JP2003109170A - Method and system for preparing schedule of transportation - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and a system for preparing a schedule for transportation in which the solution of accuracy equal to the preparation by an expert can be found within an interactively available time corresponding to various evaluation references when preparing a physical distribution schedule for a plurality of trucks. SOLUTION: A base position such as a delivery destination, luggage information on a luggage to be transported to the delivery destination, transportation means information composed of information such as trucks to be utilized for transportation and purpose information composed of priority in excess time or travel distance concerning the evaluation of the transportation schedule are inputted. On the basis of such information, a transportation route is generated. When generating the transportation route, while searching the optimal schedule by evaluating the schedules according to the purpose information and on the basis of the other input information, travel schedules for a plurality of transportation means are simultaneously prepared. In particular, the transportation schedule is prepared by using a so-called genetic algorithm for preparing an initial solution group composed of a plurality of initial solutions and improving the prepared initial solution group in genetic processing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to transportation (delivery) in a physical distribution system in which delivery is performed from one delivery point to a large number of points using a plurality of transportation trucks within a certain area.
A method and system for creating a plan.

[0002]

2. Description of the Related Art Transportation planning in a physical distribution system corresponds to a so-called traveling sales problem, and various solutions have been proposed. As a simple solution, NN (Nearest Neighbor) method and NI (Nearest I
nsertion) method (Yoshitsugu Yamamoto, Mikio Kubo, Invitation to Traveling Salesman Problem, pp.69-711997, Asakura Shoten), or improve the initial solution to find a route with higher optimality 2o
pt method (Yoshitsugu Yamamoto, Mikio Kubo, invitation to traveling salesman problem, p.94, 1997, Asakura Shoten) and LK method (Toshihide Ibaraki,
(Discrete optimization method and algorithm, p.86, 1993, Iwanami Shoten)
And so on.

Further, a mathematically exact optimum solution can be obtained.
In order to deal with complicated constraint conditions appearing in practical problems and solutions using DP (Dynamic Programming) and polyhedral approach (Toshihide Ibaraki, optimization method, pp.71-103, Kyoritsu Shuppan, 1993) Genetic Algorithm:
(Hereinafter abbreviated as GA) (Hiroaki Kitano, "Genetic Algorithm," 1993 Industrial Book Publishing, or JP-A-7-105180), Simulated Annealing method, metaheuristics such as tabu search (edited by Colin R. Reeves). , Modern Heuristics, Nikkan Kogyo Shimbun, 1997) has also been developed.

[0004]

When actually constructing a physical distribution network, it is necessary to deeply consider the cost and effectiveness of the physical distribution network in advance, which involves decision making at the strategic level of a company or the like.

[0005] Such decision making cannot be performed simply by mathematical optimization, and it can be said that multifaceted human judgments such as economic, social and environmental aspects are essential. As described above, in order to efficiently judge the distribution network from various aspects, it is necessary to simulate and examine the distribution plan by a method based on various conditions and evaluation criteria. In addition, the simulation of the physical distribution plan requires a speed at which interactive processing is possible and the optimization accuracy of the solution at the same level as a human expert.

When a human expert prepares a transportation plan, he / she does not always seek an optimum solution, but a plan including a large error is unlikely to cause practical problems. However, the above-mentioned computer-based solution method may generate a solution with high mathematical optimality in some cases, but it may generate a solution with a large error, which is a problem for practical use together with execution speed. .

Specifically, the calculation time is short if only the NN method or the NI method is used, but the accuracy of the solution is poor, which is a practical problem. The 2opt method is also fast, but cannot obtain sufficient accuracy. The exact solution method using the DP method cannot be applied to problems on a practical scale because the calculation takes a long time. The exact solution method using the polyhedron approach can be applied to large-scale traveling salesman problems with many bases, but it takes time to execute and cannot be applied to situations requiring interactive processing performance. The Simulated Annealing method can theoretically find a solution with high optimality, but for that purpose, it still takes a lot of time for calculation.

By applying GA to the traveling salesman problem, a solution for optimizing an efficiently defined objective function is obtained from a huge combination of traveling orders (Japanese Patent Laid-Open No. 7-105180). ).

However, in order to obtain a highly optimal solution in GA, it is necessary for the initial population to include sufficiently diverse solutions. If an initial population with a biased tendency is used, a local optimum solution is encountered and sufficient accuracy cannot be obtained.

On the other hand, when the initial population is created by a completely random method, there arises a problem that the convergence of the solution due to a genetic operation such as crossover or mutation becomes very slow thereafter. Furthermore, the crossover and mutation used in GA could not be realized efficiently if they were constructed only by completely random operations.

As described above, even when a physical distribution schedule is created using GA, it remains an issue to guarantee practical speed performance and optimality of the created plan at the same time.

Therefore, in order to solve the above-mentioned problems, the present invention supports various evaluation criteria for creating a distribution plan for a plurality of trucks, and provides a solution with the same accuracy as that prepared by a human expert. It is an object of the present invention to provide a transportation plan creation method and system that can be obtained in a time that can be used interactively.

