The real-time predicting method of subway train track based on Robust StrategiesThe application is application No. is 201510150289.X, and invention and created name is " the real-time prediction of subway train trackMethod ", the applying date are as follows: the divisional application of the application for a patent for invention on March 31st, 2015.
Technical field
The present invention relates to a kind of real-time predicting method of subway train track more particularly to a kind of ground based on Robust StrategiesThe real-time predicting method of iron train track.
Background technique
With being growing for China big and medium-sized cities scale, Traffic Systems are faced with the increasing pressure, energeticallyFeasibility of developing track transportation system becomes the important means for solving urban traffic congestion.National Eleventh Five-Year Plan guiding principle is it is to be noted, that there is itemThe big city and group of cities area of part are using rail traffic as Priority setting.China is just undergoing a unprecedented railRoad transport development peak period, some cities have been turned to the construction of net by the construction of line, urban mass transit network gradually shapeAt.In the complex region that Rail traffic network and train flow are intensive, is still combined using train operation plan and be based on subjective experienceTrain interval dispensing mode gradually show its backwardness, be in particular in: (1) formulation of train operation plan timetable is simultaneouslyThe influence for not considering various enchancement factors be easy to cause the management of traffic flow tactics crowded, reduces the safety of traffic system operationProperty；(2) train scheduling work lays particular emphasis on the personal distance for keeping single row workshop, not yet rises to and carries out strategic pipe to train flowThe macroscopic aspect of reason；(3) train allocation process depends on the subjective experience of a line dispatcher more, and the selection for deploying opportunity is randomProperty it is larger, lack scientific theory support；(4) the less shadow in view of external interference factor of the allotment means that dispatcher is usedIt rings, the robustness and availability of train programs are poor.
The discussion object spininess of existing documents and materials is to long-distance railway transportation, and for big flow, high density and closely-spacedThe Scientific Regulation scheme of city underground traffic system under service condition still lacks system design.Under complicated road network service conditionTrain Coordinated Control Scheme needed on strategic level in region in transportation network the operating status of single vehicles carry out calculate andOptimization, and collaborative planning is implemented to the traffic flow being made of multiple trains；And the operation conflict Resolution of multiple row vehicle is based on over the groundOn the basis of the prediction of iron train track, the operating status of train often not exclusively belongs to a certain specific motion state, at presentIt there is no the real-time predicting method of effective subway train track.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of robustness and availabilities preferably based on the ground of Robust StrategiesThe real-time predicting method of iron train track, this method are higher to the trajectory predictions precision of subway train.
Realize that the technical solution of the object of the invention is to provide a kind of the real-time pre- of the subway train track based on Robust StrategiesSurvey method, includes the following steps:
Step A, according to the plan operating parameter of each train, the topology diagram of Rail traffic network is generated；
Step B, the topology diagram based on Rail traffic network constructed by step A, analyze train flow controllability andTwo class feature of sensibility；
Step C, according to the plan operating parameter of each train, on the basis of constructing Modeling Method for Train Dynamics, according to columnVehicle operation conflict Coupling point establishes train running conflict and deploys model in advance, generates multiple row vehicle Lothrus apterus running track；
Step D, in each sampling instant t, based on the current operating status of train and historical position observation sequence, to trainThe advanced positions at certain following moment are predicted；Detailed process is as follows for it:
