CN114169560A

CN114169560A - Material scheduling control method for stereoscopic warehouse

Info

Publication number: CN114169560A
Application number: CN202011526353.7A
Authority: CN
Inventors: 罗悦; 李贺; 郝睿智; 沈金洋
Original assignee: Sichuan Hezong Yisco Pharmaceutical Co ltd
Current assignee: Sichuan Hezong Yisco Pharmaceutical Co ltd
Priority date: 2020-12-22
Filing date: 2020-12-22
Publication date: 2022-03-11
Anticipated expiration: 2040-12-22
Also published as: CN114169560B

Abstract

The invention provides a material scheduling control method for a stereoscopic warehouse, which comprises the following steps: determining an initial strategy to solve an optimization problem for dispatching a plurality of predetermined materials to a plurality of predetermined locations using a plurality of vehicles; scheduling and adding a plurality of non-priority materials into a plurality of priority routes in the current strategy; a plurality of tabu search algorithms are executed to optimize the initial strategy. The invention provides a material scheduling control method for a stereoscopic warehouse, which adaptively expands a VRP solution strategy according to the change of actual transportation requirements in the material scheduling process of the stereoscopic warehouse and obtains an optimal solution with higher efficiency.

Description

Material scheduling control method for stereoscopic warehouse

Technical Field

The invention relates to intelligent logistics, in particular to a material scheduling control method for a stereoscopic warehouse.

Background

In a logistics scenario, the VRP strategy of a material scheduling route is an important issue. VRPs typically include the problem of using multiple vehicles to transport multiple materials to multiple locations. In solving the problem, it is desirable to define a set of routes that meet vehicle capacity, time, or other constraints so that the total cost of all transportation routes can be minimized. Heuristic algorithms for tabu search have been effective in solving some of the large and practical problems. While the above-described schema is effective for solving VRPs in some applications, the VRP solution strategy cannot be extended to account for changes in actual transportation needs.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a material scheduling control method for a stereoscopic warehouse, which comprises the following steps:

determining an initial strategy to solve an optimization problem for scheduling a plurality of predetermined materials to a plurality of predetermined locations using a plurality of vehicles, the initial strategy comprising selecting one or more routes from the plurality of routes to accomplish the material scheduling and minimizing a total cost of all selected routes, the initial strategy generated independent of a priority of the materials;

adding a plurality of non-priority material schedules into a plurality of priority routes in a current strategy, wherein the non-priority material schedules are positioned in the neighborhood of the priority routes, and the priority routes are routes comprising at least one priority material schedule; and

executing a plurality of tabu search algorithms to optimize the initial strategy, the tabu search algorithms comprising one of:

dividing one or more selected priority routes in the current strategy into two new priority routes;

moving the set of pivot points from one of the two routes to the other of the two routes for one or more selected priority routes in the current strategy and exchanging the pivot points between the priority routes in the two routes; and

dividing the indirect priority route into a plurality of shortened priority routes not including the midway pivot point for a plurality of selected priority routes in the indirect priority route having the midway pivot point in the current strategy, and for the one or more selected priority routes, performing the above-mentioned steps of moving the set of pivot points from one of the two routes to the other of the two routes and exchanging the pivot points between the priority routes in the two routes, and wherein each pair of selected priority route combinations includes a shortened priority route.

Preferably, the priority material schedule is a material schedule that needs to be transported at a current time, and the non-priority material schedule is a material schedule that can be transported at a future time.

Preferably, before the non-priority materials are scheduled and added into the current task, the priority routes are reversely sorted according to the priority, and if the priority routes with the same priority exist, the priority routes are reversely sorted according to the full load rate; and selecting a priority route according to the sequence to add a non-priority material scheduling task.

Preferably, the dividing one or more selected priority routes in the current strategy into two new priority routes further comprises:

determining, for each priority route, a saving distance associated with the division of the priority route; partitioning according to a reverse ordering of saved distances; selecting a first subdivision for evaluation according to the sorted order;

according to the selected subdivision, the priority route is temporarily divided into two new priority routes, a plurality of non-priority material schedules are added into the two or more priority routes, the non-priority material schedules are located in the neighborhoods of the two priority routes, if the unit cost of all the priority routes is optimized due to the division of the two new priority routes after the non-priority material schedules are added, the division is accepted, otherwise, the division is discarded, and the next subdivision is selected for estimation.

