CN106295851A - A kind of truck information service and dispatching patcher and method of work thereof - Google Patents

Abstract

The invention discloses a kind of truck information service and dispatching patcher and method of work thereof, described system includes that truck client and terminal information are collected processing platform, described truck client and terminal information collection processing platform and connected by wireless network.Described method comprises the following steps: reservation truck collection ETA estimated time of arrival；Collection processes truck collection port subscription information；Forecast set card time of advent；Optimize stockyard secondary mould turnover order.Present invention employs truck client and terminal information collects processing platform, it is achieved that truck information service and the function of scheduling, improve dock operation efficiency, reduce harbour operation cost.Present invention employs truck client, truck driver uses the truck Ji Gang reservation APP on mobile terminal to carry out Ji Gang reservation, it is to avoid the problems such as the harbour caused blocks up of entering a port are concentrated in truck peak period.Present invention employs the truck prognoses system time of advent, more accurate collection can be doped and snap into ETA estimated time of arrival and collection snaps into port sequencing.

Description

A kind of truck information service and dispatching patcher and method of work thereof

Technical field

The present invention relates to the job scheduling decision-making technic of a kind of port and pier, the information service of a kind of wharf container-truck With dispatching method.

Background technology

The important evidence that outer truck (container truck outside harbour) is dispatched to port information as dock operation, is widely used in In the scheduling decision of harbour plant equipment, resource distribution, its utilization obtaining mode, information quality and information and container code Head production operation efficiency is closely bound up.Present stage, outside container terminal, truck enters a port at random, along with the traffic of a port increases, and lock Mouth truck queuing jam is serious, exacerbates the environmental pollution of port district, causes the traffic congestion in port district, reduces multimodal transport effect Rate, reduces dock operation efficiency, adds dock operation cost.For the pick-up operation process of harbour, its efficiency is by mould turnover Amount impact, is also distributed by the task of straddle truck and mobile route is affected.It is uncertain that outer collection snaps into ETA estimated time of arrival, carrying of outer truck Case order is not mated with goal box cause a large amount of mould turnover, outer truck to wait in the position of storing up in stockyard, affects the shifting of straddle truck yet Dynamic path and operating cost.If can obtain in time outer collection accurately snap into port information just can be in advance to suitcase mould turnover operation Process is optimized, adjusts, and improves the working performance of harbour.

Summary of the invention

For solve the above-mentioned problems in the prior art, present invention one to be designed can improve dock operation efficiency, Reduce truck information service and dispatching patcher and the method for work thereof of harbour operation cost.

To achieve these goals, technical scheme is as follows: a kind of truck information service and dispatching patcher, including Truck client and terminal information are collected processing platform, described truck client and terminal information collection processing platform and are passed through nothing Gauze network connects；

Installing truck Ji Gang in described truck client and preengage APP, truck driver uses the truck collection on mobile terminal Port reservation APP carries out Ji Gang reservation；When truck driver needs to collect port, log in truck client truck Ji Gang and preengage APP, inquiry Goods relevant information, boats and ships flight number relevant information and collection port relevant information, according to the reservation rule choice set ETA estimated time of arrival at harbour, in advance Meeting by appointment port, after completing pre-port of meeting by appointment, truck driver carries out Ji Gang according to subscription time section；

Described terminal information is collected and is installed truck collection port subscription information collection processing system implementing on processing platform, truck arrives Time prediction system and stockyard secondary mould turnover optimize system；

Described truck collection port subscription information collection processing system implementing obtains truck driver to be intended collecting ETA estimated time of arrival, utilizes truck GPS Real-time positioning is to preengaging the position of truck, and notifies that harbour sluice gate advanced processing is met by appointment the relevant letter of truck at port in advance Breath；

The truck that the described truck prognoses system time of advent provides according to truck collection port subscription information collection processing system implementing Collection port subscription information, builds the truck dynamic prediction model time of advent based on support vector machine and Kalman filtering；

Described stockyard secondary mould turnover optimizes the truck time of advent that system provides according to the truck prognoses system time of advent, With the minimum target of suitcase totle drilling cost, it is considered to the longest constraint at ETA estimated time of arrival of mould turnover time, straddle truck traveling time and truck, structure Build stockyard mould turnover and truck suitcase sequence synchronization Optimized model, optimize simultaneously the suitcase order of truck, overturning case fall case position with And the task distribution of straddle truck.