Step D1, train track data pre-process, using train initiating station stop position as coordinate origin, adopted eachThe sample moment, according to the acquired original discrete two-dimensional position sequence x=[x of train1,x2,...,xn] and y=[y1,y2,...,yn], processing is carried out to it using first-order difference method and obtains new train discrete location sequence △ x=[△ x1,△x2,...,△xn-1] and △ y=[△ y1,△y2,...,△yn-1], wherein △ xi=xi+1-xi,△yi=yi+1-yi(i=1,2 ..., n-1)；
Step D2, train track data is clustered, to train discrete two-dimensional position sequence △ x and △ y new after processing, is led toSetting cluster number M' is crossed, it is clustered respectively using genetic algorithm for clustering；
Step D3, parameter training is carried out using Hidden Markov Model to the train track data after cluster, by that will locateTrain operation track data △ x and △ y after reason is considered as the aobvious observation of hidden Markov models, by setting hidden state numberN' and parameter update period τ ', roll the newest hidden Ma Erke of acquisition according to T' nearest position detection value and using B-W algorithmHusband's model parameter λ '；Specifically: since train track sets data length obtained is dynamic change, in order in real time withThe state change of track train track, it is necessary to initial track Hidden Markov Model parameter lambda '=(π, A, B) on the basis of it is rightIt is readjusted, more accurately to speculate train in the position at certain following moment；T' every period τ ', according to newest acquisitionA observation (o1,o2,...,oT') to track Hidden Markov Model parameter lambda '=(π, A, B) reevaluated；
Step D4, it according to Hidden Markov Model parameter, is obtained corresponding to current time observation using Viterbi algorithmHidden state q；
Step D5, every the periodAccording to the Hidden Markov Model parameter lambda of newest acquisition '=(π, A, B) and nearest HA history observation (o1,o2,...,oH), the hidden state q based on train current time predicts time domain by setting in moment tH' obtains the position prediction value O of future time period train, arranges to roll supposition in each sampling instant to subway in future time periodThe track of vehicle.
Further, detailed process is as follows by step A:
Step A1, the website letter stopped in each train travelling process is extracted from the database of subway transportation control centreBreath；
Step A2, classify according to the site information that positive and negative two traffic directions stop each train, and will be sameSame site on one traffic direction merges；
Step A3, according to website amalgamation result, the multiple websites in front and back are connected according to the space layout form straight line of website.
Further, detailed process is as follows by step B:
Step Bl, the Traffic flux detection model in single subsegment is constructed；Detailed process is as follows for it:
Step Bl.1, state variable Ψ, input variable u and output variable Ω are introduced, wherein Ψ indicates phase link between websiteTrain quantity existing for certain moment in section, it includes single channel section and Multiple Sections two types, and u indicates that rail traffic dispatcher is directed toThe Operation Measures that certain section is implemented, such as adjustment train speed or change train in the station time, Ω indicates certain period sectionOn the train quantity left；
Step B1.2, by establishing time discretization shaped like Ψ (t+ △ t)=A1Ψ(t)+B1U (t) and Ω (t)=C1Ψ(t)+D1Discrete time Traffic flux detection model in the single subsegment of u (t), wherein △ t indicates sampling interval, Ψ (t) tableShow the state vector of t moment, A1、B1、C1And D1Respectively indicate the state-transition matrix, input matrix, output measurement square of t momentBattle array and direct transmission matrix；
Step B2, the Traffic flux detection model in more subsegments is constructed；Detailed process is as follows for it:
Step B2.1, according to route space layout form and train flow historical statistical data, each son of cross link is obtainedFlow proportional parameter beta in section；
Step B2.2, according to the discrete time Traffic flux detection model in flow proportional parameter and single subsegment, shape is constructedSuch as Ψ (t+ △ t)=A1Ψ(t)+B1U (t) and Ω (t)=C1Ψ(t)+D1Discrete time traffic flow control in more subsegments of u (t)Simulation；
Step B3, according to the controllable factor matrix [B of Controlling model1,A1B1,...,A1n-1B1] order and numerical value n relationship,Its controllability of qualitative analysis, according to the sensitivity coefficient matrix [C of Controlling model1(zI-A1)-1B1+D1], its input of quantitative analysis is defeatedSensibility out, wherein n indicates that the dimension of state vector, I indicate that unit matrix, z are indicated to original discrete time Traffic flux detectionThe element factor that model is converted.