Preferably, the moving the set of pivot points from one of the two routes to the other of the two routes for one or more selected priority routes in the current strategy and exchanging the pivot points between the priority routes in the two routes further comprises:

determining, for the selected priority route combination, a quadruple of movement and interchange tasks between the priority routes, the quadruple of movement tasks representing a first priority route of the route combination, a second priority route of the route combination, a set of pivot points of the first priority route to be moved, and a location where the set of pivot points is to be added to the second priority route, and the quadruple of interchange tasks representing a first priority route of the route combination, a second priority route of the route combination, a first set of pivot points of the first priority route, and a first pivot point of the second priority route to be interchanged with the first set of pivot points;

determining a saving distance associated with each quadruple; sorting the quadruples reversely according to the saved distance; selecting the first quadruple according to the sorted order for estimation; temporarily performing the move or swap task associated with the selected quadruplet to generate two new priority routes, and accepting the move or swap task if the unit cost of all priority routes is optimized for the move or swap task, otherwise discarding the move or swap task and selecting the next quadruplet for evaluation.

Preferably, a destination distance associated with each priority route is determined; the quadruplets are determined only for priority route combinations having an associated destination distance exceeding a predetermined threshold.

Compared with the prior art, the invention has the following advantages:

the invention provides a material scheduling control method for a stereoscopic warehouse, which adaptively expands a VRP solution strategy according to the change of actual transportation requirements in the material scheduling process of the stereoscopic warehouse and obtains an optimal solution with higher efficiency.

Drawings

Fig. 1 is a flowchart of a material scheduling control method for a stereoscopic warehouse according to an embodiment of the present invention.

Detailed Description

A detailed description of one or more embodiments of the invention is provided below along with accompanying figures that illustrate the principles of the invention. The invention is described in connection with such embodiments, but the invention is not limited to any embodiment. The scope of the invention is limited only by the claims and the invention encompasses numerous alternatives, modifications and equivalents. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. These details are provided for the purpose of example and the invention may be practiced according to the claims without some or all of these specific details.

One aspect of the present invention provides a material scheduling control method for a stereoscopic warehouse. Fig. 1 is a flowchart of a material scheduling control method for a stereoscopic warehouse according to an embodiment of the present invention.

The invention adopts a method for dispatching a cargo transportation route according to priority, which comprises the steps of firstly determining an initial strategy and solving the optimization problem of dispatching a plurality of materials to a plurality of places by a plurality of vehicles, wherein the initial strategy comprises the step of setting a plurality of routes so as to minimize the total cost of all the routes, and the initial strategy is generated independently of the priority of the materials.

If a material scheduling task must be executed immediately, the material scheduling task is called as priority; otherwise, the material scheduling task is said to be non-prioritized. If a route contains a priority material schedule, the route is said to be priority.

Adding a plurality of non-priority material schedules into each route of a plurality of priority routes in the current strategy, wherein the non-priority material schedules are positioned in the neighborhood of the priority routes, and the priority routes are routes containing priority material schedules. The neighborhood of the route is defined as a material scheduling task or route whose destination is within the maximum adjacency of the route's destination. Executing a plurality of tabu search algorithms to optimize the initial strategy, the tabu search algorithms comprising one of: (a) dividing each route of a plurality of selected priority routes in the current strategy into two new priority routes; (b) moving the set of pivot points from one of the two routes to the other of the two routes and/or swapping pivot points between the priority routes of the two routes for each of a plurality of selected priority routes in the current strategy; and (c) dividing a plurality of indirect priority routes having midway pivot points in the current strategy into a plurality of shortened priority routes not including the midway pivot points, and performing the process as described in (b) on a plurality of selected priority routes, wherein each selected priority route combination comprises the shortened priority routes.

In certain embodiments, for an optimized strategy that takes into account material scheduling priorities, a relatively complete route of multiple priority materials can be better indicated. This policy can be considered a first-level approximation policy for multi-stage VRPs. Preferably, a plurality of tabu search algorithms are used to identify the strategy for the problem. The tabu search algorithm is simple and effective, and can provide a better strategy for large-scale actual logistics problems within a reasonable time range.