A kind of truck information service and the method for work of dispatching patcher, comprise the following steps:

A, reservation truck collection ETA estimated time of arrival

When truck driver needs to collect port, log in client truck Ji Gang and preengage APP, inquire about goods relevant information, boats and ships Flight number relevant information and collection port relevant information, according to the reservation rule choice set ETA estimated time of arrival at harbour, port of meeting by appointment in advance, complete pre-meeting by appointment Behind port, truck driver carries out Ji Gang according to subscription time section；

B, collection process truck collection port subscription information

Truck collection port subscription information collection processing system implementing obtains truck driver to be intended collecting ETA estimated time of arrival, utilizes truck GPS fixed in real time Position is to preengaging the position of truck, in the middle of the input variable input model as the truck forecast model time of advent, and general Truck subscription information Real-time Feedback is updated to truck client；Meanwhile, harbour sluice gate advanced processing is met by appointment the phase of truck at port in advance Pass information；

C, forecast set card time of advent

For accurate forecast set card time of advent, it is thus achieved that the sequencing that truck arrives, truck prognoses system time of advent structure Build the truck dynamic prediction model time of advent based on support vector machine and Kalman filtering；Obtain utilizing supporting vector machine model On the basis of initial predicted result, update link travel time predictive value by Kalman filtering dynamic realtime, finally by Comprehensive analysis obtains the predictive value of the truck time of advent；The parameter and the variable-definition that relate to during prediction are as follows:

T: truck overall travel time, the truck time of advent of relative homeposition；

The running section set of U: truck

V: still need to running section set

a_i: section i, i=1,2,3 ..., I

a_ij: section a_iSub-section j, j=1,2,3 ..., J；

a_sm: truck current location, s, m are respectively section, current location numbering and sub-section numbering；

α_ij: 0-1 variable, if truck has run over section i section j, it is 1, is otherwise 0；

β_ij: 0-1 variable, if section i section j is contained in truck driving path, it is 1, is otherwise 0；

x_ij: section i section j running time；

The predictive value of the supporting vector machine model of section i section j running time；

z_ij: the measured value of section i section j running time；

t_ij: truck arrives the gps time during endpoint node of section i section j；

E_t: the weather condition when collection is positioned in section i section j, according to severe degree divided rank, 1 represents good Weather, 0 represents the most impassable weather, 0≤E_t≤1；

W_t: the number of weeks when collection is positioned in section i section j, W_t∈ [1,2,3,4,5,6,7], represents that Monday is to star Phase sky；

M_t: the day issue when collection is positioned in section i section j, M_t∈ [1,2,3 ..., 31], represent a middle of the month certain My god；

Φ: container type, 0 represents empty van, and 1 represents 20 chi loaded vans, and 2 represent 40 chi loaded vans；

Optimal filter estimated value

A_ij: state transfer amount parameter

H_ij: measurement error parameter

w_ij: average is the white Gaussian noise of 0, and covariance is Q

r_ij: average is the white Gaussian noise of 0, and covariance is R

x_ijThe covariance of prior uncertainty

P_i,j:x_ijThe covariance of posteriori error

K_ij: kalman gain matrix

Concrete prediction steps is as follows:

C1, based on historical data Training Support Vector Machines model

First set training set, re-use grid search and cross validation method determines that the supporting vector machine model of optimum is joined (C, v), selects number the most again to kernel function, then by input parameter and training set Training Support Vector Machines model, obtains After supporting vector machine model, then input value is inputted this supporting vector machine model trained it is predicted；Set and support vector The input variable of machine model is { U, t_ij,E_t,W_t,M_t, Φ }, the output variable of supporting vector machine model is

C2, dynamically update the predictive value time of advent based on Kalman Algorithm

Run real time information combination supporting vector machine model prediction result according to vehicle, moved by Kalman filtering iteration State updates link travel time, improves forecasting accuracy.

The truck dynamic prediction model time of advent based on Kalman filtering:

$T=\underset{i=1}{\overset{I}{\Σ}}\underset{j=1}{\overset{N_i}{\Σ}}{z}_{ij}{\mathrm{\α}}_{ij}{\mathrm{\β}}_{ij}+\underset{i=1}{\overset{I}{\Σ}}\underset{j=1}{\overset{N_i}{\Σ}}{\hat{x}}_{ij}(1-{\mathrm{\α}}_{ij}){\mathrm{\β}}_{ij}---\left(1\right)$

${\hat{x}}_{ij}=A_{ij}{\hat{x}}_{i,j-1}+K_{ij}(z_{ij}-H_{ij}{A}_{ij}{\hat{x}}_{i,j-1})---\left(2\right)$

$K_{ij}=P_{ij}^{-}{H}_{ij}^T{(H_{ij}{P}_{ij}^{-}{H}_{ij}^T+R)}^{-1}---\left(3\right)$

$P_{ij}^{-}=A_{ij}{P}_{i,j-1}{A}_{ij}^T+Q---\left(4\right)$

$P_{i,j+1}=(I-K_{ij}{H}_{ij})P_{ij}^{-}---\left(5\right)$

$P_{i0},{\hat{x}}_{i0}\∈R,{\hat{x}}_{i0}\≠0---\left(6\right)$

Formula (1) represents that truck completes to travel required All Time；Formula (2)～(5) are the recursion side of Kalman filtering Journey；