Further, detailed process is as follows by step C:
Step C1, train status transfer modeling, train are shown as between website along the process that rail traffic road network is runSwitching at runtime process is arranged according to the website in train operation plan, establishes single train switched and transferred between different websitesPetri net model: E=(g, G, Pre, Post, m) is train section metastasis model, and wherein g indicates each sub- section between website, G tableShow that the transfer point of train running speed state parameter, Pre and Post respectively indicate the front and back between each sub- section and website to connectionRelationship,Indicate operation section locating for train, wherein m indicates model identification, Z+Indicate Positive Integer Set；
Step C2, the full operation profile hybrid system modeling of train, is considered as continuous process for operation of the train between website, fromThe stress situation of train is set out, and derives kinetics equation of the train in the different operation phase according to energy model, dry in conjunction with the external worldFactor is disturbed, is established about train in a certain operation phase speed vGMapping function vG=λ (T1,T2, H, R, α), wherein T1、T2、H, R and α respectively indicates tractive force of train, braking force of train, train resistance, train gravity and train status random fluctuation parameter；
Step C3, speculated by the way of mixing emulation and solve train track, by utilizing state consecutive variations for time subdivisionCharacteristic Recursive Solution any time train in distance of a certain operation phase away from initial rest position point,Wherein J0Voyage for initial time train away from initial rest position point, △ τ are the numerical value of time window, and J (τ) is τ moment trainDistance away from initial rest position point thereby it is assumed that obtain single-row wheel paths；
Step C4, train is in station time probability distribution function modeling, for specific run route, by transferring train eachThe dwell time data at station obtain the dwell time probability distribution of different route difference website condition Trains；
Step C5, the Lothrus apterus robust track allotment of multiple row vehicle coupling, reaches the time of conflict point in advance according to each train, passes throughTime segments division, it is discontented nearby to conflict point according to scheduling rule under the premise of incorporating random factor in each sampling instant tImplement robust secondary planning in the train track that sufficient personal distance requires.
Further, in step D, the value of cluster number M' is 4, and the value of hidden state number N' is 3, and parameter updates period τ 'It is 30 seconds, T' 10,It is 30 seconds, H 10, prediction time domain h' is 300 seconds.
The present invention has the effect of positive: (1) the real-time prediction of the subway train track of the invention based on Robust StrategiesMethod is real in prediction based on the real-time position information of train under the premise of meeting rail traffic control personal distanceThe specific run state of setting train before applying, maintenance data excavate means dynamic and speculate train track.(2) the present invention is based on institute's structuresThe train operation track rolling forecast scheme built can incorporate all kinds of disturbing factors in train real time execution in time, improve columnThe accuracy of wheel paths prediction, the disadvantage for overcoming Conventional Off-line prediction scheme accuracy not high.(3) the present invention is based on rail trafficsControllability and the sensitivity analysis of network topology structure avoid as a result, scientific basis can be provided for subway transportation stream trajectory predictionsThe randomness that prediction scheme is chosen.
Detailed description of the invention
Fig. 1 is train flow analysis on Operating figure；
Fig. 2 is that Lothrus apterus 3D robust track speculates figure.
Specific embodiment
(embodiment 1)
A kind of flow-optimized control system of subway transportation, including it is wire topologies generation module, data transmission module, vehicle-mountedTerminal module, controlling terminal module and track monitoring module, track monitoring module are collected the status information of train and are supplied toControlling terminal module.
The controlling terminal module includes following submodule:
Lothrus apterus track generation module before train operation: according to Train operation plan running schedule, train dynamics are initially set upModel is learned, then train running conflict is established according to train running conflict Coupling point and deploys model in advance, ultimately produces Lothrus apterus columnVehicle running track.
Train operation middle or short term track generation module: the train real time status information provided according to track monitoring module, benefitWith data mining model, thus it is speculated that the running track of train in future time period.
Train operation situation monitoring module: in each sampling instant t, the track estimation result based on train, when between trainIt is possible that its dynamic behaviour implementing monitoring and providing warning information when being in the presence of violating safety regulation for controlling terminal.
