5.

The first step, according to D-S evidential reasoning theory, each bar record of floating vehicle travelling trace information data file all will be finished evidence to the effectively support of proposition distribution in the identification space as an evidence.

(in improved Floating Car system, the traffic information grade is divided into three kinds: the Jam that blocks up, slowly Slow and unimpeded Unblocked, thus constitute identification space Θ={ Jam, Slow, Unblocked}.

The power set in identification space constitutes the proposition space $\begin{array}{c}{2}^{\mathrm{\Θ}}=\{\mathrm{\Φ},\left\{\mathrm{Jam}\right\},\left\{\mathrm{Slow}\right\},\left\{\mathrm{Unblocked}\right\},\{\mathrm{Jam}\∪\mathrm{Slow}\},\\ \{\mathrm{Slow}\∪\mathrm{Unblocked}\},\{\mathrm{Jam}\∪\mathrm{Unblocked}\},\\ \{\mathrm{Jam}\∪\mathrm{Slow}\∪\mathrm{Unblocked}\}\}\end{array},$ Wherein Φ is not with { Jam ∪ Unblocked} has physical significance, and { Jam ∪ Slow ∪ Unblocked} is abbreviated as { Unknown} hereinafter.

According to D-S evidential reasoning theory, for each evidence S _j, its support function to the proposition space satisfies: $m_j:2^{\mathrm{\Θ}}\→\left[\mathrm{0,1}\right];{\mathrm{\Σ}}_{A\∈2^{\mathrm{\Θ}}}{m}_j\left(A\right)=1;m_j\left(\mathrm{\Φ}\right)=0.$ In view of the assessment of finishing the path culculating information credibility, the branch of the present evidence support of effective aspect function of assessment can be mixed.Finish m according to following formula _j{ the support function of Unknown} distributes.

$m_j\left\{\mathrm{Unknown}\right\}=\underset{i=1}{\overset{I}{\mathrm{\Σ}}}{\mathrm{\α}}_i\·(1-f_i\left(S_j\right))$

Wherein, I=4, f _i(S _j), i=1,2,3,4th, evidence S _jThe valuation functions of four confidence level indexs that flow process (2) the 4th step is mentioned has f _i(S _j) ∈ [0,1].α _i, i=1,2,3,4th, to the weight of four confidence level indexs, and have ${\mathrm{\Σ}}_{i=1}^I{\mathrm{\α}}_i=1.$

At known m _j{ under the prerequisite of Unknown}, finish the support of other propositions is distributed through the instantaneous velocity speed of the chain of passing by on one's way (unit: km/hour) according to vehicle.Concrete allocation rule is as follows:

● 0<speed≤15:m _j({Jam})＝1-m _j({Unknown})；

15<speed≤20：

● m _j({Jam})＝0.7×(1-m _j({Unknown}))；m _j({Jam∪Slow})＝0.3×(1-m _j({Unknown}))；

20<speed≤25：

● m _j({Jam，Slow})＝0.3×(1-m _j({Unknown}))；m _j({Slow})＝0.7×(1-m _j({Unknown}))；

● 25<speed≤35:m _j({Slow})＝1-m _j({Unknown})；

35<speed≤40：

● m _j({Slow})＝0.7×(1-m _j({Unknown}))；；

m _j({Slow，Unblocked})＝0.3×(1-m _j({Unknown}))

45<speed≤45：

● m _j({Slow，Unblocked})＝0.3×(1-m _j({Unknown}))；；

m _j({Unblocked})＝0.7×(1-m _j({Unknown}))

● 45<speed:m _j({Unblocked})＝1-m _j({Unknown})。)

In second step, for each bar road chain, each evidence calculates on evidence to the primitive proposition degree of support of (only comprising the proposition of an element, i.e. the proposition that only is made of the element in the framework of identification) in the institute that is collected in this road chain road conditions information of sign in the current period.

If the proposition collection is { A _i, i=1,2 ..., M}, evidence collection are { S _j, j=1,2 ..., N}, primitive proposition collection is { B _k, k=1,2 ..., K}.Evidence S _jSupport to the primitive proposition is: $D_j\left(B_k\right)=\underset{A_i\&cap;B_k\&NotEqual;\mathrm{\&Phi;}}{\mathrm{\&Sigma;}}\frac{m_j\left(A_i\right)}{J\left(A_i\right)},$ Wherein, J (A wherein _i) be proposition A _iThe number of the primitive proposition that comprises.

In the 3rd step,, determine the primitive proposition collection of its support according to the degree of support of evidence to each primitive proposition.

