CN105083333A - Subway traffic flow optimization control method - Google Patents

Abstract

The invention relates to a subway traffic flow optimization control method. The method comprises the following steps that a topology structural chart of a track traffic network is generated according to planned running parameters of trains; controllability and sensitivity of train flow are analyzed based on the topology structural chart; the multi-train non-collision running track is generated according to the planned running parameters of the trains; the running positions of the trains at some time are predicted at each sampling time based on the current running state of the trains and the historical position observation sequence, an observer from the continuous dynamic state of the trains to the discrete collision logic is established, and the continuous dynamic state is mapped into the collision state expressed through the discrete observation value; when it is possible for a system to violate the traffic control rule, the hybrid dynamic state of the subway traffic hybrid system is monitored, and alarm information is provided for a control center; and finally when the alarm information occurs, the robustness double-layer planning is carried out on the train running track through a self-adaptation control theory method, and a planning result is transmitted to the trains.

Description

The flow-optimized control method of a kind of subway transportation

Technical field

The present invention relates to the flow-optimized control method of a kind of subway transportation, particularly relate to a kind of flow-optimized control method of double-deck subway transportation based on Robust Strategies.

Background technology

Along with the expanding day of China's big and medium-sized cities scale, Traffic Systems is faced with the increasing pressure, and greatly developing Rail Transit System becomes the important means solving urban traffic congestion.Country's Eleventh Five-Year Plan outline is pointed out, big city with good conditionsi and group of cities area will using track traffic as Priority settings.China is just experiencing a unprecedented track traffic development peak load conditions, and some cities have turned to the construction of net by the construction of line, and urban mass transit network is progressively formed.At Rail traffic network and the intensive complex region of train flow, still the train interval dispensing mode adopting train operation plan to combine based on subjective experience demonstrates its lag gradually, be in particular in: the formulation of (1) train operation plan timetable also reckons without the impact of various enchancement factor, easily cause flow of traffic tactics to manage crowded, reduce the safety that traffic system is run; (2) train scheduling active side overweights the personal distance kept between single train, not yet rises to the macroscopic aspect of train flow being carried out to strategy management; (3) train allocation process depends on the subjective experience of a line dispatcher more, and the selection randomness of allocating opportunity is comparatively large, lacks scientific theory and supports; (4) dispatcher's less impact considering external interference factor of allotment means of using, robustness and the availability of train programs are poor.

The discussion object spininess of existing documents and materials to long-distance railway transportation, and still lacks system for the Scientific Regulation scheme of the city underground traffic system under large discharge, high density and closely-spaced condition of service.Train Coordinated Control Scheme under complicated road network condition of service needs calculate the running state of single vehicles in traffic network in region and optimize on strategic level, and implements collaborative planning to the flow of traffic be made up of multiple train; Pre-tactical level solves congestion problems by the subregional critical operational parameters in actv. monitoring mechanism adjustment traffic network top, and ensures the operating efficiency of all trains in this region; Tactical level then adjusts according to critical operational parameters the running state of relevant train, obtain single vehicles track optimizing scheme, change the headway management of train factors such as into considering train performance, scheduling rule and external environment in interior variable " microcosmic-macroscopic view-middle sight-microcosmic " Separation control mode from fixing manual type.

Summary of the invention

The technical problem to be solved in the present invention is to provide a kind of robustness and the flow-optimized control method of the good subway transportation of availability, and the method can strengthen the subject of programs formulation and can effectively prevent subway train from running conflict.

The technical scheme realizing the object of the invention is to provide the flow-optimized control method of a kind of subway transportation, comprises the steps:

Steps A, plan operational factor according to each train, the topology diagram of generator orbital traffic network;

Step B, topology diagram based on the Rail traffic network constructed by steps A, analyze controllability and sensivity two class feature of train flow;

Step C, plan operational factor according to each train, on the basis building Modeling Method for Train Dynamics, set up train operation conflict according to train operation conflict Coupling point and allocate model in advance, generate multiple row car Lothrus apterus running orbit;

Step D, at each sampling instant t, the running state current based on train and historical position observation sequence, predict the advanced positions in train certain moment following; Its detailed process is as follows:

Step D1, train track data pretreatment, with train at the stop position of originating station for the origin of coordinates, in each sampling instant, according to the train original discrete two-dimensional position sequence x=[x obtained ₁, x ₂..., x _n] and y=[y ₁, y ₂..., y _n], adopt first order difference method to carry out processing new train discrete location sequence Δ x=[the Δ x of acquisition to it ₁, Δ x ₂..., Δ x _n-1] and Δ y=[Δ y ₁, Δ y ₂..., Δ y _n-1], wherein Δ x _i=x _i+1-x _i, Δ y _i=y _i+1-y _i(i=1,2 ..., n-1);

Step D2, to train track data cluster, to new train discrete two-dimensional position sequence Δ x and Δ y after process, by setting cluster number M ', K-means clustering algorithm is adopted to carry out cluster to it respectively;

Step D3, HMM is utilized to carry out parameter training to the train track data after cluster, by the train operation track data Δ x after process and Δ y being considered as the aobvious observed value of hidden Markov models, upgrade period τ ' by setting hidden state number N ' and parameter, according to nearest T ' individual position detection value and adopt B-W algorithm to roll to obtain up-to-date HMM parameter lambda '; Specifically: because obtained train track sets data length is dynamic change, in order to the state variation of real-time tracking train track, be necessary at initial track HMM parameter lambda '=(π, A, B) basis is readjusted it, to infer the position of train in certain moment following more accurately; Every period τ ', the individual observed value (o of the T ' according to up-to-date acquisition ₁, o ₂..., o _{t '}) to track HMM parameter lambda '=(π, A, B) reappraise;

Step D4, foundation HMM parameter, adopt the hidden state q corresponding to Viterbi algorithm acquisition current time observed value;

Step D5, every the period , according to the HMM parameter lambda of up-to-date acquisition '=(π, A, B) and nearest H conception of history measured value (o ₁, o ₂..., o _h), based on the hidden state q of train current time, at moment t, by setting prediction time domain h ', obtain the position prediction value O of future time period train;

Step e, setting up the dynamic continuously observer to discrete conflict logic from train, is the conflict situation that discrete observation value is expressed by the continuous dynamic mapping of subway transportation system; When system likely violates traffic control rule, to the Hybrid dynamics behavior implementing monitoring of subway transportation hybrid system, for control center provides warning information timely;

Step F, when warning information occurs, meet train physical property, region hold stream constraint and track traffic scheduling rule prerequisite under, by setting optimizing index function, Adaptive Control Theory method is adopted to carry out robust dual layer resist to train operation track, and program results is transferred to each train, each train receives and performs train collision avoidance instruction until each train all arrives it free terminal.