[0013]

In order to achieve the above object, the invention according to claim 1 is a transportation plan creating method for creating a transportation plan by using a system having at least an input / output device, a processing device and a storage device. In addition, a base information registration step of inputting base information, which is position information of a base to which the package is delivered, and package information including a package amount of each package and a delivery site to which the package should be delivered. The package information registration step, the transportation flight information registration step of inputting transportation flight information including the limit of the load capacity of the vehicle to be a transportation flight, and the objective information indicating the evaluation criteria for the transportation plan,
A target information registration step to be entered together with the priority order, and a plan is evaluated according to the target information entered in the purpose information registration step to search for an optimal plan, and for a plurality of transportation flights based on other input information. It is characterized by including a plan creating step for creating a travel plan at one time and a plan outputting step for outputting the created travel plan.

According to a second aspect of the present invention, in the transportation plan creating method according to the first aspect, the plan creating step comprises:
A transportation plan is created using a genetic algorithm that includes an initial solution group generation step of creating an initial solution group composed of a plurality of initial solutions, and a solution group improvement step of improving the created initial solution group by genetic processing. It is characterized by being a thing.

According to a third aspect of the present invention, there is provided a transportation plan creating method for creating a transportation plan by using a system having at least an input / output device, a processing device, and a storage device, the position of a base being a delivery destination of a package. A base information registration step of inputting base information which is information, a baggage information registration step of inputting baggage information including a baggage amount of each baggage and a delivery destination base to which the baggage is to be delivered, and loading of vehicles to be transportation flights. A transportation flight information registration step for inputting transportation flight information including a limit of quantity,
Using a purpose information registration step of inputting purpose information indicating what kind of evaluation criteria a transportation plan is to be evaluated along with its priority, and the base information, the package information, the transportation flight information, and the purpose information. , A step of generating an initial solution group of travel plans for a plurality of transportation flights in an expression form in which a travel plan represented by one solution includes a travel plan for a plurality of transportation flights, and expressed in the expression form. A solution group improvement step of improving the target information to a more suitable solution group by performing genetic processing on each solution of the initial solution group, which is a solution group, and running the solution of the created solution group And a plan output step of outputting a plan.

The invention according to claim 4 is the invention according to claim 2 or 3.
In the transportation plan creating method described in 1,
One solution is to arrange the parcel indexes identifying the parcels to be delivered by the first transportation flight in the order of their delivery, then the parcel indexes identifying the parcels to be delivered by the second transportation flight in the order of their delivery, and then the third. The parcel index parts that identify the parcels to be delivered by the transport flight are arranged in the order of delivery, and the parcel index parts that are the parcel indexes of the parcels to be delivered by multiple transport flights are arranged in the order of delivery, and the identifier that identifies the transport flight. An identifier for identifying the type of the transportation flight, a start position of a package to be delivered by the transportation flight in the package index portion,
And the end position of the package to be delivered by the transport flight in the package index site, the data being expressed in an expression format including a transport flight index site arranged for each of the plurality of transport flights.

According to a fifth aspect of the present invention, in the transportation plan creating method according to any one of the second to fourth aspects, the initial solution group generating step of the transportation plan creating step is based on the amount of cargo to be transported and conditions. Based on this, the number and type of transport flights to be used are determined, and one randomly selected delivery destination is assigned to each of the determined number of transport flights. For other packages, the NI (Nearest Insertion) method is used. It is characterized in that one initial solution is generated by inserting it as a delivery destination of any of the transportation flights, and this is repeated a predetermined number of times to generate an initial solution group including a plurality of initial solutions. .

According to a sixth aspect of the present invention, in the transportation planning method according to any one of the second to fourth aspects, the genetic process is a crossover process in which the solution of the solution group is a gene. And in the intersection processing, 2 is randomly generated.
Select one gene, determine its crossing position, and the delivery route recorded before the crossing position of one gene,
It is characterized in that the delivery destination not included in the delivery route is inserted into the delivery route by the NI method according to the order in the other gene.

According to a seventh aspect of the present invention, in the transportation planning method according to any one of the second to fourth aspects, the genetic process is a mutation process in which the solution of the solution group is a gene. In the mutation treatment, one delivery destination recorded in a randomly selected gene is selected, and one randomly determined distance is selected.
The selected delivery destination and the delivery destination within the distance from the delivery destination are deleted from the selected gene and inserted again by the NI method.

The invention according to claim 8 is the method for preparing a transportation plan according to any one of claims 5 to 7, wherein the NI
When inserting a delivery destination of one package instead of inserting the delivery destination by the law, the route length between the delivery destinations before and after the insertion position before the insertion and the route length between the delivery destinations after the insertion If the increment value of the route length due to the insertion is more than a certain ratio than the route length between the delivery destinations before the insertion,
The insertion of the delivery destination of the next package is performed in advance without performing the insertion, and only the delivery destinations that have not been inserted are reordered randomly and the increment value of the route length is allowed from the original route length. It is characterized by using the multi-step NI method in which the NI method is inserted stepwise by using a larger value as the ratio and thus all the delivery destinations are inserted.