The process that the truck dynamic prediction model time of advent based on supporting vector machine model and Kalman filtering calculates is such as Under:

Step 1: initialize truck dynamic prediction model time of advent system modeAnd P_i0, make all α_i=0, β_i= 0, U={s}, truck dynamic prediction model time of advent control variable k=1；

Step 2: in supporting vector machine model, with U, t, moon number, number of weeks, time, weather condition, cargo type be Input variable, with V andOutput variable for supporting vector machine model.Update β_iIf section i is contained in U or V, then make β_i =1, it is otherwise 0；

Step 3: calculate prior uncertainty covariance matrix according to formula (4), calculate kalman gain matrix according to formula (3), System mode is updated according to formula (2)Truck overall travel time is calculated according to formula (1)；

Step 4: if meet s=U (end) simultaneously, m=J, then algorithm iteration terminates.Otherwise, systematic parameter is updated: if m= J, makes k=k+1, s=U (k), m=1；Otherwise, m=m+1 is made.Update posteriori error covariance matrix P_i,j+1, update α_i, repeat Step 3.

D, optimization stockyard secondary mould turnover order

Predicted by the truck forecast model time of advent, it is thus achieved that the truck time of advent and sequencing, set up stockyard mould turnover With truck suitcase sequence synchronization Optimized model, with the minimum target of suitcase totle drilling cost, it is considered to when mould turnover time, straddle truck move Between, the longest constraint at ETA estimated time of arrival of truck, build stockyard mould turnover and truck suitcase sequence synchronization Optimized model, optimize truck simultaneously Suitcase order, overturning case fall case position and straddle truck task distribution.For solving this model, use based on dynamic programming Heuritic approach, its basic ideas are the scheduling problems that the scheduling problem of multiple stage straddle truck resolves into multiple separate unit straddle truck, The job area assuming every straddle truck is several continuous print shellfish positions, and each shellfish position can only be serviced by a straddle truck.Use greedy Greedy Algorithm for Solving considers the separate unit straddle truck scheduling problem of mould turnover.Obtained in every straddle truck job area by greedy algorithm Suitcase totle drilling cost, returns it in dynamic programming, to optimize the job area of every straddle truck, job area internal object case The sequence of operation and overturning case dropping place, reach optimum when total suitcase cost minimization.Specifically comprise the following steps that

D1, straddle truck job area distribute: with b_kMinimum shellfish item in expression straddle truck k job area, k=1, 2 ..., m, g (i, j) represent shellfish position (i, j) in the range of all goal boxes suitcase totle drilling cost minimum and；(k j) represents gantry to f Hang the suitcase totle drilling cost of all goal boxes in k, k+1 ..., m job area minimum and, wherein r=b_k, f (1,1) i.e. represents institute Have goal box suitcase totle drilling cost minimum and.

Straddle truck k, k+1 ..., the job area of m be respectively r, r+1 ..., b_k+1-1}, { b_k+1,b_k+1+1,...,b_k+2- 1} ..., { b_m,b_m+1,...,α}；b₂,b₃,...,b_m, operating area is divided into m part, is respectively allocated to m straddle truck. Obtained by formula (7), (8) and make the b that the suitcase totle drilling cost sum of all goal boxes is minimum in the range of shellfish position (1, α)₂,b₃,..., b_m, f (1,1) i.e. determines；

F (m, r)=g (r, α), r=m, m+1 ..., α (7)

$f(k,r)=\underset{j=r+1,r+2,...,\mathrm{\&alpha;}-m+k+1}{min}\{f(k+1,j)+g(r,j-1)\},k\&le;r\&le;\mathrm{\&alpha;}-m+k,1\&le;k<m---\left(8\right)$

D2, separate unit straddle truck optimizing scheduling: use greedy algorithm solve g (i, j).It is { i, i+ for job area 1 ..., the scheduling problem of the separate unit straddle truck considering mould turnover of j}, its greedy algorithm flow process is as follows:

Step1: make t=0；Make l=i, i be job area be i, i+1 ..., the initial position of the straddle truck of j}.Make Ψ =1,2 ..., and n}, the set of the container not withdrawn in representing this straddle truck job area, n₁Represent this straddle truck operation In the range of the quantity of container.

Step2: if Ψ=φ, stop；Otherwise, for container q each in Ψ, the number of its obstruction case of t is calculated Amount n_q。

Step3: for container q each in Ψ, calculates its mould turnover cost C₁, straddle truck moved to l by l_qMobile cost C₂。

Step4: for container q each in Ψ, calculate the deadline t of its pick-up operation_q。

Step5: for container q each in Ψ, calculates t_qIn the moment, arrive port but the delay of not serviced all trucks Cost C₃。

Step6: for container q each in Ψ, calculates C₁、C₂、C₃Sum C, the container minimum by CAs the next one The goal box extracted.

