CN103246969A - Method and device for realization of logistics and allocation - Google Patents

Abstract

The invention discloses a method and a device for realization of logistics allocation. The method comprises the following steps: generating a math model used for realizing logistics allocation through logistics allocation constraint condition modeling, allocation station distance modeling, logistics allocation algorithm modeling; and applying ant colony algorithm on the basis of the math models to allocate logistics. The technology for realization of logistics allocation, disclosed by the invention, can generate a math model used for realizing logistics allocation through logistics allocation constraint condition modeling, allocation station distance modeling, logistics allocation algorithm modeling, and apply ant colony algorithm on the basis of the math models to allocate logistics, so as to generate near-optimal logistics allocation scheme at minimal amount and range, which increases the logistics allocation efficiency and saves logistics allocation cost to a large extent.

Description

A kind of implementation method of logistics deployment and device

Technical field

The present invention relates to technology of Internet of things, be specifically related to a kind of implementation method and device of logistics deployment.

Background technology

The optimizing of logistics deployment scheme belongs to the Combinatorial Optimization category, is that (Non-deterministic Polynomial, NP) problem, the solution of this problem are big research focuses to the uncertainty polynomial expression.In addition, owing to actual logistics deployment state is in the dynamic change, thereby require the Combinatorial Optimization algorithm should have good performance, can adapt to the variation characteristic of dynamic allotment simultaneously again.Some present logistics deployment algorithms adopt non-didactic traditional algorithm, there is the not high problem of efficient in performance, and the heuritic approach of other better performances reckons without the variation characteristic of allotment state, therefore can not be in time, dynamically scheme is adjusted, caused the inefficiencies of logistics deployment scheme.

Summary of the invention

In view of this, fundamental purpose of the present invention is to provide a kind of implementation method and device of logistics deployment, is used for realizing the high-level efficiency logistics deployment, and adapts to the dynamic variation characteristic in the logistics deployment process.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of implementation method of logistics deployment, this method comprises:

Generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station; On described mathematical model basis, use ant group algorithm allotment logistics.

The method of described logistics deployment constraint condition modeling is:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at moment T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)) satisfies state constraint condition S _i(T (t)) 〉=y _i(T (t)); (i=1 ... n).

This method also comprises:

By each logistics deployment station being set up and being safeguarded a finite state machine, carry out the management of logistics deployment; Logistics deployment station a _iThe state machine function definition be: St _i(T (t)); (i=1 wherein ... n);

When state machine can not satisfy the described state constraint condition of following corresponding time point, by starting the logistics deployment scheme state machine is carried out the conversion of state, make it satisfy described state constraint condition.

Described dispensing station comprises absolute distance modeling and relative distance modeling apart from modeling; Wherein,

During the absolute distance modeling, establish website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites:

During the relative distance modeling, if website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _IjIf website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

The modeling of described logistics deployment algorithm is used for to-be is predicted, comprises to-be function definition, logistics deployment defined matrix; Wherein,

When carrying out the to-be function definition, the to-be anticipation function is defined as: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods; Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value;

When carrying out the logistics deployment defined matrix, set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

This method also comprises:

With

Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine.

With

Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

The process of using ant group algorithm allotment logistics on described mathematical model basis may further comprise the steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant; Node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in; Described ant is the variable that is used for optimizing in the ant group algorithm;

Step 2: cycle index N _c=N _C+1If, N _c＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, according to state transition probability formula calculating transition probability, the next node of selecting node from then on to set out;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, finishes if this takes turns all ant routings, then jumps to step 9, otherwise jumps to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is as the optimum solution of allotment scheme.

This method also comprises:

Corresponding this suboptimum logistics deployment scheme that has generated, use real-time perception information, determine whether this scheme needs dynamically to adjust, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades; All k ant is placed on car hauler current location node, carries out optimization again and find the solution.

A kind of implement device of logistics deployment, this device comprise mathematical model generation module, logistics deployment module; Wherein,

Described mathematical model generation module is used for generating for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station;

Described logistics deployment module is used for using ant group algorithm allotment logistics on described mathematical model basis.

Described mathematical model generation module comprises logistics deployment constraint condition modeling unit, is used for:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at the moment in future T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)) satisfies state constraint condition S _i(T (t)) 〉=y _i(T (t)); (i=1 ... n).