According to a ninth aspect of the present invention, there is provided a transportation plan creation system having at least an input / output device, a processing device, and a storage device, and a base for inputting base information which is position information of a base to which a package is delivered. Information registration means, baggage information registration means for inputting baggage information including a baggage amount of each baggage and a delivery destination base to which the baggage is to be delivered, and transportation flight information including a limit of a loading capacity of a vehicle to be transported. Transportation information registration means for inputting, objective information registration means for inputting the objective information indicating what kind of evaluation criteria the transportation plan should be evaluated with, and the objective information input by the objective information registration means According to the above, the plan is evaluated and the optimum plan is searched, and the plan creation means for creating the travel plans for multiple transportation flights at once based on other input information and the created travel plan are output. Characterized by comprising an image output unit.

According to a tenth aspect of the present invention, in the transportation plan creating system according to the ninth aspect, the plan creating means creates an initial solution group creating means for creating an initial solution group consisting of a plurality of initial solutions. It is characterized in that a transportation plan is created by using a genetic algorithm using a solution group improving means for improving the initial solution group by genetic processing.

According to an eleventh aspect of the present invention, there is provided a transportation plan creation system having at least an input / output device, a processing device, and a storage device, and a base for inputting base information which is position information of a base to which a package is delivered. Information registration means, baggage information registration means for inputting baggage information including a baggage amount of each baggage and a delivery destination base to which the baggage is to be delivered, and transportation flight information including a limit of a loading capacity of a vehicle to be transported. Transportation information registration means for inputting, purpose information registration means for inputting purpose information indicating what kind of evaluation criteria the transportation plan is evaluated along with its priority, the base information, the package information, the transportation flight Using the information and the target information, the initial solution group of the travel plans of a plurality of transportation flights is expressed in a form such that the travel plan represented by one solution includes the travel plans of a plurality of transportation flights.
A means for generating and a solution group for improving the target information into a group of solutions more suitable for the target information by applying genetic processing to each solution of the initial solution group which is a group of solutions expressed in the expression format. It is characterized in that it comprises an improvement means and a plan output means for outputting a travel plan of the solution of the created solution group.

The invention according to claim 12 is the transportation plan creation system according to claim 10 or 11, wherein one solution of the travel plan is a luggage index for identifying a luggage to be delivered by a first transportation flight. Arrange in the order of delivery, then the second
The parcel indexes that specify the parcels to be delivered by the transportation route are arranged in the order of delivery, and then the parcel indexes that identify the parcels to be delivered by the third transportation route are arranged in the order of delivery, and so on. A package index part in which the package indexes of the package are arranged in the order of delivery, an identifier for identifying a transportation flight, an identifier for identifying the type of the transportation flight,
A start position of the package to be delivered by the transport flight in the package index part and an end position of the package to be delivered by the transport flight in the package index part are transport flight index parts arranged for each of the plurality of transport flights. It is characterized in that the data is expressed in an expression format including.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

In the present embodiment, it is assumed that a distribution center is used as a single starting point and a traveling plan for a plurality of trucks for transporting packages to a plurality of delivery destination bases is created.

FIG. 2 is an overall configuration diagram showing the system of this embodiment. This system consists of an input device (201), an output device (202) such as a printer, and a display device such as a display.
(203), a processing device (204), and a storage device (209). The processing device (204) includes an input processing unit (206) and a route generation unit.
(207), and a series of programs including the result output unit (208)
Execute (205). In addition, the storage device (209) stores the site information.
(210), baggage information (211), transportation flight information (212), distance table (213), purpose information (214), road map (215), and transportation route information (216) created from these are stored. To be done. Further, as the base information (210), data indicating the position of the distribution center or the delivery destination base of the package is mainly stored.

FIG. 3 shows an example of the base information (210) in FIG. The location information stores the location information of the distribution center or the delivery destination location of the package. Specifically, the location ID (301) that uniquely identifies the distribution center or the delivery destination location, the location name (302), and the latitude (303) and longitude (3
04) etc. are stored.

FIG. 4 shows the luggage information table (211) in FIG.
Shows an example. In the package information, information on each package to be transported is stored. Specifically, a package index (401) that is sequentially assigned to each package from 1, a package ID (402) that uniquely identifies the package, a delivery destination base ID (403) that indicates the delivery destination of the package, and a package volume ( 404).

FIG. 5 shows an example of the transport flight information (212) shown in FIG. The transportation information stores the transportation capacity and the limit of the truck to be used. In particular,
Track type ID (50
1), the load capacity limit value (502), the load capacity limit value (503) of that type of truck, and the departure time (504) of the truck from the distribution center and the finish time (50
It consists of 5).

FIG. 6 shows an example of the distance table (213) in FIG. The distance table (213) stores information required for route generation, which is related to the distance traveled by a transportation flight such as a truck between bases. Specifically, the ID of the departure location
(601) and arrival base ID (602), and the distance between the bases
(603) and the required (traveling) time (604) between bases.

The road map (215) of FIG. 2 shows road information of the transportation area used for calculation of the distance between bases necessary for creating the distance table (213) and output of the result.