Step7: for container q each in Ψ, if its anticipated time departure the latest is more thanThen makeBy packaging CaseThe goal box extracted as the next one.

Step8: ifPerform Step13；Otherwise, make n=1, perform Step9.

Step9: ifPerform Step13.Otherwise, Step10 is performed.

Step10: forN-th obstruction case, check whether it exists first kind candidate's stack, if it does not, perform Step11；If there is first kind candidate's stack, then as the n-th case position that falls hindering case.Otherwise, each is selected Suitcase order container the earliest in one class candidate's stack, is ranked up the suitcase order of these containers, selects the most wherein Select the stack at suitcase order container place the earliest as the n-th case position that falls hindering case.N=n+1, performs Step9.

Step11: forN-th obstruction case, check whether it exists empty stack, if it does not, perform Step12；As There is an empty stack in fruit, then as the n-th case position that falls hindering case.Otherwise, select the empty stack of the leftmost side as the n-th resistance Hinder the case position that falls of case.N=n+1, performs Step9.

Step12: select suitcase order container the earliest in each Equations of The Second Kind candidate's stack, by the suitcase of these containers Order is ranked up, and selects the stack at suitcase order container place the latest as the n-th case that falls hindering case the most wherein Position.N=n+1, performs Step9.

Step13: willReject from Φ, orderPerform Step2.

Wherein, Step10, Step11, Step12 are the heuristic rules of mould turnover.Candidate's stack is divided into three classes, and a class is to work as The suitcase order of front stored up container is all later than the stack waiting the suitcase of falling case order, referred to as first kind candidate stack.One class is to work as Before there is no the stack that container stores up, referred to as empty stack.Another kind of is an at least container in the container currently stored up Suitcase order is early than waiting the suitcase of falling case stack sequentially, referred to as Equations of The Second Kind candidate stack.Obstruction case is moved on to first kind candidate's stack with empty Stack all will not cause secondary mould turnover, and obstruction case moves on to Equations of The Second Kind candidate's stack and can cause secondary mould turnover.Therefore, fall the selection of case position Priority be: first kind candidate stack → empty stack → Equations of The Second Kind candidate's stack.

Be obtained by greedy algorithm shellfish position (i, j) in the range of the suitcase totle drilling cost sum of all goal boxes,It isMinimum and.WithAlternate form (7) and (8) g () in and f (), obtains formula (9) and (10).

$\hat{f}(m,r)=\hat{g}(r,\mathrm{\&alpha;}),r=m,m+1,...,\mathrm{\&alpha;}---\left(9\right)$

$\hat{f}(k,r)=\underset{j=r+1,r+2,...,\mathrm{\&alpha;}-m+k+1}{min}\{\hat{f}(k+1,j)+\hat{g}(r,j-1)\},k\&le;r\&le;\mathrm{\&alpha;}-m+k,1\&le;k<m---\left(10\right)$

Recursively use formula (17) and (18) obtain all goal box suitcase totle drilling costs minimum andDetermine every simultaneously The job area of platform straddle truck, the suitcase order of goal box and the case position that falls of overturning case.

According to above step, by Computer Simulation, obtain state during totle drilling cost minimum, i.e. every straddle truck Job area, the i.e. truck suitcase of the sequence of operation in job area order and overturning case dropping place position.Meanwhile, dock operation is adjusted Degree arranges straddle truck operation based on optimum results, determines that pick-up operation order, overturning case fall case position.

Further, described suitcase totle drilling cost includes that mould turnover cost, straddle truck move the delay of cost and truck and become This.

Compared with prior art, the method have the advantages that

1, owing to present invention employs truck client and terminal information collection processing platform, it is achieved that truck information service With the function of scheduling, improve dock operation efficiency, reduce harbour operation cost.

2, owing to present invention employs truck client (truck Ji Gang preengages APP), truck driver use on mobile terminal Truck Ji Gang reservation APP carry out Ji Gang reservation, it is to avoid the problems such as the harbour caused blocks up of entering a port are concentrated in truck peak period；With Time, the handling capacity of sluice gate quickly through sluice gate, can be improve when the truck preengage enters a port.On the other hand, harbour can obtain Obtain truck to preengage ETA estimated time of arrival, preengage the real time position of truck.