Further, the detailed process of steps A is as follows:

Steps A 1, extract the site information of stopping each train travelling process from the data bank of subway transportation control center;

Steps A 2, according to positive and negative two service directions, the site information that each train is stopped to be classified, and the same site on same service direction is merged;

Steps A 3, according to website amalgamation result, according to website space layout's form with straight line connect before and after multiple website.

Further, the detailed process of step B is as follows:

Step B1, the Traffic flux detection model built in single subsegment; Its detailed process is as follows:

Step B1.1, introducing state variable Ψ, input variable u and output variable Ω, wherein Ψ represents the train quantity of certain moment existence on connected section between website, it comprises single channel section and Multiple Sections two type, u represents the Operation Measures that track traffic dispatcher implements for certain section, as adjust train speed or change train in the station time etc., Ω represents the train quantity that certain period section is left;

Step B1.2, by by time discretization, set up shape as Ψ (t+ Δ t)=A ₁Ψ (t)+B ₁u (t) and Ω (t)=C ₁Ψ (t)+D ₁discrete time Traffic flux detection model in the single subsegment of u (t), wherein Δ t represents the sampling interval, and Ψ (t) represents the state vector of t, A ₁, B ₁, C ₁and D ₁represent the state-transition matrix of t, input matrix, output calculation matrix and direct transmission matrix respectively;

Step B2, the Traffic flux detection model built in many subsegments; Its detailed process is as follows:

Step B2.1, according to circuit space layout form and train flow historical statistical data, obtain the flow proportional parameter beta in cross link each subsegment;

Step B2.2, according to the discrete time Traffic flux detection model in flow proportional parameter and single subsegment, build shape as Ψ (t+ Δ t)=A ₁Ψ (t)+B ₁u (t) and Ω (t)=C ₁Ψ (t)+D ₁discrete time Traffic flux detection model in many subsegments of u (t);

Step B3, controllable factor matrix [B according to Controlling model ₁, A ₁b ₁..., A ₁ ^n-1b ₁] order and the relation of numerical value n, its controllability of qualitative analysis, according to the sensitivity coefficient matrix [C of Controlling model ₁(zI-A ₁) ^-1b ₁+ D ₁], its input and output sensivity of quantitative analysis, wherein n represents the dimension of state vector, I representation unit matrix, and z represents the element factor changed original discrete time Traffic flux detection model.

Further, the detailed process of step C is as follows:

Step C1, train status transfer modeling, train shows as the switching at runtime process between website along the process that track traffic road network runs, arrange according to the website in train operation plan, set up the Petri network model of single train switched and transferred between different website: E=(g, G, Pre, Post, m) be train section metastasis model, wherein g represents each sub-section between website, G represents the change-over point of train running speed state parameter, Pre and Post represents that front and back between each sub-section and website are to annexation respectively represent the operation section residing for train, wherein m represents model identification, Z ⁺represent Positive Integer Set;

The modeling of step C2, train full operation profile hybrid system, the operation of train between website is considered as continuous process, from the stressed situation of train, according to the kinetics equation of energy model derivation train in the different operation phase, in conjunction with external interference factor, set up about train in a certain operation phase speed v _gmapping function v _g=λ (T ₁, T ₂, H, R, α), wherein T ₁, T ₂, H, R and α represent tractive force of train, braking force of train, train resistance, train gravity and train status random fluctuation parameter respectively;

Step C3, adopt the mode mixing emulation to infer to solve train track, by by time subdivision, the distance of utilization state continually varying characteristic Recursive Solution any time train in a certain operation phase apart from initial rest position point, wherein J ₀for initial time train is apart from the voyage of initial rest position point, Δ τ is the numerical value of time window, and J (τ), for τ moment train is apart from the distance of initial rest position point, can infers thus and obtain single vehicles track;

Step C4, train, in station time probability distribution function modeling, for specific run circuit, by transferring the dwell time data of train at each station, obtain the dwell time probability distribution of different circuit different website condition Train;

The Lothrus apterus robust track allotment of step C5, multiple row car coupling, the time of conflict point is reached in advance according to each train, pass through Time segments division, at each sampling instant t, under the prerequisite incorporating random factor, according to scheduling rule, robust secondary planning is implemented to the train track not meeting personal distance requirement near conflict point.

Further, in step D, the value of cluster number M ' is 4, and the value of hidden state number N ' is 3, and it is 30 seconds that parameter upgrades period τ ', and T ' is 10, be 30 seconds, H is 10, and prediction time domain h ' is 300 seconds.

Further, the specific implementation process of step e is as follows:

Step e 1, construct conflict hypersurface collection of functions based on regulation rule: set up hypersurface collection of functions in order to reflect the contention situation of system, wherein, continuous function h relevant to single train in conflict hypersurface _ibe I type hypersurface, the continuous function h relevant to two trains _iIit is II type hypersurface;

Step e 2, set up by train continuous state to the observer of discrete conflict situation, build the safety rule collection d that need meet when train runs in traffic network _ij(t)>=d _min, wherein d _ijt () represents the actual interval of train i and train j in t, d _minrepresent the minimum safety interval between train;

Step e 3, based on person machine system theoretical and complication system hierarchical control principle, according to train operation pattern, build people at the real-time monitoring mechanism of the train of loop, the operation of guarantee system is in safe reachable set, design the discrete monitor from conflict to conflict Resolution means, when the discrete observation vector of observer shows that safety rule rally is breached, send corresponding warning information at once.