Described logistics deployment constraint condition modeling unit also is used for:

By each logistics deployment station being set up and being safeguarded a finite state machine, carry out the management of logistics deployment; Logistics deployment station a _iThe state machine function definition be: St _i(T (t)); (i=1 wherein ... n);

When state machine can not satisfy the described state constraint condition of following corresponding time point, state machine is carried out the conversion of state.

Described mathematical model generation module comprises dispensing station apart from modeling unit, is used for carrying out absolute distance modeling and relative distance modeling; Wherein,

During the absolute distance modeling, establish website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites:

During the relative distance modeling, if website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _IjIf website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

Described mathematical model generation module comprises logistics deployment algorithm modeling unit, is used for to-be is predicted, comprises to-be function definition, logistics deployment defined matrix; Wherein,

When carrying out the to-be function definition, the to-be anticipation function is defined as: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods; Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value;

When carrying out the logistics deployment defined matrix, set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

Described logistics deployment algorithm modeling unit also is used for:

With

Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine.

With

Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

When described logistics deployment module is used ant group algorithm allotment logistics on the mathematical model basis, be used for carrying out following steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant; Node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in; Described ant is the variable that is used for optimizing in the ant group algorithm;

Step 2: cycle index N _c=N _C+1If, N _c＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, according to state transition probability formula calculating transition probability, the next node of selecting node from then on to set out;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, finishes if this takes turns all ant routings, then jumps to step 9, otherwise jumps to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is as the optimum solution of allotment scheme.

Described logistics deployment module also is used for:

Corresponding this suboptimum logistics deployment scheme that has generated, use real-time perception information, determine the level of enforcement of this scheme, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades; All k ant is placed on car hauler current location node, carries out optimization again and find the solution.

The present invention realizes that the technology of logistics deployment can generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station; And on described mathematical model basis, use ant group algorithm and allocate logistics, therefore can adapt to the dynamic variation characteristic in the logistics deployment process, in minimum number, minimum zone, generate the best logistics deployment scheme that approaches, this can improve logistics deployment efficient to a great extent, realize the high-level efficiency logistics deployment, and save the logistics deployment cost.

Description of drawings

Fig. 1 realizes the general flow chart of logistics deployment for the embodiment of the invention;

Fig. 2 realizes the installation drawing of logistics deployment for the embodiment of the invention.

Embodiment

The present invention adopts a kind of dynamic, improved ant group algorithm to allocate the dynamic optimization of scheme, and this algorithm is a kind of heuritic approach.Heuritic approach claims intelligence computation again, be also referred to as " soft calculating ", it is the inspiration that people are subjected to the natural law, the algorithm that problem is found the solution in imitation according to the natural law, such algorithm has that the global optimization performance is good, strong robustness, highly versatile, be suitable for advantages such as parallel processing, and because of the efficient optimization performance, need not advantage such as problem specific information, be widely used in computer science, optimized fields such as scheduling, transportation problem, Combinatorial Optimization.Ant group algorithm (ant colony algorithm) is that people such as Maniezzo at first put forward in early 1990s by Italian scholar Dorigo.Ant group algorithm is that later another of first heuristic search algorithm such as simulated annealing, genetic algorithm, tabu search algorithm, artificial neural network algorithm that continue are applied to the heuristic search algorithm of combinatorial optimization problem.Characteristic because ant group algorithm itself possesses can adapt to the dynamic problem in the searching process preferably.

Foundation based on the intelligent management system of technology of Internet of things, make the real-time requirement amount of logistics deployment station article inventory amount, Item Title, type, article, the obtaining in real time of information such as particular location of logistics deployment car become possibility, analyze and predict by the perception data that these are got access in real time, can generate and approach best logistics deployment scheme, also can realize real-time update, dynamically adjustment to the logistics deployment scheme.

The present invention establishes mathematical model and improves based on ant group algorithm, and it is respond well to have provided optimizing, the logistics deployment scheme that efficient is higher.The present invention studies the logistics deployment problem based on technology of Internet of things.At first be designed for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station, on the mathematical model basis, the logistics deployment problem is studied then, comprised and use ant group algorithm to the research of static logistics deployment algorithm and the research of dynamic logistics deployment algorithm.