3, owing to present invention employs truck collection port subscription information collection processing system implementing, truck driver can be obtained and intend Ji Gang Time, utilize truck GPS real-time positioning to preengaging the position of truck, and notify that harbour sluice gate advanced processing is met by appointment port in advance The relevant information of truck；

4, owing to present invention employs the truck prognoses system time of advent, build based on support vector machine and Kalman filtering The truck dynamic prediction model time of advent, can according to truck collection port subscription information collection processing system implementing provide truck Ji Gang Subscription information, it was predicted that go out more accurate collection and snap into ETA estimated time of arrival and collection snaps into port sequencing.

5, owing to present invention employs by Kalman filtering, it is easy by surrounding environment influence to process in truck running Cause reflection and the decay of gps signal, bring random noise problem, dynamically adjust support vector machine according to truck real time information The prediction output of model, thus improve precision and the robustness arriving time prediction.

6, owing to present invention employs stockyard secondary mould turnover optimization system, can carry according to the truck prognoses system time of advent The truck time of advent of confession, optimizing the task distribution of fall case position and the straddle truck of the suitcase order of truck, overturning case, decision-making goes out Job scheduling scheme during the pick-up operation of inlet box stockyard.Improve storage yard operation efficiency, reduce dock operation cost.

Accompanying drawing explanation

Fig. 1 is the flow chart of the present invention.

Fig. 2 is the reservation query system schematic of client truck Ji Gang of the present invention reservation APP.

Fig. 3 is that the truck of client truck Ji Gang of the present invention reservation APP enters a port one of reservation interface schematic diagram.

Fig. 4 is that the truck of client truck Ji Gang of the present invention reservation APP enters a port the two of reservation interface schematic diagram.

Fig. 5 is the flow chart of the container terminal of the present invention truck Forecasting Methodology time of advent.

Fig. 6 is that in truck of the present invention arrival forecast model, truck based on support vector machine predicts flow chart the time of advent.

Fig. 7 is dynamically to update the predictive value method time of advent in truck of the present invention arrival forecast model based on Kalman Algorithm Schematic diagram.

Fig. 8 is the flow chart of stockyard of the present invention secondary mould turnover optimization method.

Detailed description of the invention

Below in conjunction with the accompanying drawings the present invention is described further.

As it is shown in figure 1, the present invention designs a kind of truck letter that can improve dock operation efficiency, reduction harbour operation cost Breath service and dispatching patcher, collect processing platform, described truck client and harbour including truck client and terminal information Information management platform is connected by wireless network.

As it is shown in figure 1, a kind of truck information service and the method for work of dispatching patcher, comprise the following steps:

A, reservation truck collection ETA estimated time of arrival, be three kinds of application interfaces of truck Ji Gang reservation APP shown in Fig. 2,3,4.

B, collection process truck collection port subscription information

C, forecast set card time of advent, be the flow chart of the truck Forecasting Methodology time of advent shown in Fig. 5.It it is collection shown in Fig. 6 Card arrives truck based on supporting vector machine model in forecast model and predicts flow chart the time of advent.Shown in Fig. 7 based on karr Graceful algorithm dynamically updates the predictive value method schematic diagram time of advent.The flow chart of the stockyard secondary mould turnover optimization method shown in Fig. 8.

The above, the only present invention preferably detailed description of the invention, but protection scope of the present invention is not limited thereto, Any those familiar with the art in the technical scope that the invention discloses, according to technical scheme and Inventive concept equivalent or change in addition, all should contain within protection scope of the present invention.

Claims (3)

1. a truck information service and dispatching patcher, it is characterised in that: include that truck client and terminal information collection process Platform, described truck client and terminal information are collected processing platform and are connected by wireless network；

Installing truck Ji Gang in described truck client and preengage APP, truck driver uses the truck Ji Gang on mobile terminal pre- About APP carries out Ji Gang reservation；When truck driver needs to collect port, log in truck client truck Ji Gang and preengage APP, inquire about goods Relevant information, boats and ships flight number relevant information and collection port relevant information, according to the reservation rule choice set ETA estimated time of arrival at harbour, meet by appointment in advance Port, after completing pre-port of meeting by appointment, truck driver carries out Ji Gang according to subscription time section；

Described terminal information is collected and is installed truck collection port subscription information collection processing system implementing, truck time of advent on processing platform Prognoses system and stockyard secondary mould turnover optimize system；

Described truck collection port subscription information collection processing system implementing obtains truck driver to be intended collecting ETA estimated time of arrival, utilizes truck GPS real-time Navigate to preengage the position of truck, and notify that harbour sluice gate advanced processing is met by appointment the relevant information of truck at port in advance；

The truck Ji Gang that the described truck prognoses system time of advent provides according to truck collection port subscription information collection processing system implementing Subscription information, builds the truck dynamic prediction model time of advent based on support vector machine and Kalman filtering；

Described stockyard secondary mould turnover optimizes the truck time of advent that system provides according to the truck prognoses system time of advent, to carry This minimum target of box assembly, it is considered to the longest constraint at ETA estimated time of arrival of mould turnover time, straddle truck traveling time and truck, builds heap Field mould turnover and truck suitcase sequence synchronization Optimized model, optimize the suitcase order of truck, fall case position and the dragon of overturning case simultaneously The task distribution of gantry crane.