Further, the detailed process of step F is as follows:

Step F 1, analysis result based on step B and step e, determine the flow of traffic regulation measure specifically taked, and comprises the running velocity of adjustment train and/or adjustment train in station time two class measure, and adopt specified place and the opportunity of above regulation measure;

Step F 2, the termination reference point locations P of setting train collision avoidance planning, collision avoidance policy control time domain Θ, trajectory predictions time domain Υ;

Step F 3, operation conflict Resolution process model building, be considered as the inside and outside dual planning problem based on both macro and micro aspect by the operation conflict Resolution in above-listed for Rail traffic network workshop, wherein represent outer plan model, i.e. train flow flow-Density and distribution problem on track traffic road network, represent internal layer plan model, namely on track traffic section, the state of single vehicles adjusts problem; F, x ₁and u ₁the objective function of outer planning problem, state vector and decision vector respectively, G (x ₁, u ₁)≤0 is the constraint condition of outer planning, f, x ₂and u ₂the objective function of internal layer planning problem, state vector and decision vector respectively, g (x ₂, u ₂)≤0 is the constraint condition of internal layer planning, using the reference input that the outer program results of macroscopic aspect is planned as microcosmic point internal layer;

Step F 4, run the modeling of conflict Resolution variable bound, build and comprise adjustable train quantity a, train speed ω and train at variablees such as station time γ in interior both macro and micro constraint condition: the variable bound that wherein t need implement the section k of conflict Resolution can be described as: a _k(t)≤a _m, ω _k(t)≤ω _m, γ _k(t)≤γ _m, a _m, ω _m, γ _mbe respectively maximum adjustable train quantity, maximum train running speed and most long line car in the station time, this type of frees the constraint that variable can be subject to the aspects such as flow of traffic distribution, train physical property and personal distance;

Step F 5, Multi-objective Robust optimum road network flow allocation plan solves: based on cooperative collision avoidance trajectory planning thought, for different performance figure, by selecting different conflict Resolution objective functions, solving the multiple goal flow of traffic flowrate optimization allocation plan based on Euler's network model in flow of traffic operation macroscopic aspect and respectively controlling section in Rolling Planning interval, only implementing its first Optimal Control Strategy;

Step F 6, Multi-objective Robust optimum section train operation state adjustment: according to each section or zone flow configuration result, mix the single vehicles controlling quantity of evolutionary model and Lagrangian fit plan model acquisition optimum based on train operation, generate optimum single vehicles running orbit and respectively regulate and control train in Rolling Planning interval, only implement its first Optimal Control Strategy;

Step F 7, each train receive and perform train collision avoidance instruction;

Step F 8, in next sampling instant, repeat step F 5 to F7 free terminal until each train all arrives it.

Further, in step F 2, stopping reference point locations P is that the next one of train stops website, and the value of parameter Θ is 300 seconds, and the value of Υ is 300 seconds.

Further, the detailed process of step F 5 is as follows: order

$d_{\mathrm{it}}^2={\left|\right|P_i\left(t\right)-P_i^f\left|\right|}_2^2={(x_{\mathrm{it}}-x_i^f)}^2+{(y_{\mathrm{it}}-y_i^f)}^2,$

Wherein represent the distance between the t current position of train i and next website square, P _i(t)=(x _it, y _it) represent the two-dimensional coordinate value of t train i, represent the two-dimensional coordinate value of next stop website of train i, so the priority index of t train i can be set as:

${\mathrm{\λ}}_{\mathrm{it}}=100\frac{d_{\mathrm{it}}^{-2}}{\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{d}_{\mathrm{it}}^{-2}},$

Wherein n _trepresent train number t section existing conflict, from the implication of priority index, train is nearer apart from next website, and its priority is higher;

Setting optimizing index

$\begin{array}{c}{J}^{*}(u_1^{\left(t\right)},u_1^{(t+\mathrm{\Δt})},...,u_1^{(t+\mathrm{p\Δt})},...,u_{n_t}^{\left(t\right)},u_{n_t}^{(t+\mathrm{\Δt})},...,u_{n_t}^{(t+\mathrm{\Π\Δt})})\\ =\underset{s=1}{\overset{\mathrm{\Π}}{\mathrm{\Σ}}}\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{\mathrm{\λ}}_{\mathrm{it}}{\left|\right|P_i(t+\mathrm{s\Δt})-P_i^f\left|\right|}_2^2\\ =\underset{s=1}{\overset{\mathrm{\Π}}{\mathrm{\Σ}}}\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{(P_i(t+\mathrm{s\Δt})-P_i^f)}^T{Q}_{\mathrm{it}}(P_i(t+\mathrm{s\Δt})-P_i^f)\end{array},$

Wherein i ∈ I (t) represent train code and I (t)=1,2 ..., n _t, P _i(t+s Δ t) represents the position vector of train at moment (t+s Δ t), and ∏ represents control time, i.e. the time span of Future Trajectory planning from current time, u _irepresent the optimal control sequence of train i to be optimized, Q _itfor positive definite diagonal matrix, its diagonal element is the priority index λ of train i in t _it, and $Q_{\mathrm{it}}=\left[\begin{array}{cc}{\mathrm{\λ}}_{\mathrm{it}}& 0\\ 0& {\mathrm{\λ}}_{\mathrm{it}}\end{array}\right].$

The present invention has positive effect: the flow-optimized control method of (1) subway transportation of the present invention is under the prerequisite meeting track traffic control personal distance, based on the real-time position information of train, maintenance data excavates means and dynamically infers train track; According to track traffic regulation rule, alarm is implemented to the conflict that may occur, give each train planning conflict Resolution track according to train performance data and related constraint condition; When being configured timetable, considering the probability distribution of all kinds of random factors and the robustness of timetable that affect train, strengthening the availability of configuration result.

(2) the present invention is based on controllability and the susceptivity analysis result of Rail traffic network topological structure, can be subway transportation stream allotment the time, allotment place and allotment means selection scientific basis is provided, avoid the randomness that regulation and control scheme is chosen.

(3) the present invention is based on the scene monitoring mechanism of constructed " people is at loop ", effecting reaction can be made in time alternately to train inside continuous variable and the frequent of external discrete event, overcome the shortcoming of conventional open loop monitored off-line scheme.

(4) what the dual layer resist scheme of train flow of the present invention can not only reduce Optimal Control Problem solves dimension, the practicality of regulation and control scheme can also be strengthened, the model and algorithm overcome in existing document only pay close attention to train AT STATION to the time of sending out, and the defect of the control lacked when train is run on concrete railroad section and prediction.