The present invention obtains the above-described VRP approximation solution for multiple epochs by solving only the VRP for the current epoch. The stereoscopic warehouse logistics system is designed with the goal of minimizing the unit cost of the priority route and maximizing the total traffic volume of the efficient route. The problem addressed according to the present invention is the above-mentioned VRP containing priority constraints.

In the initial stage, the information representing the materials to be transported is determined and used, and the initial strategy of the basic VRP is generated according to the information of the transported materials. The initial strategy includes all material scheduling without regard to priority. The initial strategy schedules all material in the problem space to be equal, i.e. either prioritized or non-prioritized material scheduling. Parameters for an initial policy to perform a plurality of calculations are determined. In the modeling phase, future scheduled task calculations are performed, filling each priority route with future scheduled tasks in the neighborhood of the priority route. Preferably, if the initial strategy contains all material schedules, the future schedule task calculation can minimize the total cost of all transport routes by moving some non-priority material schedules from non-priority routes to non-fully loaded priority routes.

In one embodiment, when adding future scheduled tasks to a priority route associated with a current strategy, all priority routes are first sorted in the current strategy according to priority and all priority routes are sorted in the priority according to decreasing order of fullness for allowing the most important and most laden priority route to be filled with material scheduling.

And selecting the first priority route after sorting, determining the neighborhood of the first priority route by using the maximum adjacent distance parameter, and determining all future scheduling tasks of which the destinations are all positioned within the maximum adjacent distance of the destinations of the priority route and material scheduling which is not positioned in the priority route. The future scheduled tasks are ordered according to decreasing priority, increasing distance proportion and decreasing full load rate. The distance increasing proportion refers to the rate of increasing the distance of a route after a specific pivot point is added with new material scheduling.

And selecting a first future scheduling task, and adding the future material scheduling into the priority route. If the added task does not increase the total cost by the value of the maximum allowed threshold parameter, the added task is accepted.

If another priority route needs to be considered, the above calculation is repeated until all priority routes are considered.

In the improvement phase, a plurality of tabu search algorithms may be executed to improve the initial strategy. The tabu search algorithm starts with a successful strategy stereoscopic warehouse and then uses an improvement process to search for a better strategy in the neighborhood of the strategy. As soon as there is an improved policy, it is taken immediately and then the search of its neighborhood is repeated to find a new policy. The tabu search algorithm will stop when a local optimization condition is reached.

Preferably, in the improvement phase, one of a plurality of the following tabu search algorithms is executed: (1) dividing a priority route: if enough future scheduling tasks are available to generate two valid routes in the neighborhood of the priority route, dividing one priority route into two routes; (2) optimizing a priority route: moving a set of pivot points from one route to another, or swapping pivot points between two routes, removing future scheduled tasks to make room for a priority material schedule, or adding future scheduled tasks to fill a priority route); and (3) re-optimizing the priority route: an unloaded priority route is segmented into a set of shortened routes and the re-optimized priority route is applied to the new route and all other priority routes.

For the calculation of the prioritized routes, each prioritized route is performed for one prioritized route. The most important and most loaded priority route is divided first. Once the prioritized route calculation has been applied to all the prioritized routes in the present optimal strategy, a new optimal strategy is selected.

Likewise, under the present optimal strategy, each priority route optimization calculation is performed for a single priority route combination for the calculation of the priority route optimization according to a strategy description program or other suitable instructions. Once the priority route optimization calculation has been applied to all suitable priority routes in the current optimal strategy, a new optimal strategy will be selected.

Likewise, for the calculation of the priority route re-optimization, each priority route re-optimization calculation is performed for one priority route. According to priority, to the priority routes such that the most important and most loaded priority route is interrupted first. When in the present optimal strategy, the priority route re-optimization is applied to all priority routes, a new optimal strategy will be selected.