Particularly, ant group algorithm is that a kind of heuristic optimization algorithm (claims intelligence computation again, also the someone is referred to as soft calculating), have characteristics such as efficient optimization performance, global optimization performance, strong robustness, highly versatile, be widely used in computer science, optimized fields such as scheduling, transportation problem, Combinatorial Optimization.

Need to prove that logistics deployment is extremely important, because have only required goods in time to allocate, could guarantee normal operation, improve the quality of logistics deployment.The application of technology of Internet of things, make real-time perception information obtain and the prediction of information becomes possibility.Technology such as applied forcasting technology of the present invention, real-time perception generate in minimum number, minimum zone and approach best logistics deployment scheme, can save the logistics deployment cost to a great extent.

When carrying out mathematical modeling, can carry out following operation:

The modeling of logistics deployment constraint condition:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at following certain moment T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)), in order not influence carrying out smoothly of producing and conclude the business, system must satisfy following state constraint condition:

S _i(T(t))≥y _i(T(t))；(i＝1…n) (1)

For setting forth better and dealing with problems, introduce Finite State Machine.By each logistics deployment station being set up and being safeguarded a finite state machine, carry out the management of logistics deployment.Finite state machine comprises and the information safeguarded comprises: the logistics goods demand of corresponding logistics distribution station when the logistics goods tank farm stock of corresponding logistics distribution station, this time point when time, certain time point.When state machine can not satisfy the state constraint condition (1) of following corresponding time point, just need carry out the conversion of state to state machine.The conditional function that can impel logistics deployment station state machine state to change has f, g etc.; Wherein,

G: logistics deployment function;

F: other influence factors.

Logistics deployment station a _iThe state machine function definition be: St _i(T (t)); (i=1 wherein ... n).

Dispensing station is apart from modeling:

The logistics deployment site distance from modeling, can be divided into absolute distance modeling and relative distance modeling.

1) absolute distance modeling

If website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites.Since the singularity of road, website a _iTo website a _jDistance and website a _jTo website a _iBetween distance may be different, thereby the formed matrix of modeling is not symmetric matrix:

2) relative distance modeling

All logistics distribution station points are formed a big internet, may directly link to each other between this website that wherein has, and may also have other websites between the website that has.

In order to express the relative distance between the different websites more intuitively, set up the relative distance matrix in addition, set following rule:

If website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _Ij, d _IjCan be the absolute distance value, also can be the relative distance value.

If website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

The modeling of logistics deployment algorithm:

Determine the logistics deployment scheme, factors such as tank farm stock, demand, state variation need look to the future with holding water.Therefore need set up the to-be function:

1) to-be function definition

At following certain quantity of goods that need allocate constantly, can predict by features such as empirical value, historical data, following variations.The to-be function definition is: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods.Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value.

2) logistics deployment defined matrix

Set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

Can use Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine.

Can use Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

When carrying out concrete logistics deployment, can use static logistics deployment algorithm, can also use the dynamic logistics deployment algorithm.Particularly, when the target state machine at logistics deployment station can not satisfy constraint condition, just need start the allotment that logistics deployment system carries out goods at the appointed time.Static logistics deployment algorithm is the deployment algorithm that logistics deployment is formulated when not beginning as yet; The dynamic logistics deployment algorithm is when the allotment task is carried out, by real-time perception to information determine that situation changes, thereby former allotment scheme is carried out the adjustment algorithm of real time modifying.

Static logistics deployment algorithm totally comprises finding the solution and the scheme optimization of startup, allotment scheme:

1) startup of static logistics deployment algorithm

According to the information of real-time perception, detect distribution station and whether the situation that the corresponding state machine does not satisfy constraint condition will take place in the following time, determine whether to start static logistics deployment algorithm accordingly.

2) allotment scheme is found the solution

Static logistics deployment algorithm is found the solution the logistics deployment matrix under constraint condition, and allotment scheme coverage minimum, the trucking costs that need satisfy generation simultaneously are minimum, namely allocate the optimization problem of scheme.Concrete operation is: the solution that satisfies condition for each group of the logistics deployment matrix that satisfies target state machine constraint condition, this group solution is carried out logistics deployment scheme optimization computation, thereby draw the optimum allotment scheme that this group is taken off, optimum allotment scheme to all solutions compares at last, draws the allotment scheme of global optimum.