2. a truck information service and the method for work of dispatching patcher, it is characterised in that: comprise the following steps:

A, reservation truck collection ETA estimated time of arrival

When truck driver needs to collect port, log in client truck Ji Gang and preengage APP, inquire about goods relevant information, boats and ships flight number Relevant information and collection port relevant information, according to the reservation rule choice set ETA estimated time of arrival at harbour, port of meeting by appointment in advance, after completing pre-port of meeting by appointment Truck driver carries out Ji Gang according to subscription time section；

B, collection process truck collection port subscription information

Truck collection port subscription information collection processing system implementing obtains truck driver to be intended collecting ETA estimated time of arrival, utilizes truck GPS real-time positioning to arrive Preengage the position of truck, in the middle of the input variable input model as the truck forecast model time of advent, and by truck Subscription information Real-time Feedback is updated to truck client；Meanwhile, harbour sluice gate advanced processing is met by appointment the relevant letter of truck at port in advance Breath；

C, forecast set card time of advent

For accurate forecast set card time of advent, it is thus achieved that the sequencing that truck arrives, the truck prognoses system time of advent builds base The truck dynamic prediction model time of advent in support vector machine and Kalman filtering；At the beginning of utilizing supporting vector machine model to obtain On the basis of beginning to predict the outcome, update link travel time predictive value by Kalman filtering dynamic realtime, finally by comprehensively Analyze the predictive value obtaining the truck time of advent；The parameter and the variable-definition that relate to during prediction are as follows:

T: truck overall travel time, the truck time of advent of relative homeposition；

The running section set of U: truck

V: still need to running section set

a_i: section i, i=1,2,3 ..., I

a_ij: section a_iSub-section j, j=1,2,3 ..., J；

a_sm: truck current location, s, m are respectively section, current location numbering and sub-section numbering；

α_ij: 0-1 variable, if truck has run over section i section j, it is 1, is otherwise 0；

β_ij: 0-1 variable, if section i section j is contained in truck driving path, it is 1, is otherwise 0；

x_ij: section i section j running time；

The predictive value of the supporting vector machine model of section i section j running time；

z_ij: the measured value of section i section j running time；

t_ij: truck arrives the gps time during endpoint node of section i section j；

E_t: the weather condition when collection is positioned in section i section j, according to severe degree divided rank, 1 represents good weather, 0 Represent the most impassable weather, 0≤E_t≤1；

W_t: the number of weeks when collection is positioned in section i section j, W_t∈ [1,2,3,4,5,6,7], represents that Monday is to week My god；

M_t: the day issue when collection is positioned in section i section j, M_t∈ [1,2,3 ..., 31], represent certain sky in a middle of the month；

Φ: container type, 0 represents empty van, and 1 represents 20 chi loaded vans, and 2 represent 40 chi loaded vans；

Optimal filter estimated value

A_ij: state transfer amount parameter

H_ij: measurement error parameter

w_ij: average is the white Gaussian noise of 0, and covariance is Q

r_ij: average is the white Gaussian noise of 0, and covariance is R

x_ijThe covariance of prior uncertainty

P_i,j:x_ijThe covariance of posteriori error

K_ij: kalman gain matrix

Concrete prediction steps is as follows:

C1, based on historical data Training Support Vector Machines model

First set training set, re-use grid search and cross validation method determines optimum supporting vector machine model parameter (C, v), selects kernel function the most again, then by input parameter and training set Training Support Vector Machines model, is propped up After holding vector machine model, then input value is inputted this supporting vector machine model trained it is predicted；Set support vector machine The input variable of model is { U, t_ij,E_t,W_t,M_t, Φ }, the output variable of supporting vector machine model is

C2, dynamically update the predictive value time of advent based on Kalman Algorithm

Real time information combination supporting vector machine model prediction result is run, by Kalman filtering ofaiterative, dynamic more according to vehicle New link travel time, improves forecasting accuracy；

The truck dynamic prediction model time of advent based on Kalman filtering:

$T=\underset{i=1}{\overset{I}{\&Sigma;}}\underset{j=1}{\overset{N_i}{\&Sigma;}}{z}_{ij}{\mathrm{\&alpha;}}_{ij}{\mathrm{\&beta;}}_{ij}+\underset{i=1}{\overset{I}{\&Sigma;}}\underset{j=1}{\overset{N_i}{\&Sigma;}}{\hat{x}}_{ij}(1-{\mathrm{\&alpha;}}_{ij}){\mathrm{\&beta;}}_{ij}---\left(1\right)$