FIG. 7 shows an example of the gene expression used in the calculation of the transportation route by the genetic algorithm used in the route generation unit (207) in FIG. The gene has a transportation index part and a luggage index part. In the transport flight index part, the transport flight index assigned to each transport flight for each truck used (701)
And the truck type ID (702) of the transportation flight and the start position of the luggage index part that represents the luggage delivered by the truck
(703) and end position (704) are stored. The transport flight index is sequentially assigned to each transport flight. 701 to 704 are data relating to the first transport flight, data relating to the second transport flight, data relating to the third transport flight, data relating to each transport flight, and so on.

The luggage index portion stores the luggage index (705) of the luggage delivered by each transportation flight. For example, the transport flight specified by the transport flight index 0 is a truck of the type with the truck type ID 0,
Depart from the logistics center, deliver the parcels of the parcel index from the position designated by the parcel index start position 0 to the position designated by the end position 0 in the parcel index part in that order, and return to the parcel center. It means to come. Therefore, this gene expression represents a transportation route for a plurality of transportation flights. By treating all luggage as a single array at the luggage index site,
Delivery routes can be created without considering the number of trucks used.

FIG. 8 shows an example of the transportation route information (216) shown in FIG. The transportation route information (216) stores information indicating a location where each truck travels in a transportation plan created by the method according to the present invention and the order thereof. Specifically, the transport flight index (801), the time when the truck of the transport flight index leaves the distribution center (802), the site ID of the delivery destination site (803), and the arrival time at that site.
It stores (804). The base ID and the arrival time are the base ID 1 (803) of the first delivery destination base and the arrival time 1 (804) at that base, and the base ID 2 (805) and the arrival time 2 (806) of the second delivery destination base. , ...
In this way, the data regarding the delivery destination bases are stored in the order of delivery. In addition, the total luggage amount (8
07) and total mileage (808) are stored. The above data is stored for each transport flight, that is, for each transport flight index.

FIG. 9 shows an example of the structure of the objective information (214) shown in FIG. The purpose information (214) is the transportation route generation unit (20
In 7), the information that determines how to handle each element such as transportation time, margin time, loading ratio, etc. when evaluating the optimality of the transportation route is shown. The target information consists of element names that can be calculated from the transportation route information and their priority or weight, and the user sets these. The transportation route generation unit (207) evaluates each transportation route according to this information and uses it as a guideline for creating an optimal transportation route.

FIG. 10 is a flow chart showing an outline of the program (205) executed by the processing device (204) in FIG. First, in the input processing step (1001), the site information (210),
Package information (211), transportation information (212), and purpose information (21
4) The user inputs the information required to create a transportation plan such as. Details of this processing will be described later with reference to FIG. The next route generation step (1002) is the input processing step (1001)
Determine a transportation plan that meets the conditions given in. Here, the decision of the transportation plan is the schedule (transportation route information) that the truck that delivers all the input packages to the delivery destination follows.
Refers to determining. Details of this processing will be described later with reference to FIG. Next, in the result output processing step (1003), the created route is output. Details of this processing will be described later with reference to FIG.

FIG. 11 shows the input processing step (1
It is a flowchart showing the details of (001). First, in the base information registration step (1101), the base information (210) shown in FIG.
To register. Next, in the luggage information registration step (1102),
The package information (211) shown in FIG. 4 is registered. Next, in the truck information registration step (1103), the transportation information shown in FIG.
Register (212). Next, the purpose information registration step (110
In 4), the purpose information (214) shown in FIG. 9 is registered. Next, in a distance table generation processing step (1105), the distance between bases and the required travel time by truck are calculated with reference to the road map (215), and the distance information (213) shown in FIG. 6 is generated.
It is assumed that the road map (215) is prepared in advance.

FIG. 12 is a flow chart showing the details of the route generation processing step (1002) in FIG. GA is used for the route generation process. First, an initial solution group including a plurality of initial solutions is created (step 1201). Details of this processing will be described later with reference to FIGS. 16 and 17. Next, the solution group is improved by GA (step 1202). Details of this processing will be described later with reference to FIG. At least a part of the solution group is changed by the improvement processing of the solution group, and step 1202
By repeating the processing of ~ 1205, the solution group approaches the optimal solution. After the processing of step 1202, the solution group is evaluated.
(Step 1203). The evaluation is performed by weighting and prioritizing elements such as the total number of transportation flights and transportation distance calculated from the transportation route information (216) according to the purpose information (214), and using the results, Rank each solution. Details of this processing will be described later with reference to FIG.

Next, it is checked whether the conditions for ending the route generation processing are satisfied (step 1204). This end condition consists of conditions such as the restriction on the calculation time, the restriction on the number of GA calculation generations, and the achievement of the evaluation value given as a goal. If the end condition is not satisfied in this step 1204, the solution group to be subjected to the next genetic processing is determined from the solution groups according to the evaluation rank performed in step 1203 (step 1205), and the process returns to step 1202. When the solution group is increased by the process of step 1202, the size of the solution group is kept constant by selecting an individual having a good evaluation value in the process of step 1205. If the termination conditions are met in step 1204,
At that time, the solution with the highest evaluation is output as the best solution in the solution group (step 1206), and the route generation process is terminated.