When dividing the priority route into a plurality of new priority routes, three parameters, namely a minimum full load rate, a maximum adjacent distance and a maximum distance increasing proportion, are determined firstly. The minimum fullness parameter is used to define whether the route is valid. The maximum adjacent distance and the maximum distance increasing proportion are used for calculating the future scheduling task after the priority route is divided. Then, under the current strategy, all the priority routes are searched in a sorted manner according to the priority, and all the priority routes are searched in a sorted manner according to the decreasing of the full load rate. A first priority route having a plurality of pivot points and a full load rate greater than the minimum full load rate is selected.

Preferably, if the route has N pivot points, only a subdivision of the priority route into two routes of respective pivot points (1, 2.. k) and (k + 1...., N) is considered. I.e. there are N-1 subdivisions. For a subdivision of pivot point k, the distance savings for that subdivision may be calculated as follows:

dis_save＝dis(tr_k,tr_k+1)

wherein dis_saveIndicating the distance saved, tr_k,tr_k+1Are respectively provided withIs the position of the kth and (k + 1) th pivot points, dis () is the distance between the two parameters.

All the subdivisions of N-1 possible priority routes are sorted according to the saved distance. Estimating whether there are subdivisions by calculating all future scheduled tasks in the neighborhood of the priority route can generate two valid routes. Any subdivisions that result in invalid routes are discarded.

Selecting dis with the largest saving distance from all sub-partitions_saveTo divide the priority route into two new routes and to perform future scheduled task calculations using the parameters with said maximum adjacency distance and maximum distance-increase ratio to fill the divided two new routes with future scheduled tasks.

If the full rate of both new routes is greater than the minimum full rate and the unit cost of all priority routes is reduced, then all corrections associated with the subdivisions are accepted. Otherwise, the current correction is discarded, and then the next subdivision is entered until all subdivisions are traversed. Where the unit cost is expressed as a cost per distance unit per unit of full rate.

Wherein, in calculating the priority route optimization, the method comprises a moving sub-step to move the set of pivot points from one priority route to another priority route, or a swapping sub-step to swap pivot points between two priority routes, and further comprises removing space for future scheduled tasks to generate future scheduled tasks, or adding future scheduled tasks to fill in the priority routes.

The move sub-step is represented by a quadruple (a, B, S, P), where S is the set of pivot points of route a to be added at position P of route B. The swapping sub-step is represented by a quadruple (A, B, S1, S2), where S1 is the set of pivot points for route A to be swapped with the set of pivot points for route B, S2.

When optimizing the priority route, firstly determining the following parameters: maximum adjacency distance, maximum distance-increase ratio, maximum destination distance, minimum full load rate, and minimum distance-saved parameter. A first pair of priority routes, referred to as routes A and B, is then selected. If the destination distance is greater than the maximum destination distance parameter, then two priority routes are discarded. Otherwise, the quadruplets are generated for all possible movement and exchange sub-steps between the two priority routes. Two valid routes may be generated by calculating all future scheduled tasks in the two priority route neighborhoods to estimate if there are other sub-steps. And ordering all the quadruples for all the route combinations according to the saved distances calculated according to the priority-dividing routes.

Selecting the quadruple with the largest saved distance, and performing correction on the two routes according to the largest saved distance to generate two new routes. If the full rate of both new routes exceeds the minimum full rate and the unit cost of all priority routes is reduced, all corrections associated with the sub-steps will be accepted; otherwise, the sub-steps are rolled back and the next quadruple is performed.

To segment the priority routes of the unloaded pivot points into a set of direct priority routes and apply the calculation of priority route optimization to the new direct priority route and all other priority routes, in the current strategy all priority routes are sorted according to priority and all priority routes are sorted in reverse of the load factor, and then the first priority route having a number of pivot points and having a load factor greater than the minimum load factor is selected. Dividing the priority route into a set S of direct priority routes, generating four-tuple for all priority routes which are not contained in the set S by taking the direct priority route as a root for each direct priority route in the set S, and carrying out optimization calculation on the priority route similar to the optimization calculation on the priority route to continue operation.