(5) the present invention is based on constructed train operation track rolling forecast scheme, all kinds of disturbing factors in train real time execution can be incorporated in time, improve the accuracy of train trajectory predictions, overcome the shortcoming that Conventional Off-line prediction scheme accuracy rate is not high.

Accompanying drawing explanation

Fig. 1 is train flow analysis on Operating figure;

Fig. 2 is Lothrus apterus 3D robust track supposition figure;

Fig. 3 is that train operation state mixes monitoring figure;

Fig. 4 is that train operation conflict optimum frees figure;

Fig. 5 is the schematic diagram of the double-deck allocation plan of flow of traffic.

Detailed description of the invention

(embodiment 1)

The flow-optimized control system of a kind of subway transportation, comprise wire topologies generation module, data transmission module, car-mounted terminal module, control terminal module and track monitoring module, track monitoring module is collected the status information of train and is supplied to control terminal module.

Described control terminal module comprises following submodule:

Lothrus apterus Track Pick-up module before train operation: according to the Train operation plan table time of running, first set up Modeling Method for Train Dynamics, then sets up train operation conflict according to train operation conflict Coupling point and allocates model in advance, finally generate Lothrus apterus train operation track.

Train operation Track Pick-up a middle or short term module: the train real time status information provided according to track monitoring module, utilizes data mining model, infers the running orbit of train in future time period.

Train operation situation monitoring module: at each sampling instant t, based on the track estimation result of train, when likely occurring violating the situation of safety rule when between train, provides warning information to its dynamic behaviour implementing monitoring and for control terminal.

Train collision avoidance track optimizing module: when train operation situation monitoring module sends warning information, meet train physical property, region hold stream constraint and track traffic scheduling rule prerequisite under, by setting optimizing index function, adopt Adaptive Control Theory method to carry out robust dual layer resist by control terminal module to train operation track, and by data transmission module, program results is transferred to the execution of car-mounted terminal module.Train collision avoidance track optimizing module comprises internal layer planning and outer planning two class planning process.

Apply the flow-optimized control method of subway transportation of the flow-optimized control system of above-mentioned subway transportation, comprise the following steps:

Steps A, plan operational factor according to each train, the topology diagram of generator orbital traffic network; Its detailed process is as follows:

Steps A 1, extract the site information of stopping each train travelling process from the data bank of subway transportation control center;

Steps A 2, according to positive and negative two service directions, the site information that each train is stopped to be classified, and the same site on same service direction is merged;

Steps A 3, according to website amalgamation result, according to website space layout's form with straight line connect before and after multiple website.

Step B, topology diagram based on the Rail traffic network constructed by steps A, analyze controllability and sensivity two class feature of train flow; Its detailed process is as follows:

Step B1, see Fig. 1, build the Traffic flux detection model in single subsegment; Its detailed process is as follows:

Step B1.1, introducing state variable Ψ, input variable u and output variable Ω, wherein Ψ represents the train quantity of certain moment existence on connected section between website, it comprises single channel section and Multiple Sections two type, u represents the Operation Measures that track traffic dispatcher implements for certain section, as adjust train speed or change train in the station time etc., Ω represents the train quantity that certain period section is left;

Step B1.2, by by time discretization, set up shape as Ψ (t+ Δ t)=A ₁Ψ (t)+B ₁u (t) and Ω (t)=C ₁Ψ (t)+D ₁discrete time Traffic flux detection model in the single subsegment of u (t), wherein Δ t represents the sampling interval, and Ψ (t) represents the state vector of t, A ₁, B ₁, C ₁and D ₁represent the state-transition matrix of t, input matrix, output calculation matrix and direct transmission matrix respectively;

Step B2, the Traffic flux detection model built in many subsegments; Its detailed process is as follows:

Step B2.1, according to circuit space layout form and train flow historical statistical data, obtain the flow proportional parameter beta in cross link each subsegment;

Step B2.2, according to the discrete time Traffic flux detection model in flow proportional parameter and single subsegment, build shape as Ψ (t+ Δ t)=A ₁Ψ (t)+B ₁u (t) and Ω (t)=C ₁Ψ (t)+D ₁discrete time Traffic flux detection model in many subsegments of u (t);

Step B3, controllable factor matrix [B according to Controlling model ₁, A ₁b ₁..., A ₁ ^n-1b ₁] order and the relation of numerical value n, its controllability of qualitative analysis, according to the sensitivity coefficient matrix [C of Controlling model ₁(zI-A ₁) ^-1b ₁+ D ₁], its input and output sensivity of quantitative analysis, wherein n represents the dimension of state vector, I representation unit matrix, and z represents the element factor changed original discrete time Traffic flux detection model;

Step C, see Fig. 2, according to the plan operational factor of each train, on the basis building Modeling Method for Train Dynamics, set up train operation conflict according to train operation conflict Coupling point and allocate model in advance, generate multiple row car Lothrus apterus running orbit; Its detailed process is as follows:

Step C1, train status transfer modeling, train shows as the switching at runtime process between website along the process that track traffic road network runs, arrange according to the website in train operation plan, set up the Petri network model of single train switched and transferred between different website: E=(g, G, Pre, Post, m) be train section metastasis model, wherein g represents each sub-section between website, G represents the change-over point of train running speed state parameter, Pre and Post represents that front and back between each sub-section and website are to annexation respectively represent the operation section residing for train, wherein m represents model identification, Z ⁺represent Positive Integer Set;

The modeling of step C2, train full operation profile hybrid system, the operation of train between website is considered as continuous process, from the stressed situation of train, according to the kinetics equation of energy model derivation train in the different operation phase, in conjunction with external interference factor, set up about train in a certain operation phase speed v _gmapping function v _g=λ (T ₁, T ₂, H, R, α), wherein T ₁, T ₂, H, R and α represent tractive force of train, braking force of train, train resistance, train gravity and train status random fluctuation parameter respectively;

Step C3, adopt the mode mixing emulation to infer to solve train track, by by time subdivision, the distance of utilization state continually varying characteristic Recursive Solution any time train in a certain operation phase apart from initial rest position point, wherein J ₀for initial time train is apart from the voyage of initial rest position point, Δ τ is the numerical value of time window, and J (τ), for τ moment train is apart from the distance of initial rest position point, can infers thus and obtain single vehicles track;

Step C4, train, in station time probability distribution function modeling, for specific run circuit, by transferring the dwell time data of train at each station, obtain the dwell time probability distribution of different circuit different website condition Train;

The Lothrus apterus robust track allotment of step C5, multiple row car coupling, the time of conflict point is reached in advance according to each train, pass through Time segments division, at each sampling instant t, under the prerequisite incorporating random factor, according to scheduling rule, robust secondary planning is implemented to the train track not meeting personal distance requirement near conflict point.