${\hat{x}}_{ij}=A_{ij}{\hat{x}}_{i,j-1}+K_{ij}(z_{ij}-H_{ij}{A}_{ij}{\hat{x}}_{i,j-1})---\left(2\right)$

$K_{ij}=P_{ij}^{-}{H}_{ij}^T{(H_{ij}{P}_{ij}^{-}{H}_{ij}^T+R)}^{-1}---\left(3\right)$

$P_{ij}^{-}=A_{ij}{P}_{i,j-1}{A}_{ij}^T+Q---\left(4\right)$

$P_{i,j+1}=(I-K_{ij}{H}_{ij})P_{ij}^{-}---\left(5\right)$

$P_{i0},{\hat{x}}_{i0}\&Element;R,{\hat{x}}_{i0}\&NotEqual;0---\left(6\right)$

Formula (1) represents that truck completes to travel required All Time；Formula (2)～(5) are the recurrence equation of Kalman filtering；

The process that the truck dynamic prediction model time of advent based on supporting vector machine model and Kalman filtering calculates is as follows:

Step 1: initialize truck dynamic prediction model time of advent system modeAnd P_i0, make all α_i=0, β_i=0, U= { s}, truck dynamic prediction model time of advent control variable k=1；

Step 2: in supporting vector machine model, with U, t, moon number, number of weeks, time, weather condition, cargo type for input Variable, with V andOutput variable for supporting vector machine model；Update β_iIf section i is contained in U or V, then make β_i=1, It is otherwise 0；

Step 3: calculate prior uncertainty covariance matrix according to formula (4)Kalman gain matrix is calculated according to formula (3), according to Formula (2) updates system modeTruck overall travel time is calculated according to formula (1)；

Step 4: if meet s=U (end) simultaneously, m=J, then algorithm iteration terminates；Otherwise, systematic parameter is updated: if m=J, make K=k+1, s=U (k), m=1；Otherwise, m=m+1 is made；Update posteriori error covariance matrix P_i,j+1, update α_i, repeat step 3；

D, optimization stockyard secondary mould turnover order

Predicted by the truck forecast model time of advent, it is thus achieved that the truck time of advent and sequencing, set up stockyard mould turnover and collection Card suitcase sequence synchronization Optimized model, with the minimum target of suitcase totle drilling cost, it is considered to mould turnover time, straddle truck traveling time, collection Block the longest constraint at ETA estimated time of arrival, build stockyard mould turnover and truck suitcase sequence synchronization Optimized model, optimize carrying of truck simultaneously Case order, overturning case fall case position and straddle truck task distribution；For solving this model, use inspiration based on dynamic programming Formula algorithm, its basic ideas are the scheduling problems that the scheduling problem of multiple stage straddle truck resolves into multiple separate unit straddle truck, it is assumed that The job area of every straddle truck is several continuous print shellfish positions, and each shellfish position can only be serviced by a straddle truck；Greediness is used to calculate Method solves the separate unit straddle truck scheduling problem considering mould turnover；The suitcase in every straddle truck job area is obtained by greedy algorithm Totle drilling cost, returns it in dynamic programming, to optimize the job area of every straddle truck, the operation of job area internal object case Order and overturning case dropping place, reach optimum when total suitcase cost minimization；Specifically comprise the following steps that

D1, straddle truck job area distribute: with b_kRepresent the minimum shellfish item in straddle truck k job area, k=1,2 ..., m, g (i, j) represent shellfish position (i, j) in the range of all goal boxes suitcase totle drilling cost minimum and；(k j) represents straddle truck k, k+ to f 1 ..., in m job area the suitcase totle drilling cost of all goal boxes minimum and, wherein r=b_k, f (1,1) i.e. represents all targets The minimum of case suitcase totle drilling cost and；

Straddle truck k, k+1 ..., the job area of m be respectively r, r+1 ..., b_k+1-1}, { b_k+1,b_k+1+1,...,b_k+2- 1} ..., { b_m,b_m+1,...,α}；b₂,b₃,...,b_m, operating area is divided into m part, is respectively allocated to m straddle truck； Obtained by formula (7), (8) and make the b that the suitcase totle drilling cost sum of all goal boxes is minimum in the range of shellfish position (1, α)₂,b₃,..., b_m, f (1,1) i.e. determines；

F (m, r)=g (r, α), r=m, m+1 ..., α (7)

$f(k,r)=\underset{j=r+1,r+2,...,\mathrm{\&alpha;}-m+k+1}{min}\{f(k+1,j)+g(r,j-1)\},k\&le;r\&le;\mathrm{\&alpha;}-m+k,1\&le;k<m---\left(8\right)$