If evidence S _jThe maximum primitive attribute of supporting is: $B_{j,\max }=\underset{k}{\max }\left(D_j\left(B_k\right)\right)$ , in order to reduce the susceptibility of synthetic result, definition threshold value G=0.5 to the evidence subtle change.Evidence S _jThe primitive proposition collection of supporting is ${\mathrm{\&Omega;}}_j=\left\{B_k\right|\frac{D_j\left(B_k\right)}{D_j\left(B_{k,\max }\right)}\&GreaterEqual;G\},k=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;K.$

In the 4th step,, determine the supporting evidence collection of each primitive proposition of its correspondence to each bar road chain.For primitive proposition B _k, k=1,2 ... K, support its evidence collection to be: $T_{B_k}=\left\{S_j\right|B_k\&Element;{\mathrm{\&Omega;}}_j,j=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;N\}.$ The definitions set class: ${\mathrm{\&Gamma;}}_T=\{T_{B_k},k=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;K\},$ ${\mathrm{\&Gamma;}}_D=\left\{T_{B_k}\right|T_{B_k}\&SubsetEqual;T_{B_l}\},k,l\&Element;\{\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;K\},$ From Γ _TIn leave out Γ _D: Γ=Γ _T/ Γ _DEach element among the collection class Γ is all represented the result of a classification of evidence, utilizes this strategy can effectively avoid the bigger evidence branch of two conflicts within one group, thereby has avoided the synthetic of conflicting evidence.

The 5th step, the evidence in each classification results is carried out the D-S evidence synthesize, obtain one group of combined evidence $m_f^l,l=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;L,$ L≤K wherein.

D-S evidence composition rule is (be fused to example with two evidences, fusion results and the 3rd evidence of preceding two evidences can be merged when three evidences merge, many evidences merge analogizes):

$m\left(A\right)=\frac{1}{1-k}{\mathrm{\&Sigma;}}_{A_i,B_j,A_i\&cap;B_j=A}{m}_1\left(A_i\right)m_2\left(B_j\right),$ Wherein, m ₁And m ₂Be respectively that two evidences or two evidences are gathered the support function after the framework of identification independent judgment, $k={\mathrm{\&Sigma;}}_{A_i,B_j,A_i\&cap;B_j=\mathrm{\&Phi;}}{m}_1\left(A_i\right)m_2\left(B_j\right)$ Be the conflict factor, reflected the degree of conflicting between the evidence.

In the 6th step, the synthetic evidence of all classification is done weighted mean obtains final synthetic evidence: $m_F=\underset{l}{\mathrm{\&Sigma;}}{d}_l{m}_f^l.$ Wherein, d _l, l=1,2 ..., L is the normalization weighting factor of the synthetic evidence of each classification.This weighting factor is got the average of the degree of confidence of concentrated each evidence of primitive proposition supporting evidence on the road chain.For the evidence that is subordinated in the individual paths estimation result, its degree of confidence is higher, is made as 1; Otherwise if this evidence is subordinated to the path culculating result between certain two continuous sampled point, then its degree of confidence depends on the average of these two sampled points and corresponding road section map match.

In the 7th step, calculate final synthetic evidence m _FSupport to each primitive proposition: $\sup \left(B_k\right)=\underset{A_i\&cap;B_k\&NotEqual;\mathrm{\&Phi;}}{\mathrm{\&Sigma;}}\frac{m_F\left(A_i\right)}{J\left(A_i\right)},$ Final court verdict is $\left\{B_k\right|\sup \left(B_k\right)=\underset{K}{\max }\left\{\sup \left(B_k\right)\right\}\}$ Pairing primitive is assigned a topic, and this primitive proposition is made as the road condition grade of this road chain.

The 8th step, support the velocity information of the evidence correspondence of final primitive proposition to average those, thereby obtain in the current period by this road chain average velocity and hourage mathematical expectation.

The content that is not described in detail in the instructions of the present invention belongs to this area professional and technical personnel's known prior art.

The above only is a preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (4)

1, the floating vehicle information processing method under a kind of parallel road network structure is characterized in that step is as follows:

The first step, employing svm classifier device is judged the matching relationship between sampled point and the road, carries out map match, and map matching result is submitted to follow-up heuristic path culculating algorithm;

Second step, match information according to the map, adopt the heuristic path culculating algorithm to infer the driving path that vehicle is possible, whether travel in parallel road network structure according to the road chain determined property vehicle that comprises the road chain in the path, the situation of branch is selected the shortest driving trace of output or two driving traces of main and side road, and the average overall travel speed information that vehicle provides is carried out the assessment of confidence level;

The 3rd step, employing is based on the D-S evidential reasoning theory of classification, the traffic information that all Floating Car of passing through a certain road chain in the current period are provided merges, and in the process that merges, consider the confidence level of these traffic informations, to obtain in the current period by the average velocity of this road chain and the mathematical expectation of hourage.