Step D, at each sampling instant t, the running state current based on train and historical position observation sequence, predict the advanced positions in train certain moment following; Its detailed process is as follows:

Step D1, train track data pretreatment, with train at the stop position of originating station for the origin of coordinates, in each sampling instant, according to the train original discrete two-dimensional position sequence x=[x obtained ₁, x ₂..., x _n] and y=[y ₁, y ₂..., y _n], adopt first order difference method to carry out processing new train discrete location sequence Δ x=[the Δ x of acquisition to it ₁, Δ x ₂..., Δ x _n-1] and Δ y=[Δ y ₁, Δ y ₂..., Δ y _n-1], wherein Δ x _i=x _i+1-x _i, Δ y _i=y _i+1-y _i(i=1,2 ..., n-1);

Step D2, to train track data cluster, to new train discrete two-dimensional position sequence Δ x and Δ y after process, by setting cluster number M ', K-means clustering algorithm is adopted to carry out cluster to it respectively;

Step D3, HMM is utilized to carry out parameter training to the train track data after cluster, by the train operation track data Δ x after process and Δ y being considered as the aobvious observed value of hidden Markov models, upgrade period τ ' by setting hidden state number N ' and parameter, according to nearest T ' individual position detection value and adopt B-W algorithm to roll to obtain up-to-date HMM parameter lambda '; Specifically: because obtained train track sets data length is dynamic change, in order to the state variation of real-time tracking train track, be necessary at initial track HMM parameter lambda '=(π, A, B) basis is readjusted it, to infer the position of train in certain moment following more accurately; Every period τ ', the individual observed value (o of the T ' according to up-to-date acquisition ₁, o ₂..., o _{t '}) to track HMM parameter lambda '=(π, A, B) reappraise;

Step D4, foundation HMM parameter, adopt the hidden state q corresponding to Viterbi algorithm acquisition current time observed value;

Step D5, every the period , according to the HMM parameter lambda of up-to-date acquisition '=(π, A, B) and nearest H conception of history measured value (o ₁, o ₂..., o _h), based on the hidden state q of train current time, at moment t, by setting prediction time domain h ', obtain the position prediction value O of future time period train;

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 upgrades period τ ', and T ' is 10, be 30 seconds, H is 10, and prediction time domain h ' is 300 seconds.

Step e, seeing Fig. 3, set up the dynamic continuously observer to discrete conflict logic from train, is the conflict situation that discrete observation value is expressed by the continuous dynamic mapping of subway transportation system; When system likely violates traffic control rule, to the Hybrid dynamics behavior implementing monitoring of subway transportation hybrid system, for control center provides warning information timely;

The specific implementation process of described step e is as follows:

Step e 1, construct conflict hypersurface collection of functions based on regulation rule: set up hypersurface collection of functions in order to reflect the contention situation of system, wherein, continuous function h relevant to single train in conflict hypersurface _ibe I type hypersurface, the continuous function h relevant to two trains _iIit is II type hypersurface;

Step e 2, set up by train continuous state to the observer of discrete conflict situation, build the safety rule collection d that need meet when train runs in traffic network _ij(t)>=d _min, wherein d _ijt () represents the actual interval of train i and train j in t, d _minrepresent the minimum safety interval between train;

Step e 3, based on person machine system theoretical and complication system hierarchical control principle, according to train operation pattern, build people at the real-time monitoring mechanism of the train of loop, the operation of guarantee system is in safe reachable set, design the discrete monitor from conflict to conflict Resolution means, when the discrete observation vector of observer shows that safety rule rally is breached, send corresponding warning information at once.

Step F, see Fig. 4, when warning information occurs, meet train physical property, region hold stream constraint and track traffic scheduling rule prerequisite under, by setting optimizing index function, Adaptive Control Theory method is adopted to carry out robust dual layer resist to train operation track, and program results is transferred to each train, each train receives and performs train collision avoidance instruction until each train all arrives it free terminal; Its detailed process is as follows:

Step F 1, analysis result based on step B3 and step e 3, determine the flow of traffic regulation measure specifically taked, and comprises the running velocity of adjustment train and/or adjustment train in station time two class measure, and adopt specified place and the opportunity of above regulation measure;

Step F 2, the termination reference point locations P of setting train collision avoidance planning, collision avoidance policy control time domain Θ, trajectory predictions time domain Υ;

Stopping reference point locations P is that the next one of train stops website, and the value of parameter Θ is 300 seconds, and the value of Υ is 300 seconds;

Step F 3, operation conflict Resolution process model building, be considered as the inside and outside dual planning problem based on both macro and micro aspect by the operation conflict Resolution in above-listed for Rail traffic network workshop, see Fig. 5, wherein represent outer plan model, i.e. train flow flow-Density and distribution problem on track traffic road network, represent internal layer plan model, namely on track traffic section, the state of single vehicles adjusts problem; F, x ₁and u ₁the objective function of outer planning problem, state vector and decision vector respectively, G (x ₁, u ₁)≤0 is the constraint condition of outer planning, f, x ₂and u ₂the objective function of internal layer planning problem, state vector and decision vector respectively, g (x ₂, u ₂)≤0 is the constraint condition of internal layer planning, using the reference input that the outer program results of macroscopic aspect is planned as microcosmic point internal layer;

Step F 4, run the modeling of conflict Resolution variable bound, build and comprise adjustable train quantity a, train speed ω and train at variablees such as station time γ in interior both macro and micro constraint condition: the variable bound that wherein t need implement the section k of conflict Resolution can be described as: a _k(t)≤a _m, ω _k(t)≤ω _m, γ _k(t)≤γ _m, a _m, ω _m, γ _mbe respectively maximum adjustable train quantity, maximum train running speed and most long line car in the station time, this type of frees the constraint that variable can be subject to the aspects such as flow of traffic distribution, train physical property and personal distance;