D2, separate unit straddle truck optimizing scheduling: use greedy algorithm solve g (i, j)；For job area be i, i+1 ..., j} The scheduling problem of separate unit straddle truck considering mould turnover, its greedy algorithm flow process is as follows:

Step1: make t=0；Make l=i, i be job area be i, i+1 ..., the initial position of the straddle truck of j}；Make Ψ= 1,2 ..., and n}, the set of the container not withdrawn in representing this straddle truck job area, n₁Represent this straddle truck operation model Enclose the quantity of interior container；

Step2: if Ψ=φ, stop；Otherwise, for container q each in Ψ, quantity n of its obstruction case of t is calculated_q；

Step3: for container q each in Ψ, calculates its mould turnover cost C₁, straddle truck moved to l by l_qMobile cost C₂；

Step4: for container q each in Ψ, calculate the deadline t of its pick-up operation_q；

Step5: for container q each in Ψ, calculates t_qIn the moment, arrive port but the tardiness cost of not serviced all trucks C₃；

Step6: for container q each in Ψ, calculates C₁、C₂、C₃Sum C, the container minimum by CExtract as the next one Goal box；

Step7: for container q each in Ψ, if its anticipated time departure the latest is more thanThen makeBy containerMake The goal box extracted for the next one；

Step8: ifPerform Step13；Otherwise, make n=1, perform Step9；

Step9: ifPerform Step13；Otherwise, Step10 is performed；

Step10: forN-th obstruction case, check whether it exists first kind candidate's stack, if it does not, perform Step11；If there is first kind candidate's stack, then as the n-th case position that falls hindering case；Otherwise, each is selected Suitcase order container the earliest in one class candidate's stack, is ranked up the suitcase order of these containers, selects the most wherein Select the stack at suitcase order container place the earliest as the n-th case position that falls hindering case；N=n+1, performs Step9；

Step11: forN-th obstruction case, check whether it exists empty stack, if it does not, perform Step12；If deposited An empty stack, then as the n-th case position that falls hindering case；Otherwise, the empty stack selecting the leftmost side hinders case as n-th The case position that falls；N=n+1, performs Step9；

Step12: select suitcase order container the earliest in each Equations of The Second Kind candidate's stack, by the suitcase order of these containers It is ranked up, selects the stack at suitcase order container place the latest as the n-th case position that falls hindering case the most wherein；n =n+1, performs Step9；

Step13: willReject from Φ, orderPerform Step2；

Wherein, Step10, Step11, Step12 are the heuristic rules of mould turnover；Candidate's stack is divided into three classes, and a class is current institute The suitcase order of the container stored up all is later than the stack waiting the suitcase of falling case order, referred to as first kind candidate stack；One class is currently not have There are the stack that container is stored up, referred to as empty stack；Another kind of is the suitcase of an at least container in the container currently stored up Order is early than waiting the suitcase of falling case stack sequentially, referred to as Equations of The Second Kind candidate stack；Obstruction case is moved on to first kind candidate's stack equal with empty stack Secondary mould turnover will not be caused, and obstruction case is moved on to Equations of The Second Kind candidate's stack and can cause secondary mould turnover；Therefore, fall the excellent of case position selection First order is: first kind candidate stack → empty stack → Equations of The Second Kind candidate's stack；

Be obtained by greedy algorithm shellfish position (i, j) in the range of the suitcase totle drilling cost sum of all goal boxes,It isMinimum and；WithWithG in alternate form (7) and (8) () and f (), obtain formula (9) and (10)；

$\hat{f}(m,r)=\hat{g}(r,\mathrm{\&alpha;}),r=m,m+1,...,\mathrm{\&alpha;}---\left(9\right)$

$\hat{f}(k,r)=\underset{j=r+1,r+2,...,\mathrm{\&alpha;}-m+k+1}{min}\{\hat{f}(k+1,j)+\hat{g}(r,j-1)\},k\&le;r\&le;\mathrm{\&alpha;}-m+k,1\&le;k<m---\left(10\right)$

Recursively use formula (17) and (18) obtain all goal box suitcase totle drilling costs minimum andDetermine every dragon simultaneously The job area of gantry crane, the suitcase order of goal box and the case position that falls of overturning case；

According to above step, by Computer Simulation, obtain the operation of state during totle drilling cost minimum, i.e. every straddle truck Scope, the i.e. truck suitcase of the sequence of operation in job area order and overturning case dropping place position；Meanwhile, dock operation scheduling with Arrange straddle truck operation based on optimum results, determine that pick-up operation order, overturning case fall case position.

A kind of truck information service the most according to claim 2 and the method for work of dispatching patcher, it is characterised in that: described Suitcase totle drilling cost include that mould turnover cost, straddle truck move the tardiness cost of cost and truck.