It is estimated whether it is possible to generate two valid routes by calculating all future scheduled tasks in the neighborhood of the priority route combination associated with the quadruple, and discarding the quadruple that caused the invalid route. The quadruple with the largest saved distance is selected and modified to generate two new routes. If the preset capacity limit is not met, all future scheduled tasks will be removed from the two new routes and then the two new routes are filled with future scheduled tasks using the future scheduled task calculation process. If the preset capacity limit is met, it is directly determined whether the full load rates of the two new routes exceed the minimum full load rate and the unit costs of all priority routes are reduced, and if so, all temporary modifications associated with the sub-step are accepted. If not, the current sub-step is rejected, the temporary modification is discarded, and an attempt is made to process the next quad.

Preferably, if all temporary modifications associated with this sub-step are accepted, i.e. routes a and B are changed to routes a 'and B', all remaining quadruplets containing a or B are discarded and the prioritized routes are then sorted again.

According to another aspect of the invention, after the material dispatching route is optimized, the operation efficiency of the dispatching strategy is re-estimated. The method comprises the following steps: receiving the quantity to be scheduled, time constraint and cost constraint of the materials input by a user; applying the quantity to be scheduled to a first optimal route of the logistics network; calculating a first success probability of the current material in the first optimal route of the logistics network according to the quantity to be scheduled, the time constraint and the cost constraint; if the first optimal route is in a failure state, applying the quantity to be scheduled to a second optimal route of the logistics network, and calculating a first failure probability of the current material transported in the first optimal route of the logistics network; calculating a second success probability of the material transported in a second optimal route of the logistics network according to the quantity to be scheduled, the time constraint and the cost constraint; and integrating the first success probability, the first failure probability and the second success probability into the material scheduling success probability of the logistics network through a preset calculation process, and taking the material scheduling success probability as the operation efficiency of the three-dimensional warehouse logistics system.

Wherein the step of applying the quantity to be scheduled to the optimal route comprises determining a plurality of optimal routes of the logistics network, wherein the optimal route is an ordered set of a plurality of priority route segments between the stereoscopic warehouse and the destination without any circulation; calculating the flow of each optimal route; and converting the flow of each optimal route into the current full rate of each priority route segment.

Wherein the step of calculating the flow of each optimal route comprises providing lead time for each priority route segment of the optimal route; solving the flow of the optimal route by using an objective function with scheduling time less than or equal to time constraint, wherein the scheduling time is equal to the sum of the result of dividing the number to be scheduled by the flow of the optimal route and a plurality of lead times; and when the flow of the optimal route is smaller than the capacity constraint of the optimal route, judging that the optimal route has an optimal vector.

The step of calculating the success probability comprises the steps of solving the maximum traffic volume transmitted by each priority route segment of the optimal route in each unit time according to the number to be scheduled, the time constraint and the cost constraint, and an objective function that the scheduling time for transmitting the number to be scheduled by each optimal route is less than or equal to the time constraint;

forming a full load vector by a plurality of full load rates of a plurality of priority route segments of each optimal route, setting the numerical values of the full load rates to be randomly changed so as to correspond to the flow distribution state of the three-dimensional warehouse logistics system; performing a cost check to check whether the scheduling cost of transmitting the quantity to be scheduled per optimal route exceeds a cost constraint; when the scheduling cost of each optimal route is less than or equal to the cost constraint and the scheduling time is less than or equal to the time constraint, calculating the probability that the full load vector of any route of the logistics network is greater than or equal to the optimization vectors of the optimal routes, and defining the probability as the success probability of the logistics network.

The cost checking process comprises the steps of calculating the scheduling cost when the optimal route transmits the materials of the quantity to be scheduled; comparing the scheduling cost with the value of the cost constraint; and judging whether the optimal route has an optimal vector according to the size relation between the scheduling cost and the cost constraint.

After the logistics model is established, the invention preferably uses an artificial bee colony algorithm, adaptively adjusts the search span according to the pheromone concentration of the bee colony, promotes the algorithm to quickly converge to the global optimal solution, and enhances the optimizing capability of the algorithm.

Firstly, the search span is dynamically adjusted according to the pheromone concentration of the bee colony, then the bee colony position is updated under the guidance of the self-adaptive search span, and meanwhile, the global and local optimal solution searching capability of the algorithm is improved.

Step a 1: parameters and the location of the bee colony are initialized. A bee individual is mapped to a solution. Setting the size N of the bee colony, the maximum generation number G of the algorithm, and the initial position (X) of the bee colony_axis，Y_axis) And pheromone concentration in the bee colony.