FIG. 13 shows the initial solution group generation process in FIG.
9 is a flowchart showing the details of the processing of the NI method (nearest neighbor insertion method) used for (step 1201) and improvement of the solution population by genetic processing (step 1202). Specifically, the NI method is used in step 1405 of FIG. 14, step 1607 of FIG. 16, step 1908 of FIG. 19, step 2014 of FIG.

In this process, the luggage index portion of the gene information is copied in advance to each element of the array R. Also,
Set the number of baggage indexes stored in array R to variable LEN, and then set the baggage index of the newly added baggage to a variable.
It shall be set to D in advance.

First, the luggage index is D in the variable Lmin.
The total distance when the delivery destination of the package is added to the end of the route represented by R [0] to R [LEN] is substituted (step 1301).
At this time, regarding the transport flight index portion, it is assumed that the corresponding luggage index end position of the last transport flight in this portion is treated as +1 and the traveling distance is obtained. After that, when the baggage of the baggage index D is inserted at the Ith position of the baggage index portion, it is treated as being added to the transportation flight carrying the baggage at the I-1th position. However, at the beginning (1
No.) When inserting into R [0], it is treated as an addition to the transportation flight carrying the package in R [0] before insertion.

Next, LEN + 1 is assigned to the variable P (step 1
302). Next, 1 is assigned to the variable I (step 1303). Next, the processes of steps 1305 to 1309 are repeated until I exceeds (LEN + 1) (step 1304).

In the iterative processing of steps 1305 to 1309, the traveling distance when the luggage specified by the index D is inserted into the I-th route of the array R is obtained, and the traveling distance becomes the minimum among them. Ask for I and insert the delivery address of the package at that position. Specifically, first, the total traveled distance of the route when the baggage designated by the index D is inserted into the I-th route of the array R is substituted into the variable L (step 1305). Next, the value of the variable Lmin and the value of the variable L are compared (step 1306). If L is smaller than Lmin, the value of I is assigned to the variable P (step 1307). Next, the value of the variable L is substituted for Lmin (step 1308), and the process proceeds to step 1309. If L is not smaller than Lmin, the process proceeds to step 1309. Next, 1 is added to the variable I (step 1309) and the processing of steps 1305 to 1309 is repeated for the next insertion position I.

At the end of this process, the variable P represents the best position for inserting the parcel of index D. Therefore, next, the delivery destination of the package indicated by the index D is added to the insertion position indicated by the variable P, and along with that, the gene transport flight index also carries the package at the position of P-1 for the package of the index D. Modify to be added to the transport (step 1
310).

FIG. 14 is a processing flow of the multi-step NI method. The multi-step NI method is shown in step 17 of FIG.
04, step 1908 of FIG. 19, and step 20 of FIG.
Use in 14.

In FIG. 14, in the array M, the increment ratio of the distance allowed at the time of insertion by the NI method at one site is set in advance in each stage of the multi-step NI method. This value is M
[1] <M [2] <M [3] <... These values M [1], M [2], M [3], ... Correspond to k described in FIG. The number of increment ratios set in the array M is set in the variable Max. Further, the number of delivery destinations registered in the insertion destination route (that is, the number of delivery destinations (packages) in the transportation route represented by the gene expression in FIG. 7) is set in the variable LEN.

First, 1 is assigned to the variable I (step 140)
1). Next, I and Max are compared (step 1402). If I is small, insertion processing is performed at the increment ratio of M [I] (step 1403). Details of this insertion processing are shown in FIG. Next, 1 is added to the variable I (step 1404), the process returns to step 1402 and the process is repeated. When I reaches Max in step 1402, the luggage of the luggage index that is not inserted in the route is inserted into the route by the process of the NI method (step 14
05).

FIG. 15 shows the insertion process of FIG. 14 (step 14
It is a flow chart showing details of (03).

First, the delivery destination (luggage) not yet inserted in the route
Then, it is checked whether there is any baggage that has not been tried to be inserted into the route at the increment ratio M [I] (step 1501).
If such packages are left, one package (delivery destination) is randomly selected from the packages.
The parcel index on the parcel information table for this parcel is set to the variable J (step 1502). Next, 1 is assigned to the variable K indicating the insertion candidate position of the delivery destination and 0 is assigned to P (step 150
3). In the subsequent steps, the package to be inserted (delivery destination)
The Kth insertion is considered, but the position of the insertion candidate is represented by the variable K. The variable P is a variable that stores the position evaluated to be the best for insertion among the 1st, 2nd, ..., K-1th examined so far. Next, K is compared with LEN + 1 (step 1504). The value of K is LEN + 1
If it does not exceed, the possibility of insertion at that position is checked.