Step F 5, Multi-objective Robust optimum road network flow allocation plan solves: based on cooperative collision avoidance trajectory planning thought, for different performance figure, by selecting different conflict Resolution objective functions, solving the multiple goal flow of traffic flowrate optimization allocation plan based on Euler's network model in flow of traffic operation macroscopic aspect and respectively controlling section in Rolling Planning interval, only implementing its first Optimal Control Strategy; Its detailed process is as follows: order

$d_{\mathrm{it}}^2={\left|\right|P_i\left(t\right)-P_i^f\left|\right|}_2^2={(x_{\mathrm{it}}-x_i^f)}^2+{(y_{\mathrm{it}}-y_i^f)}^2,$

Wherein represent the distance between the t current position of train i and next website square, P _i(t)=(x _it, y _it) represent the two-dimensional coordinate value of t train i, next stops the two-dimensional coordinate value of website to represent train i, and the priority index of that so t train i can be set as:

${\mathrm{\λ}}_{\mathrm{it}}=100\frac{d_{\mathrm{it}}^{-2}}{\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{d}_{\mathrm{it}}^{-2}},$

Wherein n _trepresent train number t section existing conflict, from the implication of priority index, train is nearer apart from next website, and its priority is higher;

Setting optimizing index

$\begin{array}{c}{J}^{*}(u_1^{\left(t\right)},u_1^{(t+\mathrm{\Δt})},...,u_1^{(t+\mathrm{p\Δt})},...,u_{n_t}^{\left(t\right)},u_{n_t}^{(t+\mathrm{\Δt})},...,u_{n_t}^{(t+\mathrm{\Π\Δt})})\\ =\underset{s=1}{\overset{\mathrm{\Π}}{\mathrm{\Σ}}}\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{\mathrm{\λ}}_{\mathrm{it}}{\left|\right|P_i(t+\mathrm{s\Δt})-P_i^f\left|\right|}_2^2\\ =\underset{s=1}{\overset{\mathrm{\Π}}{\mathrm{\Σ}}}\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{(P_i(t+\mathrm{s\Δt})-P_i^f)}^T{Q}_{\mathrm{it}}(P_i(t+\mathrm{s\Δt})-P_i^f)\end{array},$

Wherein i ∈ I (t) represent train code and I (t)=1,2 ..., n _t, P _i(t+s Δ t) represents the position vector of train at moment (t+s Δ t), and ∏ represents control time, i.e. the time span of Future Trajectory planning from current time, u _irepresent the optimal control sequence of train i to be optimized, Q _itfor positive definite diagonal matrix, its diagonal element is the priority index λ of train i in t _it, and $Q_{\mathrm{it}}=\left[\begin{array}{cc}{\mathrm{\λ}}_{\mathrm{it}}& 0\\ 0& {\mathrm{\λ}}_{\mathrm{it}}\end{array}\right];$

Step F 6, Multi-objective Robust optimum section train operation state adjustment: according to each section or zone flow configuration result, mix the single vehicles controlling quantity of evolutionary model and Lagrangian fit plan model acquisition optimum based on train operation, generate optimum single vehicles running orbit and respectively regulate and control train in Rolling Planning interval, only implement its first Optimal Control Strategy;

Step F 7, each train receive and perform train collision avoidance instruction;

Step F 8, in next sampling instant, repeat step F 5 to F7 free terminal until each train all arrives it.

Obviously, above-described embodiment is only for example of the present invention is clearly described, and is not the restriction to embodiments of the present invention.For those of ordinary skill in the field, can also make other changes in different forms on the basis of the above description.Here exhaustive without the need to also giving all embodiments.And these belong to spirit institute's apparent change of extending out of the present invention or change and are still among protection scope of the present invention.

Claims (9)

1. the flow-optimized control method of subway transportation, is characterized in that comprising the steps:

Steps A, plan operational factor according to each train, the topology diagram of generator orbital traffic network;

Step B, topology diagram based on the Rail traffic network constructed by steps A, analyze controllability and sensivity two class feature of train flow;

Step C, plan operational factor according to each train, on the basis building Modeling Method for Train Dynamics, set up train operation conflict according to train operation conflict Coupling point and allocate model in advance, generate multiple row car Lothrus apterus running orbit;

Step D, at each sampling instant t, the running state current based on train and historical position observation sequence, predict the advanced positions in train certain moment following; Its detailed process is as follows:

Step D1, train track data pretreatment, with train at the stop position of originating station for the origin of coordinates, in each sampling instant, according to the train original discrete two-dimensional position sequence x=[x obtained ₁, x ₂..., x _n] and y=[y ₁, y ₂..., y _n], adopt first order difference method to carry out processing new train discrete location sequence Δ x=[the Δ x of acquisition to it ₁, Δ x ₂..., Δ x _n-1] and Δ y=[Δ y ₁, Δ y ₂..., Δ y _n-1], wherein Δ x _i=x _i+1-x _i, Δ y _i=y _i+1-y _i(i=1,2 ..., n-1);

Step D2, to train track data cluster, to new train discrete two-dimensional position sequence Δ x and Δ y after process, by setting cluster number M ', K-means clustering algorithm is adopted to carry out cluster to it respectively;

Step D3, HMM is utilized to carry out parameter training to the train track data after cluster, by the train operation track data Δ x after process and Δ y being considered as the aobvious observed value of hidden Markov models, upgrade period τ ' by setting hidden state number N ' and parameter, according to nearest T ' individual position detection value and adopt B-W algorithm to roll to obtain up-to-date HMM parameter lambda '; Specifically: because obtained train track sets data length is dynamic change, in order to the state variation of real-time tracking train track, be necessary at initial track HMM parameter lambda '=(π, A, B) basis is readjusted it, to infer the position of train in certain moment following more accurately; Every period τ ', the individual observed value (o of the T ' according to up-to-date acquisition ₁, o ₂..., o _{t '}) to track HMM parameter lambda '=(π, A, B) reappraise;