Train collision avoidance track optimizing module: when train operation situation monitoring module issues warning information, meeting trainUnder the premise of physical property, region hold stream constraint and rail traffic scheduling rule, by setting optimizing index function, use is adaptiveIt answers control theory method to carry out robust dual layer resist to train operation track by controlling terminal module, and passes through data transmission moduleProgram results are transferred to car-mounted terminal module to execute.Train collision avoidance track optimizing module includes internal layer planning and outer layer planning twoClass planning process.
Using the real-time prediction of the subway train track based on Robust Strategies of the flow-optimized control system of above-mentioned subway transportationMethod, comprising the following steps:
Step A, according to the plan operating parameter of each train, the topology diagram of Rail traffic network is generated；It is specificProcess is as follows:
Step A1, the website letter stopped in each train travelling process is extracted from the database of subway transportation control centreBreath；
Step A2, classify according to the site information that positive and negative two traffic directions stop each train, and will be sameSame site on one traffic direction merges；
Step A3, according to website amalgamation result, the multiple websites in front and back are connected according to the space layout form straight line of website.
Step B, the topology diagram based on Rail traffic network constructed by step A, analyze train flow controllability andTwo class feature of sensibility；Detailed process is as follows for it:
Step Bl, see Fig. 1, construct the Traffic flux detection model in single subsegment；Detailed process is as follows for it:
Step Bl.1, state variable Ψ, input variable u and output variable Ω are introduced, wherein Ψ indicates phase link between websiteTrain quantity existing for certain moment in section, it includes single channel section and Multiple Sections two types, and u indicates that rail traffic dispatcher is directed toThe Operation Measures that certain section is implemented, such as adjustment train speed or change train in the station time, Ω indicates certain period sectionOn the train quantity left；
Step B1.2, by establishing time discretization shaped like Ψ (t+ △ t)=A1Ψ(t)+B1U (t) and Ω (t)=C1Ψ(t)+D1Discrete time Traffic flux detection model in the single subsegment of u (t), wherein △ t indicates sampling interval, Ψ (t) tableShow the state vector of t moment, A1、B1、C1And D1Respectively indicate the state-transition matrix, input matrix, output measurement square of t momentBattle array and direct transmission matrix；
Step B2, the Traffic flux detection model in more subsegments is constructed；Detailed process is as follows for it:
Step B2.1, according to route space layout form and train flow historical statistical data, each son of cross link is obtainedFlow proportional parameter beta in section；
Step B2.2, according to the discrete time Traffic flux detection model in flow proportional parameter and single subsegment, shape is constructedSuch as Ψ (t+ △ t)=A1Ψ(t)+B1U (t) and Ω (t)=C1Ψ(t)+D1Discrete time traffic flow control in more subsegments of u (t)Simulation；
Step B3, according to the controllable factor matrix [B of Controlling model1,A1B1,...,A1n-1B1] order and numerical value n relationship,Its controllability of qualitative analysis, according to the sensitivity coefficient matrix [C of Controlling model1(zI-A1)-1B1+D1], its input of quantitative analysis is defeatedSensibility out, wherein n indicates that the dimension of state vector, I indicate that unit matrix, z are indicated to original discrete time Traffic flux detectionThe element factor that model is converted；
Step C, see Fig. 2, according to the plan operating parameter of each train, on the basis of constructing Modeling Method for Train Dynamics,Train running conflict is established according to train running conflict Coupling point and deploys model in advance, generates multiple row vehicle Lothrus apterus running track；ItsDetailed process is as follows:
Step C1, train status transfer modeling, train are shown as between website along the process that rail traffic road network is runSwitching at runtime process is arranged according to the website in train operation plan, establishes single train switched and transferred between different websitesPetri net model: E=(g, G, Pre, Post, m) is train section metastasis model, and wherein g indicates each sub- section between website, G tableShow that the transfer point of train running speed state parameter, Pre and Post respectively indicate the front and back between each sub- section and website to connectionRelationship,Indicate operation section locating for train, wherein m indicates model identification, Z+Indicate Positive Integer Set；