X_axis＝rand()

Y_axis＝rand()

Wherein rand () represents a random number of [0, 1 ].

Step a 2: updating the search span, and dynamically correcting the search span V of the ith bee according to the concentration of pheromone_i。

When Taste_i≥Taste_avgThe method comprises the following steps:

V_i＝V₂-[(V₂-V₁)(Taste_i-Taste_avg)]/Taste_max-Taste_avg()

when Taste_i＜Taste_avgThe method comprises the following steps:

V_i＝V₂

wherein, V_iAnd Taste_iThe span and pheromone concentration of the ith colony, respectively. V₁And V₂Respectively, the boundary values of the search span variation range. Taste_avgAnd Taste_maxMean and maximum pheromone concentrations in the population are shown, respectively.

In the initial stage of the algorithm, the bee colony is far from the optimal solution, the pheromone concentration is small, at the moment, the search span is large, and the bee colony is promoted to move towards food quickly, so that the search speed and the global optimization capability of the algorithm are improved. Along with evolution iteration of the algorithm, the bee colony is gradually concentrated near the optimal solution, and the search span is gradually reduced, so that the convergence precision and the local optimization capability of the algorithm are improved.

Step a 3: and updating the bee colony position. In the search span V_iUnder the guidance of (2), dynamically updating the position (X) of the ith bee_i，Y_i)：

X_i＝X_axis+V_i

Y_i＝Y_axis+V_i，

Step a 4: pheromone concentration parameters were calculated. Pheromone concentration parameter S_iFor the value of the optimal solution for the colony distance, the values of the colony and origin distance are Dit_iIs composed of

S_i＝1/Dit_i

Step a 5: the pheromone concentration was calculated. Pheromone concentration function Taste_iAnd (3) representing the optimization effect of the bee colony, and taking the objective function min of the logistics network as an pheromone concentration function.

Step a 6: and d, determining an optimal bee colony, searching the bee colony with the minimum pheromone concentration value in the current population according to the result of the step a5, and taking the bee colony as the optimal bee colony.

[optiTaste，optiIndex]＝min[Taste_i()]，

Step a 7: recording the optimal solution information, and recording the pheromone concentration value and the position of the optimal bee colony according to the result of the step a 6.

Tasteopti＝optiTaste，

X_axis＝X(optiIndex)，

Y_axis＝Y(optiIndex)，

Step a 8: and when the algorithm evolves to a designated generation number, the algorithm is ended, and a result is output. Otherwise, steps a3-a6 are repeated.

In summary, the invention provides a material scheduling control method for a stereoscopic warehouse, which adaptively expands a VRP solution strategy according to the change of the actual transportation demand in the material scheduling process of the stereoscopic warehouse, and obtains an optimal solution with higher efficiency.

It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented in a general purpose computing system, centralized on a single computing system, or distributed across a network of computing systems, and optionally implemented in program code that is executable by the computing system, such that the program code is stored in a storage system and executed by the computing system. Thus, the present invention is not limited to any specific combination of hardware and software.

It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.

Claims

1. A material scheduling control method for a stereoscopic warehouse is characterized by comprising the following steps:

2. The method of claim 1, wherein the priority material schedule is a material schedule that requires transportation at a current time and the non-priority material schedule is a material schedule that may be transported at a future time.

3. The method of claim 1, further comprising:

before non-priority materials are scheduled and added into a current task, the priority routes are reversely sorted according to the priority, and if the priority routes with the same priority exist, the priority routes are reversely sorted according to the full load rate; and selecting a priority route according to the sequence to add a non-priority material scheduling task.

4. The method of claim 1, wherein the dividing one or more selected priority routes in the current strategy into two new priority routes further comprises:

5. The method of claim 1, wherein moving the set of pivot points from one of the two routes to the other of the two routes and swapping pivot points between the two routes for one or more selected priority routes in the current strategy further comprises:

6. The method of claim 1, further comprising: determining a destination distance associated with each priority route; the quadruplets are determined only for priority route combinations having an associated destination distance exceeding a predetermined threshold.