Particularly, allotment the finding the solution of scheme, be with the status predication function calculation go out for the value of negative makes it become 0 by additive operation, while record coversion process generates corresponding conversion scheme, this conversion scheme is exactly a logistics deployment scheme.The record of conversion process is to be that zero two-dimensional matrix is realized entirely by an init state, corresponding additive operation only occurs between the positive negative value, vector carries out an additive operation, then in the allotment matrix, also mark in corresponding position accordingly, finish also just having generated corresponding allotment matrix like this when linear transformation.Different conversion processes will produce different allotment matrixes, and different allotment schemes just are the optimal scheduling schemes as for which allotment scheme wherein, also need further to use the further optimizing of allotment scheme optimization algorithm, generate near the optimal scheduling scheme at last.

3) allotment scheme optimization

At first generate best traffic program for each allotment scheme.For simplicity, establish the logistics deployment expense only with the scheduling distance dependent, same same goods of time allotment of car hauler and allotment polylith goods have same-cost.The allotment scheme matrix to generating at first, whether the scheduling of goods exists overlapping route between the different websites with the relative distance matrix computations of applications distances matrix, goods on the overlapping route can be taken car by the way, travelling expenses have so just been saved, the overlapping route of car merges taking by the way, allotment scheme optimization problem repeat just to transform into after circuit merges a traveling salesman problem special, that condition is limited (Traveling Salesman Problem, TSP).

There is optimum solution in theory in TSP, but does not also find effective method to try to achieve this value in actual applications, can only approach to optimum solution gradually, namely can only obtain near the suboptimal solution of theoretical optimum solution, and it is typical np problem that TSP finds the solution problem.Allotment scheme optimization problem is the TSP special, that condition is limited, this is because the solution procedure of TSP does not have the restriction of node visit order, and there is the restriction of access order in allotment scheme optimization problem, need arrive the website of calling in goods from accessing the website of goods.

Use ant group algorithm below and find the solution above TSP special, that condition is limited.

If m ' is the quantity of node in the node set of setting out, the quantity that ant is set accordingly also is m ', and m ' ant is placed on the individual node that sets out of m '.Described ant is the variable that is used for optimizing in the ant group algorithm, n ' expression set out node set and destination node gather in all interstitial contents, d _IjBe the distance between node i and the node j, τ _Ij(t) be t path (i, j) the pheromones amount on, initial time τ constantly _Ij(t) get constant const, τ _Ij(t) determining the calculating of state transition probability.Be the state transition probability computing formula below:

$p_{\mathrm{ij}}^k\left(t\right)=\left\{\begin{array}{cc}\frac{{\left[{\mathrm{\τ}}_{\mathrm{ij}}\left(t\right)\right]}^{\mathrm{\α}}\×{\left[{\mathrm{\η}}_{\mathrm{ik}}\left(t\right)\right]}^{\mathrm{\β}}}{\underset{s\⋐\mathrm{allowedk}}{\mathrm{\Σ}}{\left[{\mathrm{\τ}}_{\mathrm{is}}\left(t\right)\right]}^{\mathrm{\α}}\×{\left[{\mathrm{\η}}_{\mathrm{is}}\left(t\right)\right]}^{\mathrm{\β}}},& j\∈{\mathrm{allowed}}_k\\ 0,& \mathrm{else}\end{array}\right.---\left(2\right)$

In the formula (2),

Be illustrated in t constantly k ant transfer to the state transition probability that the node j that the limit links to each other is arranged with i, allowed by node i _kThe node set that to be all have the limit to link to each other with the i node.α is the heuristic factor of information, the relative importance of expression track has reflected information role when ant moves that ant accumulates in motion process, its value is more big, then this ant more tends to select the path of other ant process, and cooperation is more strong between the ant.β is the heuristic factor of expectation, and the relative importance of expression visibility has reflected that ant heuristic information in motion process selects the attention degree of being subjected in the path ant.η _Ij(t) be heuristic function, if this computing node does not have then η of historical information _Ij(t) get empirical value.