2, the floating vehicle information processing method under the parallel road network structure according to claim 1 is characterized in that: in the described first step, employing svm classifier device is accurately judged the matching relationship between sampled point and the road, and the process of carrying out map match is as follows:

(1) grid at positioned vehicle place, and obtain all interior highway sections of this grid;

(2) to each bar highway section, at first calculating sampling is put the angle angle apart from distance and sampled point direction and highway section direction in this highway section, if distance≤30 meter and angle≤30 degree are then concentrated the candidate matches highway section of adding sampled point to the highway section to;

(3) each bar highway section of concentrating for the candidate matches highway section, extract a five-tuple＜distance, angle, speed, state, link-type 〉, wherein the implication of distance and angle is above providing, and speed is the instantaneous velocity information of vehicle, and state is the carrying status information of vehicle, link-type has identified the grade of affiliated road, this highway section chain, and above-mentioned five-tuple is corresponding to the data characteristics vector of a four-dimension $\stackrel{\&RightArrow;}{s}=\&lang;p_1,p_2,p_3,p_4\&rang;$ , wherein each dimension of vector all is the posterior probability of judging sampled point and highway section matching relationship, i.e. a p ₁=p (match|distance), p ₂=p (match|angle), p ₃=p (match|link-type, speed), p ₄=p (match|link-type, state), its basic meaning is for when certain characteristic condition is satisfied in sampled point and highway section, and sampled point matches the probable value on the chain of road;

(4), obtain the optimal classification normal vector and be according to the training result of svm classifier device

, threshold value is b, if $\stackrel{\&RightArrow;}{\mathrm{\&omega;}}\&CenterDot;\stackrel{\&RightArrow;}{s}+b\&GreaterEqual;0$ , think that then sampled point successfully matches on the highway section, otherwise think that it fails to match;

(5) highway section that will it fails to match is concentrated from candidate road section and is rejected, and will mate the highway section result and submit to follow-up path culculating algorithm.

3, the floating vehicle information processing method under the parallel road network structure according to claim 1 is characterized in that: in described second step, after adopting the heuristic path culculating algorithm to infer the path candidate collection, select the process of vehicle driving trace as follows:

(1) the path candidate estimation result of at first judging vehicle is concentrated several paths is arranged, if having only one, then exports this path, otherwise changes step (2);

(2) choose shortest path, main road path and bypass path, do not comprise bypass road chain in main road road chain or the bypass path, the output shortest path if do not comprise in the main road path; Otherwise change step (3);

(3) similarity in shortest path and main road path and bypass path is relatively respectively replaced the path similar to it with shortest path, and exports with the driving trace that another paths constitutes vehicle jointly;

(4) the driving trace Reliability of Information of output is assessed, the parameter of assessment comprises: the probability of the respective stretch map match that comprises in bar number, sampled point and the path of outgoing route, vehicle are issued the number percent that distance that time interval of being and vehicle travel accounts for the road chain length through highway section and traffic information on the chain of road.

4, the floating vehicle information processing method under the parallel road network structure according to claim 1 is characterized in that: the implementation procedure in described the 3rd step is as follows:

(1) according to D-S evidential reasoning theory, each bar record of floating vehicle travelling trace information data file all will be finished evidence to the effectively support of proposition distribution in the identification space as an evidence;

(2) for each bar road chain, the degree of support of each evidence to the primitive proposition calculated on evidence by the institute that is collected in this road chain road conditions information of sign in the current period;

(3), determine the primitive proposition collection of its support according to the degree of support of evidence to each primitive proposition;

(4), determine the supporting evidence collection of each primitive proposition of its correspondence for each bar road chain;

(5) evidence in each classification results is carried out the D-S evidence and synthesize, obtain one group of combination;

(6) the synthetic evidence of all classification is done weighted mean and obtain final synthetic evidence;

(7) calculate the support of final synthetic evidence to each primitive proposition;

(8) those are supported the velocity information of the evidence correspondence of final primitive proposition average, thus obtain in the current period by this road chain average velocity and hourage mathematical expectation.

Images