The checking of the insertability is the following processing. First, in the variable L1, route (luggage index)
The distance between the delivery destination of the (K-1) th package and the delivery destination of the Kth package is substituted (step 1505). If the K-1th position in the baggage index part of the gene is the end position of the baggage index corresponding to one of the freight flights in the freighter index part, it will be added to the end of the route of the freighter. Therefore, the distance is calculated using the distribution center as the Kth delivery destination in the route. That is, in this case, "K" in this step and the following steps.
The "th delivery destination" is not the delivery destination of the Kth luggage of the luggage index portion (actually, the Kth luggage portion of the luggage index portion is the first luggage of the next transportation flight), but the distribution center. To do.

Next, in the variable L2, the distance between the (K-1) th delivery destination and the delivery destination of the package J to be added, and the distance between the delivery destination to be added and the Kth delivery destination. Substitute the sum of (step 1506). Next, L1 * M [I] is compared with (L2-L1) (step 1507). If (L2-L1) is large, the delivery destination is
No insertion is made between the Kth and K + 1th roots, and the process proceeds to step 1514. Then add 1 to K (step 1514),
Return to step 1404 to move on to the possibility of inserting the next position in the route.

If (L2-L1) is equal to or less than L1 * M [I] in the comparison in step 1507, whether or not the increment of the route length due to the insertion at that point is smaller than the insertion position evaluated so far. To determine. First, the variable P is compared with 0 (step 150
8). If the value of the variable P is 0, this means that this is the first process, so the value of K is assigned to the variable P (step
1509) and substitute the value of L2-L1 into the variable L0 (step 151
0), and proceeds to step 1514. If the value of the variable P is not 0 in step 1508, the values of L0 and L2-L1 are compared (step 1511). Of the insertion positions evaluated up to this point, L0
The value of L2-L1 is stored although L2-L1 is the smallest. If L2-L1 is smaller than L0, it means that the distance L2-L1 is smaller at this insertion position.
(Step 1512) and the value of L2-L1 to L0
(Step 1513) and the process proceeds to step 1514. In step 1511
If L0> L2-L1 is not satisfied, the process proceeds to step 1514.

If K is larger than LEN + 1 in the determination processing of step 1504, all positions in the route have been checked, so if a position that can be inserted is found, it is inserted into the route. First, it is determined whether or not the insertable position is recorded in the variable P, that is, whether or not the value of the variable P is 0.
(Step 1515). If the value of P is 0, it means that there is no insertion position that satisfies the condition, so the process proceeds to the next delivery destination. If the value of P is not 0, the insertion at the position P is performed (step 1516). After step 1516, the process returns to step 1501 to move to the processing of the next delivery destination. After processing all the shipping destinations (packages) to be inserted in step 1501,
finish.

Next, an example in which the random NI method is used in the initial solution group generation processing (step 1201) in FIG. 12 will be described.

In order to obtain a highly optimal solution by GA, the initial population must include sufficiently random individuals. If it consists only of solutions that have a certain tendency,
Even if the initial solution (initial planning group) has a high optimality, the possibility of falling into a local optimal solution in the solution search process increases, and the optimality required for practical use cannot be realized.

Further, if the initial solution is constructed completely randomly, a large amount of so-called lethal genes, which cannot satisfy the given constraint conditions, occur and the search efficiency decreases. Even if the initial solution satisfies the constraint conditions, starting from only a solution group with low optimality would result in poor efficiency of solution group improvement due to genetic operations such as GA crossovers and mutations. It becomes impossible to achieve both the response speed required for a conventional system and a solution that secures practical accuracy.

Therefore, by obtaining solutions with high randomness and optimality and constructing an initial design ensemble from those solutions, the convergence speed of the solution by GA is improved and the lower limit of the accuracy of the solution is guaranteed. I am trying to give. That is, in the present embodiment, in order to secure the randomness of the initial solution group, first, the delivery destination bases are randomly assigned and assigned to each transportation flight, and the constitutive traveling salesman problem is solved according to the random order. Create a delivery route using the NI method, which is the solution. By using the NI method, the lower bound of accuracy is guaranteed and an initial population with randomness is created. The initial solution group generation process using such a random NI method will be described below.

FIG. 16 shows the initial solution group generation process in FIG.
6 is a flowchart showing details when the random NI method is used in (step 1201). It should be noted that the traveling schedule (transportation route) of a plurality of trucks is registered in one solution (transport flight index portion and luggage index portion in FIG. 7), and the solution group further includes a plurality of solutions.

When this processing is executed, the variable SIZE is set in advance with the number of individuals of genes to be registered in the initial population.
In addition, the number of packages (delivery destination bases) registered in the route is set in the variable LEN.

First, 0 is substituted for the variable I (step 160
1). Then while I is less than SIZE, steps 1602 through 161
The process of 0 is repeated to generate the number of genes specified by SIZE.

First, in order to generate the format of the I-th gene, the number and type of transportation flights to be used are determined from the amount of cargo to be transported and conditions (total weight, total capacity, operating time, etc.), and only the determined number is determined. A plurality of transportation flight indexes corresponding to one randomly selected baggage index for each transportation flight are created. Then, a luggage index part including only the luggage selected there is created (step 160).
3). The corresponding transport index site is also created.