Step C2, the full operation profile hybrid system modeling of train, is considered as continuous process for operation of the train between website, fromThe stress situation of train is set out, and derives kinetics equation of the train in the different operation phase according to energy model, dry in conjunction with the external worldFactor is disturbed, is established about train in a certain operation phase speed vGMapping function vG=λ (T1,T2, H, R, α), wherein T1、T2、H, R and α respectively indicates tractive force of train, braking force of train, train resistance, train gravity and train status random fluctuation parameter；
Step C3, speculated by the way of mixing emulation and solve train track, by continuously being become using state by time subdivisionThe characteristic Recursive Solution any time train of change in distance of a certain operation phase away from initial rest position point,Wherein J0Voyage for initial time train away from initial rest position point, △ τ are the numerical value of time window, and J (τ) is τ moment trainDistance away from initial rest position point thereby it is assumed that obtain single-row wheel paths；
Step C4, train is in station time probability distribution function modeling, for specific run route, by transferring train eachThe dwell time data at station obtain the dwell time probability distribution of different route difference website condition Trains；
Step C5, the Lothrus apterus robust track allotment of multiple row vehicle coupling, reaches the time of conflict point in advance according to each train, passes throughTime segments division, it is discontented nearby to conflict point according to scheduling rule under the premise of incorporating random factor in each sampling instant tImplement robust secondary planning in the train track that sufficient personal distance requires.
Step D, in each sampling instant t, based on the current operating status of train and historical position observation sequence, to trainThe advanced positions at certain following moment are predicted；Detailed process is as follows for it:
Step D1, train track data pre-process, using train initiating station stop position as coordinate origin, adopted eachThe sample moment, according to the acquired original discrete two-dimensional position sequence x=[x of train1,x2,...,xn] and y=[y1,y2,...,yn], processing is carried out to it using first-order difference method and obtains new train discrete location sequence △ x=[△ x1,△x2,...,△xn-1] and △ y=[△ y1,△y2,...,△yn-1], wherein △ xi=xi+1-xi,△yi=yi+1-yi(i=1,2 ..., n-1)；
Step D2, train track data is clustered, to train discrete two-dimensional position sequence △ x and △ y new after processing, is led toSetting cluster number M' is crossed, it is clustered respectively using genetic algorithm for clustering；
Step D3, parameter training is carried out using Hidden Markov Model to the train track data after cluster, by that will locateTrain operation track data △ x and △ y after reason is considered as the aobvious observation of hidden Markov models, by setting hidden state numberN' and parameter update period τ ', roll the newest hidden Ma Erke of acquisition according to T' nearest position detection value and using B-W algorithmHusband's model parameter λ '；Specifically: since train track sets data length obtained is dynamic change, in order in real time withThe state change of track train track, it is necessary to initial track Hidden Markov Model parameter lambda '=(π, A, B) on the basis of it is rightIt is readjusted, more accurately to speculate train in the position at certain following moment；T' every period τ ', according to newest acquisitionA observation (o1,o2,...,oT') to track Hidden Markov Model parameter lambda '=(π, A, B) reevaluated；
Step D4, it according to Hidden Markov Model parameter, is obtained corresponding to current time observation using Viterbi algorithmHidden state q；
Step D5, every the periodAccording to the Hidden Markov Model parameter lambda of newest acquisition '=(π, A, B) and nearest HA history observation (o1,o2,...,oH), the hidden state q based on train current time predicts time domain by setting in moment tH' obtains the position prediction value O of future time period train, arranges to roll supposition in each sampling instant to subway in future time periodThe track of vehicle；
The value of above-mentioned cluster number M' is 4, and the value of hidden state number N' is 3, and it is 30 seconds that parameter, which updates period τ ', and T' is10,It is 30 seconds, H 10, prediction time domain h' is 300 seconds.
Obviously, the above embodiment is merely an example for clearly illustrating the present invention, and is not to of the inventionThe restriction of embodiment.For those of ordinary skill in the art, it can also be made on the basis of the above descriptionIts various forms of variation or variation.There is no necessity and possibility to exhaust all the enbodiments.And these belong to this hairThe obvious changes or variations that bright spirit is extended out are still in the protection scope of this invention.