Step D4, foundation HMM parameter, adopt the hidden state q corresponding to Viterbi algorithm acquisition current time observed value;

Step D5, every the period according to the HMM parameter lambda of up-to-date acquisition '=(π, A, B) and nearest H conception of history measured value (o ₁, o ₂..., o _h), based on the hidden state q of train current time, at moment t, by setting prediction time domain h ', obtain the position prediction value O of future time period train;

Step e, setting up the dynamic continuously observer to discrete conflict logic from train, is the conflict situation that discrete observation value is expressed by the continuous dynamic mapping of subway transportation system; When system likely violates traffic control rule, to the Hybrid dynamics behavior implementing monitoring of subway transportation hybrid system, for control center provides warning information timely;

Step F, when warning information occurs, meet train physical property, region hold stream constraint and track traffic scheduling rule prerequisite under, by setting optimizing index function, Adaptive Control Theory method is adopted to carry out robust dual layer resist to train operation track, and program results is transferred to each train, each train receives and performs train collision avoidance instruction until each train all arrives it free terminal.

2. the flow-optimized control method of a kind of subway transportation according to claim 1, is characterized in that: the detailed process of steps A is as follows:

Steps A 1, extract the site information of stopping each train travelling process from the data bank of subway transportation control center;

Steps A 2, according to positive and negative two service directions, the site information that each train is stopped to be classified, and the same site on same service direction is merged;

Steps A 3, according to website amalgamation result, according to website space layout's form with straight line connect before and after multiple website.

3. the flow-optimized control method of a kind of subway transportation according to claim 1, is characterized in that: the detailed process of step B is as follows:

Step R1, the Traffic flux detection model built in single subsegment; Its detailed process is as follows:

Step R1.1, introducing state variable Ψ, input variable u and output variable Ω, wherein Ψ represents the train quantity of certain moment existence on connected section between website, it comprises single channel section and Multiple Sections two type, u represents the Operation Measures that track traffic dispatcher implements for certain section, as adjust train speed or change train in the station time etc., Ω represents the train quantity that certain period section is left;

Step R1.2, by by time discretization, set up shape as Ψ (t+ Δ t)=A ₁Ψ (t)+B ₁u (t) and Ω (t)=C ₁Ψ (t)+D ₁discrete time Traffic flux detection model in the single subsegment of u (t), wherein Δ t represents the sampling interval, and Ψ (t) represents the state vector of t, A ₁, B ₁, C ₁and D ₁represent the state-transition matrix of t, input matrix, output calculation matrix and direct transmission matrix respectively;

Step B2, the Traffic flux detection model built in many subsegments; Its detailed process is as follows:

Step B2.1, according to circuit space layout form and train flow historical statistical data, obtain the flow proportional parameter beta in cross link each subsegment;

Step B2.2, according to the discrete time Traffic flux detection model in flow proportional parameter and single subsegment, build shape as Ψ (t+ Δ t)=A ₁Ψ (t)+B ₁u (t) and Ω (t)=C ₁Ψ (t)+D ₁discrete time Traffic flux detection model in many subsegments of u (t);

Step B3, controllable factor matrix [B according to Controlling model ₁, A ₁b ₁..., A ₁ ^n-1b ₁] order and the relation of numerical value n, its controllability of qualitative analysis, according to the sensitivity coefficient matrix [C of Controlling model ₁(zI-A ₁) ^-1b ₁+ D ₁], its input and output sensivity of quantitative analysis, wherein n represents the dimension of state vector, I representation unit matrix, and z represents the element factor changed original discrete time Traffic flux detection model.

4. the flow-optimized control method of a kind of subway transportation according to claim 1, is characterized in that: the detailed process of step C is as follows:

Step C1, train status transfer modeling, train shows as the switching at runtime process between website along the process that track traffic road network runs, arrange according to the website in train operation plan, set up the Petri network model of single train switched and transferred between different website: E=(g, G, Pre, Post, m) be train section metastasis model, wherein g represents each sub-section between website, G represents the change-over point of train running speed state parameter, Pre and Post represents that front and back between each sub-section and website are to annexation respectively represent the operation section residing for train, wherein m represents model identification, Z ⁺represent Positive Integer Set;

The modeling of step C2, train full operation profile hybrid system, the operation of train between website is considered as continuous process, from the stressed situation of train, according to the kinetics equation of energy model derivation train in the different operation phase, in conjunction with external interference factor, set up about train in a certain operation phase speed v _gmapping function v _g=λ (T ₁, T ₂, H, R, α), wherein T ₁, T ₂, H, R and α represent tractive force of train, braking force of train, train resistance, train gravity and train status random fluctuation parameter respectively;

Step C3, adopt the mode mixing emulation to infer to solve train track, by by time subdivision, the distance of utilization state continually varying characteristic Recursive Solution any time train in a certain operation phase apart from initial rest position point, wherein J ₀for initial time train is apart from the voyage of initial rest position point, Δ τ is the numerical value of time window, and J (τ), for τ moment train is apart from the distance of initial rest position point, can infers thus and obtain single vehicles track;

Step C4, train, in station time probability distribution function modeling, for specific run circuit, by transferring the dwell time data of train at each station, obtain the dwell time probability distribution of different circuit different website condition Train;

The Lothrus apterus robust track allotment of step C5, multiple row car coupling, the time of conflict point is reached in advance according to each train, pass through Time segments division, at each sampling instant t, under the prerequisite incorporating random factor, according to scheduling rule, robust secondary planning is implemented to the train track not meeting personal distance requirement near conflict point.

5. the flow-optimized control method of a kind of subway transportation according to claim 1, is characterized in that: in step D, and the value of cluster number M ' is 4, and the value of hidden state number N ' is 3, and it is 30 seconds that parameter upgrades period τ ', and T ' is 10, be 30 seconds, H is 10, and prediction time domain h ' is 300 seconds.