Next, 1 is assigned to the variable J (step 160
Four). J is a variable indicating how many packages are being processed.
Next, the values of J and LEN are compared (step 1605). If J
If is less than or equal to LEN, the processes of steps 1606 to 1608 are repeated. That is, one randomly selected luggage index that is not registered in the I-th gene from the luggage information (211) and substitutes the value of the luggage index into the variable d.
(Step 1606). Next, the parcel whose luggage index is d is NI
Method is added to the track route expressed by the I-th gene (step 1607). Then add 1 to the variable J and step
Return to 1605 and repeat the process.

If J exceeds LEN in step 1605, it means that all packages to be delivered are registered in the I-th gene, and the production of one gene is completed. In that case, it is checked whether the transportation route information represented by the gene violates the restrictions (conditions such as loading weight limit, loading capacity limit, working hours, etc.) of the transportation flight (step 1609), and it is broken. If so, increase the number of tracks to be used, return to step 1603, and repeat the process. If the condition is not violated in step 1609, add 1 to the variable I (step 161
0), the process returns to step 1602, and genes are sequentially generated.

Next, a method of generating an initial ensemble using the multi-step NI method, which is another example of the initial solution ensemble generation processing by the random NI method of FIG. 16, will be described.

A transportation route created by using a mechanical method using a computer (a schedule to be followed by each transportation flight in the plan) has a large mathematical optimality, but is not as large as a route created by a human expert. The shape was different, which was a problem for practical use. For this reason, there is a demand for a method of creating a route close to the route shape created by a human together with the optimality. Further, when a route is created by the above-mentioned random NI method, the distance between bases is not taken into consideration, so that a route with many zigzags as shown in FIG. 1A is generated. When a person creates a route, the route is created by giving priority to delivery destination bases close to the created route to some extent so that the route will be smooth, and therefore smooth as shown in FIG. 1B with less zigzag. A new route is created.

Therefore, in the present embodiment, in order to suppress the occurrence of such a zigzag, the initial group of FIG.
Random NI solution with less zigzag as shown in (b)
Try to mix with the initial solution by law. Therefore, in the initial solution generation method by the NI method, when a new delivery destination C is added between two bases AB as shown in Fig. 1 (c), the distance between AB and the distance when C is inserted increase. The ratio of the minutes AC + BC-AB is calculated, and if this ratio is above a certain level, the site C is not inserted and the next delivery destination is inserted. The delivery destination, which cannot be inserted in the random order set at the beginning, relaxes the allowable range of the increase rate of the distance by the insertion and tries the insertion again. Even with this re-insertion, if there remains a delivery destination that could not be inserted, the allowable range is relaxed again. In this way, by gradually relaxing the allowable range of the distance increase ratio, it is possible to create a route with less zigzag and high optimality. The delivery destination that has not been inserted will try the insertion again by the NI method after the insertion in the random order defined at the beginning is completed. By using such an initial solution generation method, routes with less zigzag can be added to the initial population.

The specific procedure of the above-mentioned method is the multi-step NI method described in FIG. In the following, as an example of the initial population generation process different from that in FIG.
A specific example of the initial group generation process using the method will be described.

FIG. 17 is a flow chart showing a method for generating an initial population using the multi-step NI method. This is another example (from FIG. 16) of the initial population generation process (step 1201) in FIG. 12, and is a detailed flow when the multi-step NI method is used. When this process is executed, the variable SIZE is set in advance with the number of individuals of genes to be registered in the initial population.

First, 1 is assigned to the variable I (step 170
1). Next, it is determined whether I is smaller than SIZE (step 170
2). If I is less than or equal to SIZE, a transportation flight index part and a luggage index part are created in the same manner as described in step 1603 of FIG. 16 (step 1703). Next, the delivery destinations registered in the package information table are shown in FIG. 14 and FIG.
Create a route by inserting it into an empty route using the multi-step NI method described in 5. Next, it is checked whether the transportation route information represented by the gene violates the restrictions of the transportation flight (loading weight limit, loading capacity limit, working hours, etc.) (step 1704), and it is violated. In this case, the number of tracks to be used is increased and the process is repeated from step 1703. If the condition is not violated in step 1704, 1 is added to the variable I, and the process returns to step 1705 to proceed to the generation of the next gene.

FIG. 18 is a flow chart showing the details of the improvement of the solution group (step 1202) by the genetic processing in FIG. Genetic processing changes the solution population and makes improvements
This is the central processing of GA. First, one solution (gene) is selected from the solution group (step 1801), and it is checked whether or not cross processing is performed (step 1802). The intersection process is a process of creating a new solution that inherits the properties of the two different solutions. Whether or not the intersection process is performed is randomly determined with a predetermined probability. When performing intersection processing,
Choose another different solution from the solution set (step 1805),
A new solution is created based on the two selected solutions (step 1806). Details of this processing will be described later with reference to FIG.

Next, it is checked whether or not the mutation process is performed (step 1803). The mutation process is a process of partially changing one solution to create a new solution. Whether or not the mutation process is performed is also randomly determined with a predetermined probability. When performing mutation processing, a new solution is created by changing a part of the solution (step 180).
7). Details of this processing will be described later with reference to FIG.