T+n constantly, node (i, j) update rule of pheromones is as follows on the segmental arc:

τ _ij(t+n)＝(1-ρ)×τ _ij(t)+ρ×Δτ _ij；

$\mathrm{\Δ}{\mathrm{\τ}}_{\mathrm{ij}}=\underset{k=1}{\overset{m^{\′}}{\mathrm{\Σ}}}{\mathrm{\Δ\τ}}_{\mathrm{ij}}^k;$

$\mathrm{\Δ}{\mathrm{\τ}}_{\mathrm{ij}}^k=\left\{\begin{array}{c}\frac{Q}{L_k}\\ 0\end{array}\right.;$

K ant passes through node i constantly at moment t and t+n, j, otherwise be 0.Wherein, τ _Ij(t) expression (i, the j) value of t time information element on the segmental arc,

Expression t+n is certain computing node (i, j) value of the pheromones amount on that ant is newly experienced constantly.ρ represents the pheromones renewal factor, and 1-ρ represents the residual factor of pheromones, and ρ represents the τ that computing node arrives with great degree reference new experience here _Ij(t) value, 1-ρ represents to use for reference τ before with great degree _Ij(t) value.

According to internodal one-to-one relationship in node set and the destination node set of setting out, the destination node of the node correspondence of setting out is put into corresponding tabv _k(k=1,2 ..., m ') in.

In order to obtain the allotment scheme of global optimization, can carry out following steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant, the value of const is advisable not influence the subsequent algorithm operation result, and the node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in;

Step 2: cycle index N _c=N _C+1If, N _c＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, calculate transition probability according to the state transition probability formula, the next node of selecting node from then on to set out is namely selected the definite computing node of this node and next node;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, and the plain summation of the newly-added information here is the totalling of each child node newly-added information element; Finish if this takes turns all ant routings, then jump to step 9, otherwise jump to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is exactly the optimum solution of allotment scheme.

By all scheduling schemes are carried out finding the solution of best transportation route and minimum trucking costs, finally obtain the optimal scheduling scheme of this scheduling.

By the above as seen, static logistics deployment algorithm started before the logistics deployment program start, was used for this allotment is generated best allotment scheme.But because the state of system is among the variation always, even thereby in the process that this logistics deployment scheme has generated and carried out, also need to carry out dynamic, real-time adjustment according to change in information, to generate dynamic logistics allotment scheme.The dynamic logistics deployment algorithm is on the basis of static logistics deployment algorithm, increases the dynamic design of algorithm, dynamically deletion of node or increase node and change that corresponding algorithm flow realizes.

The dynamic logistics deployment algorithm can comprise following operation:

Corresponding this suboptimum logistics deployment scheme that has generated, use real-time perception information, determine the level of enforcement of this scheme, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades;

In addition, when initialization all k ant is placed on car hauler current location node, the static logistics deployment algorithm of application of aforementioned carries out optimization again and finds the solution.

In conjunction with the above as seen, the present invention realizes that the operation thinking of logistics deployment can represent flow process as shown in Figure 1, and this flow process may further comprise the steps:

Step 110: generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station;

Step 120: on described mathematical model basis, use ant group algorithm allotment logistics.

In order to guarantee that aforesaid operations can carry out smoothly, device as shown in Figure 2 can be set, this device comprises mathematical model generation module, the logistics deployment module that can link to each other; Wherein, the mathematical model generation module can generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station; The logistics deployment module can be used ant group algorithm allotment logistics on described mathematical model basis.

As seen, the present invention can generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station; On described mathematical model basis, use ant group algorithm allotment logistics.

The method of described logistics deployment constraint condition modeling can for:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at moment T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)) satisfies state constraint condition S _i(T (t)) 〉=y _i(T (t)); (i=1 ... n).

Can also carry out the management of logistics deployment by each logistics deployment station being set up and being safeguarded a finite state machine; Logistics deployment station a _iThe state machine function definition be: St _i(T (t)); (i=1 wherein ... n);

When state machine can not satisfy the described state constraint condition of following corresponding time point, by starting the logistics deployment scheme state machine is carried out the conversion of state, make it satisfy described state constraint condition.