6. the flow-optimized control method of a kind of subway transportation according to claim 1, is characterized in that: the specific implementation process of step e is as follows:

Step e 1, construct conflict hypersurface collection of functions based on regulation rule: set up hypersurface collection of functions in order to reflect the contention situation of system, wherein, continuous function h relevant to single train in conflict hypersurface _ibe I type hypersurface, the continuous function h relevant to two trains _iIit is II type hypersurface;

Step e 2, set up by train continuous state to the observer of discrete conflict situation, build the safety rule collection d that need meet when train runs in traffic network _ij(t)>=d _min, wherein d _ijt () represents the actual interval of train i and train j in t, d _minrepresent the minimum safety interval between train;

Step e 3, based on person machine system theoretical and complication system hierarchical control principle, according to train operation pattern, build people at the real-time monitoring mechanism of the train of loop, the operation of guarantee system is in safe reachable set, design the discrete monitor from conflict to conflict Resolution means, when the discrete observation vector of observer shows that safety rule rally is breached, send corresponding warning information at once.

7. the flow-optimized control method of a kind of subway transportation according to claim 1, is characterized in that: the detailed process of step F is as follows:

Step F 1, analysis result based on step B and step e, determine the flow of traffic regulation measure specifically taked, and comprises the running velocity of adjustment train and/or adjustment train in station time two class measure, and adopt specified place and the opportunity of above regulation measure;

Termination reference point locations P, collision avoidance policy control time domain Θ, the trajectory predictions time domain of step F 2, setting train collision avoidance planning

Step F 3, operation conflict Resolution process model building, be considered as the inside and outside dual planning problem based on both macro and micro aspect by the operation conflict Resolution in above-listed for Rail traffic network workshop, wherein represent outer plan model, i.e. train flow flow-Density and distribution problem on track traffic road network, represent internal layer plan model, namely on track traffic section, the state of single vehicles adjusts problem; F, x ₁and u ₁the objective function of outer planning problem, state vector and decision vector respectively, G (x ₁, u ₁)≤0 is the constraint condition of outer planning, f, x ₂and u ₂the objective function of internal layer planning problem, state vector and decision vector respectively, g (x ₂, u ₂)≤0 is the constraint condition of internal layer planning, using the reference input that the outer program results of macroscopic aspect is planned as microcosmic point internal layer;

Step F 4, run the modeling of conflict Resolution variable bound, build and comprise adjustable train quantity a, train speed ω and train at variablees such as station time γ in interior both macro and micro constraint condition: the variable bound that wherein t need implement the section k of conflict Resolution can be described as: a _k(t)≤a _m, ω _k(t)≤ω _m, γ _k(t)≤γ _m, a _m, ω _m, γ _mbe respectively maximum adjustable train quantity, maximum train running speed and most long line car in the station time, this type of frees the constraint that variable can be subject to the aspects such as flow of traffic distribution, train physical property and personal distance;

Step F 5, Multi-objective Robust optimum road network flow allocation plan solves: based on cooperative collision avoidance trajectory planning thought, for different performance figure, by selecting different conflict Resolution objective functions, solving the multiple goal flow of traffic flowrate optimization allocation plan based on Euler's network model in flow of traffic operation macroscopic aspect and respectively controlling section in Rolling Planning interval, only implementing its first Optimal Control Strategy;

Step F 6, Multi-objective Robust optimum section train operation state adjustment: according to each section or zone flow configuration result, mix the single vehicles controlling quantity of evolutionary model and Lagrangian fit plan model acquisition optimum based on train operation, generate optimum single vehicles running orbit and respectively regulate and control train in Rolling Planning interval, only implement its first Optimal Control Strategy;

Step F 7, each train receive and perform train collision avoidance instruction;

Step F 8, in next sampling instant, repeat step F 5 to F7 free terminal until each train all arrives it.

8. the flow-optimized control method of a kind of subway transportation according to claim 7, is characterized in that: in step F 2, and stopping reference point locations P is that the next one of train stops website, and the value of parameter Θ is 300 seconds, value be 300 seconds.

9. the flow-optimized control method of a kind of subway transportation according to claim 7 or 8, is characterized in that: the detailed process of step F 5 is as follows: order

$d_{\mathrm{it}}^2={\left|\right|P_i\left(t\right)-P_i^f\left|\right|}_2^2={(x_{\mathrm{it}}-x_i^f)}^2+{(y_{\mathrm{it}}-y_i^f)}^2,$

Wherein represent the distance between the t current position of train i and next website square, P _i(t)=(x _it, y _it) represent the two-dimensional coordinate value of t train i, next stops the two-dimensional coordinate value of website to represent train i, and the priority index of that so t train i can be set as:

${\mathrm{\λ}}_{\mathrm{it}}=100\frac{d_{\mathrm{it}}^{-2}}{\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{d}_{\mathrm{it}}^{-2}},$

Wherein n _trepresent train number t section existing conflict, from the implication of priority index, train is nearer apart from next website, and its priority is higher;

Setting optimizing index

$\begin{array}{c}{J}^{*}(u_1^{\left(t\right)},u_1^{(t+\mathrm{\Δt})},...,u_1^{(t+\mathrm{p\Δt})},...,u_{n_t}^{\left(t\right)},u_{n_t}^{(t+\mathrm{\Δt})},...,u_{n_t}^{(t+\mathrm{\Π\Δt})})\\ =\underset{s=1}{\overset{\mathrm{\Π}}{\mathrm{\Σ}}}\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{\mathrm{\λ}}_{\mathrm{it}}{\left|\right|P_i(t+\mathrm{s\Δt})-P_i^f\left|\right|}_2^2\\ =\underset{s=1}{\overset{\mathrm{\Π}}{\mathrm{\Σ}}}\underset{i=1}{\overset{n_t}{\mathrm{\Σ}}}{(P_i(t+\mathrm{s\Δt})-P_i^f)}^T{Q}_{\mathrm{it}}(P_i(t+\mathrm{s\Δt})-P_i^f)\end{array},$

Wherein i ∈ I (t) represent train code and I (t)=1,2 ..., n _t, P _i(t+s Δ t) represents the position vector of train at moment (t+s Δ t), and ∏ represents control time, i.e. the time span of Future Trajectory planning from current time, u _irepresent the optimal control sequence of train i to be optimized, Q _itfor positive definite diagonal matrix, its diagonal element is the priority index λ of train i in t _it, and $Q_{\mathrm{it}}=\left[\begin{array}{cc}{\mathrm{\λ}}_{\mathrm{it}}& 0\\ 0& {\mathrm{\λ}}_{\mathrm{it}}\end{array}\right].$