Described dispensing station can comprise absolute distance modeling and relative distance modeling apart from modeling; Wherein,

During the absolute distance modeling, establish website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites:

During the relative distance modeling, if website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _IjIf website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

The modeling of described logistics deployment algorithm can be used for to-be is predicted, comprise to-be function definition, logistics deployment defined matrix; Wherein,

When carrying out the to-be function definition, the to-be anticipation function is defined as: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods; Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value;

When carrying out the logistics deployment defined matrix, set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

In addition, can use

Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine.

With

Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

The process of using ant group algorithm allotment logistics on described mathematical model basis can may further comprise the steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant; Node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in; Described ant is the variable that is used for optimizing in the ant group algorithm;

Step 2: cycle index N _c=N _C+1If, Nc＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, according to state transition probability formula calculating transition probability, the next node of selecting node from then on to set out;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, finishes if this takes turns all ant routings, then jumps to step 9, otherwise jumps to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is as the optimum solution of allotment scheme.

This suboptimum logistics deployment scheme that can also corresponding generate, use real-time perception information, determine whether this scheme needs dynamically to adjust, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades; All k ant is placed on car hauler current location node, carries out optimization again and find the solution.

Logistics deployment implement device of the present invention can comprise mathematical model generation module, logistics deployment module; Wherein,

Described mathematical model generation module is used for generating for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station;

Described logistics deployment module is used for using ant group algorithm allotment logistics on described mathematical model basis.

Described mathematical model generation module can comprise logistics deployment constraint condition modeling unit, is used for:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at the moment in future T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)) satisfies state constraint condition S _i(T (t)) 〉=y _i(T (t)); (i=1 ... n).

Described logistics deployment constraint condition modeling unit can also be used for:

By each logistics deployment station being set up and being safeguarded a finite state machine, carry out the management of logistics deployment; Logistics deployment station a _iThe state machine function definition be: St _i(T (t)); (i=1 wherein ... n);

When state machine can not satisfy the described state constraint condition of following corresponding time point, state machine is carried out the conversion of state.

Described mathematical model generation module can comprise dispensing station apart from modeling unit, is used for carrying out absolute distance modeling and relative distance modeling; Wherein,

During the absolute distance modeling, establish website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites:

During the relative distance modeling, if website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _IjIf website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

Described mathematical model generation module can comprise logistics deployment algorithm modeling unit, is used for to-be is predicted, comprises to-be function definition, logistics deployment defined matrix; Wherein,

When carrying out the to-be function definition, the to-be anticipation function is defined as: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods; Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value;

When carrying out the logistics deployment defined matrix, set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

Described logistics deployment algorithm modeling unit can also be used for:

With Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine.

With

Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

When described logistics deployment module is used ant group algorithm allotment logistics on the mathematical model basis, can carry out following steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant; Node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in; Described ant is the variable that is used for optimizing in the ant group algorithm;

Step 2: cycle index N _c=N _C+1If, N _c＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, according to state transition probability formula calculating transition probability, the next node of selecting node from then on to set out;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, finishes if this takes turns all ant routings, then jumps to step 9, otherwise jumps to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is as the optimum solution of allotment scheme.

This suboptimum logistics deployment scheme that described logistics deployment module can also corresponding generate, use real-time perception information, determine the level of enforcement of this scheme, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades; All k ant is placed on car hauler current location node, carries out optimization again and find the solution.

In sum as seen, no matter be method or the device of this method of realization, the present invention realizes that the technology of logistics deployment can generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station; And on described mathematical model basis, use ant group algorithm and allocate logistics, therefore can adapt to the dynamic variation characteristic in the logistics deployment process, in minimum number, minimum zone, generate the best logistics deployment scheme that approaches, this can improve logistics deployment efficient to a great extent, realize the high-level efficiency logistics deployment, and save the logistics deployment cost.

The above is preferred embodiment of the present invention only, is not for limiting protection scope of the present invention.

Claims (16)

1. the implementation method of a logistics deployment is characterized in that, this method comprises:

Generate for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station; On described mathematical model basis, use ant group algorithm allotment logistics.

2. method according to claim 1 is characterized in that, the method for described logistics deployment constraint condition modeling is:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at moment T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)) satisfies state constraint condition S _i(T (t)) 〉=y _i(T (t)); (i=1 ... n).

3. method according to claim 2 is characterized in that, this method also comprises:

By each logistics deployment station being set up and being safeguarded a finite state machine, carry out the management of logistics deployment; Logistics deployment station a _iThe state machine function definition be: St _i(T (t)); (i=1 wherein ... n);

When state machine can not satisfy the described state constraint condition of following corresponding time point, by starting the logistics deployment scheme state machine is carried out the conversion of state, make it satisfy described state constraint condition.

4. method according to claim 1 is characterized in that, described dispensing station comprises absolute distance modeling and relative distance modeling apart from modeling; Wherein,

During the absolute distance modeling, establish website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites:

During the relative distance modeling, if website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _IjIf website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

5. method according to claim 1 is characterized in that, the modeling of described logistics deployment algorithm is used for to-be is predicted, comprises to-be function definition, logistics deployment defined matrix; Wherein,

When carrying out the to-be function definition, the to-be anticipation function is defined as: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods; Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value;

When carrying out the logistics deployment defined matrix, set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

6. method according to claim 5 is characterized in that, this method also comprises:

With

Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine;

With

Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

7. according to each described method of claim 1 to 6, it is characterized in that the process of using ant group algorithm allotment logistics on described mathematical model basis may further comprise the steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant; Node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in; Described ant is the variable that is used for optimizing in the ant group algorithm;

Step 2: cycle index N _c=N _C+1If, N _c＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, according to state transition probability formula calculating transition probability, the next node of selecting node from then on to set out;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, finishes if this takes turns all ant routings, then jumps to step 9, otherwise jumps to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is as the optimum solution of allotment scheme.

8. method according to claim 7 is characterized in that, this method also comprises:

Corresponding this suboptimum logistics deployment scheme that has generated, use real-time perception information, determine whether this scheme needs dynamically to adjust, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades; All k ant is placed on car hauler current location node, carries out optimization again and find the solution.

9. the implement device of a logistics deployment is characterized in that, this device comprises mathematical model generation module, logistics deployment module; Wherein,

Described mathematical model generation module is used for generating for the mathematical model that realizes logistics deployment apart from modeling, the modeling of logistics deployment algorithm by the modeling of logistics deployment constraint condition, distribution station;

Described logistics deployment module is used for using ant group algorithm allotment logistics on described mathematical model basis.

10. device according to claim 9 is characterized in that, described mathematical model generation module comprises logistics deployment constraint condition modeling unit, is used for:

If logistics deployment system is total n logistics deployment station always, each distribution station is expressed as a _i(i=1 ... n), at the moment in future T (t), distribution station a _iCertain logistics goods tank farm stock be S _i(T (t)), and this constantly this distribution station be y to the demand of this logistics goods _i(T (t)) satisfies state constraint condition S _i(T (t)) 〉=y _i(T (t)); (i=1 ... n).

11. device according to claim 10 is characterized in that, described logistics deployment constraint condition modeling unit also is used for:

By each logistics deployment station being set up and being safeguarded a finite state machine, carry out the management of logistics deployment; Logistics deployment station a _iThe state machine function definition be: St _iT (t)); (i=1 wherein ... n);

When state machine can not satisfy the described state constraint condition of following corresponding time point, state machine is carried out the conversion of state.

12. device according to claim 9 is characterized in that, described mathematical model generation module comprises dispensing station apart from modeling unit, is used for carrying out absolute distance modeling and relative distance modeling; Wherein,

During the absolute distance modeling, establish website a _iWith website a _jBetween distance be d _Ij, set up the two-dimensional matrix of distance between the different websites:

During the relative distance modeling, if website a _iTo website a _jBetween path on do not have other websites, then establish website a _iTo website a _jBetween distance be d _IjIf website a _iTo website a _jBetween have the 3rd site k, then d _Ij=d _Ik+ d _Kj

And the like, the distance table between all different websites be shown the nondecomposable dot spacing of minimum station in twos from and:

Wherein, d ' _IjExpression website a _iTo website a _jBetween relative distance, be expressed as minimum station dot spacing in twos from relative value.

13. device according to claim 9 is characterized in that, described mathematical model generation module comprises logistics deployment algorithm modeling unit, is used for to-be is predicted, comprises to-be function definition, logistics deployment defined matrix; Wherein,

When carrying out the to-be function definition, the to-be anticipation function is defined as: Num _i=fc (st _i(T (t)), y _i(T (t)), S _i(T (t)));

If Num _i＞0, expression distribution station a _iRemaining logistics goods is arranged, and number is Num _iIf Num _i=0, expression distribution station a _iDo not remain the logistics goods; Num _i〉=0 o'clock, expression distribution station a _iDo not need to replenish the logistics goods by the logistics deployment program; If Num _i＜0, expression distribution station a _iNeed to start the logistics deployment program, and be formulated to distribution station a _iThe logistics quantity of goods be Num _iAbsolute value;

When carrying out the logistics deployment defined matrix, set up the logistics deployment two-dimensional matrix between distribution station, if distribution station a _iDo not need to distribution station a _jThe allotment goods is then represented with 0; Allocate if desired, then be expressed as the logistics goods numerical value t of allotment _Ij:

14. device according to claim 13 is characterized in that, described logistics deployment algorithm modeling unit also is used for:

With

Expression transports distribution station a to _jThe summation of all logistics goods, need satisfy condition:

Namely transport distribution station a to _jThe logistics goods want to satisfy the requirement of to-be machine;

With

Expression is from distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy condition:

Be distribution station a _iTransport the logistics goods summation of other all distribution stations to, need satisfy distribution station a _iFollowing state machine constraint condition.

15. according to each described device of claim 9 to 14, it is characterized in that, when described logistics deployment module is used ant group algorithm allotment logistics on the mathematical model basis, be used for carrying out following steps:

Step 1: set maximum cycle N _{C max}, ant number m ', initialization time sheet t=0, loop control variable N _c=0, ant loop variable k=0, taboo table tabu _k(k=1,2 ..., m '), the dynamic adjustment form tabv of node _k(k=1,2 ..., m '), routing table path _k(k=1,2 ..., m '), m ' ant is placed on the individual node that sets out of m ', make the pheromones τ on the continuous arc of different nodes _Ij(t)=and const, wherein const is constant; Node in all destination node set is put into the dynamic adjustment form tabv of node _k(k=1,2 ..., m ') in; Described ant is the variable that is used for optimizing in the ant group algorithm;

Step 2: cycle index N _c=N _C+1If, N _c＞N _{C max}, withdraw from circulation, jump to step 11;

Step 3: ant number k=k+1 if k＞m withdraws from the k circulation, jumps to step 8;

Step 4: determine the node at the present place of ant, detect this node and whether belong to destination node set, if, then with the destination node of this node correspondence from the dynamic adjustment form tabv of node _kMiddle deletion, otherwise jump to next step;

Step 5: seek the path link to each other with ant place node, according to state transition probability formula calculating transition probability, the next node of selecting node from then on to set out;

Step 6: revise taboo table tabu _k, the node of newly selecting is put into the taboo table; And carry out the renewal of pheromones;

Step 7: if taboo table tabu _kComprised all nodes in all set out node set and destination node set, then this routing of ant finishes, and jumps to step 9, otherwise jumps to step 5;

Step 8: this takes turns the plain summation of the newly-added information of selecting node to calculate and record ant, finishes if this takes turns all ant routings, then jumps to step 9, otherwise jumps to step 3;

Step 9: the plain summation of newly-added information of the combination computing node that all ants are selected in relatively should taking turns, get the computing node composite sequence of the plain total value maximum of newly-added information, the pheromones on the respective paths is carried out the renewal of pheromones according to the pheromones update rule;

Step 10: empty taboo table tabu _k, routing table path _k, jump to step 2;

Step 11: the allotment route that the pheromones maximum path is represented is as the optimum solution of allotment scheme.

16. device according to claim 15 is characterized in that, described logistics deployment module also is used for:

Corresponding this suboptimum logistics deployment scheme that has generated, use real-time perception information, determine the level of enforcement of this scheme, for allocating finishing of task, the corresponding node of deletion follow node set and the destination node set, the corresponding path of deletion from the optimal scheduling path; For the current scheduler task that will delete, increase newly, revise, deletion in node set and the destination node set of setting out, newly-increased corresponding node; Determine the node location at the current place of car hauler or the node location that be about to arrive, carry out generating corresponding node state figure after up-to-date state upgrades; All k ant is placed on car hauler current location node, carries out optimization again and find the solution.

Images