Next, it is checked whether all the solutions have been checked in steps 1802 and 1803 (step 1804), and if there are unchecked solutions, the process returns to step 1801. If all the solutions have been checked, the process ends.

Next, the gene crossover processing will be described. In GA, it is necessary to use genetic manipulations such as crossover and mutation that match the characteristics of the problem in order to efficiently and accurately obtain a solution. Using only simple two-point crossing or mutation by exchanging random values is no different from random search. Apply to Traveling Salesman Problem
In GA, when a child gene is created at the intersection, it is necessary to inherit the good characteristics of two parent genes to the child gene.
In the traveling salesman problem, inheriting a traveling order to a child allows good inheritance of features. The following intersection method (hereinafter referred to as NI intersection method) is used in order to perform high-speed intersection processing that inherits this cyclic order and obtain a good solution in a short time.

First, two genes are taken out randomly. Next, another intersection of two genes is randomly determined. In the route consisting of genes in the first half of the intersection of one gene, take out the delivery destination that does not appear in the route from the beginning of the other gene and use it as the route to NI
Insert by method. In this way, a new route is created from the routes expressed by the two genes. Note that the accuracy of the solution can be further improved by using the above-mentioned multi-step NI method.

A specific example of such intersection processing will be described below.

FIG. 19 is a flow chart showing the details of the intersection processing of step 1806 of FIG. 18, that is, the NI intersection method. The variable N is set in advance with the gene size, that is, the number of packages registered in the package information table.

First, two genes randomly selected from the solution population are copied to the structures G1 and G2 (step 190
1). After that, the copied gene G2's luggage index site
The I-th baggage index is represented as G2 [I]. Next, an integer of 1 or more and N or less is randomly selected and assigned to the variable P indicating the intersection position (step 1902). Next, the Pth and subsequent elements of the array G1 are deleted, and the transport flight index portion is corrected accordingly (step 1903). If there is a transport flight that has no luggage here, randomly select one from the packages not included in G1 and assign it to the empty transport flight
(Step 1904). Next, 1 is assigned to the variable I (step 190
Five).

Next, while I is equal to or less than N, steps 1907 to 19 are executed.
Repeat the process of 09. Specifically, it is determined whether the load of G2 [I] is included in the elements in the load index part of G1 (step 1907). If not included,
The delivery destination of the package of the package index represented by G2 [I] is added to G1 (step 1908). This process is performed by the NI shown in FIG.
The method or the multi-step NI method shown in FIGS. 13 and 14 is used. Next, 1 is added to the variable I (step 1909), the process returns to step 1906 to repeat the process.

When I reaches N in step 1906, it means that all the packages registered in the package information table have been registered and new genes have been generated. It is checked whether or not the route information violates the restrictions of the transportation service (load weight limit, load capacity limit, working hours, etc.) (step 1910). If they are protected, register the new gene (step 191
1), the processing ends. If not protected, the process ends without registering the gene generated by this process.

Next, the mutation process will be described. In the mutation process as well, similar to the above-described crossover process, it is necessary to use a process that matches the characteristics of the problem. In order to prevent falling into a local solution, it is necessary not only to improve the solution but also to significantly change the solution. For this purpose, the following sudden displacement processing method is used.

First, one planned site is randomly selected. Next, the distance is randomly determined, and one planned base determined earlier and the planned base whose distance from that point is closer than the randomly determined distance are deleted from the plan. Next, the site removed from the plan is reinserted by the NI method. Note that the accuracy of the solution can be further improved by using the above-mentioned multi-step NI method.

A specific example of such a sudden displacement process will be described below.

FIG. 20 is a flow chart showing details of the mutation process in step 1807 of FIG. Variable in advance
Set the gene size in N.

First, a gene randomly selected as a mutation target is copied to sequence G (step 2001). Hereinafter, the I-th load index of the copied load index site of gene G will be referred to as G [I]. Next, the distance between the distribution center and the delivery destination base farthest from the distribution center is substituted into the variable L (step 2002). Next, a real number of 0 or more and 0.3 or less is randomly selected and set as the variable R (step 1903). Next, the product of L and R is substituted into the variable L1 (step 2004). Next, a positive integer of 1 or more and N or less is randomly selected and assigned to the variable P (step 2005). Then the variable I
Substitute 1 for (step 2006). Substitute 0 for the variable J (step 2007).

If the value of I is N or less, from step 2009
The process of 2012 is repeated. Specifically, the distance between the delivery destination base corresponding to G [P] and the delivery destination base corresponding to G [I] is obtained by referring to the distance table, and it is determined whether the value is L or less.
(Step 2009). If the distance between the delivery destination bases is less than or equal to L, the value of G [I] is substituted for H [J] and -1 is substituted for G [I] (step 2010). Then add 1 to J (step 2011),
Add 1 to I (step 2012) and return to step 2008.

Next, in the luggage index part of G, the value is-
The element of 1 is deleted from the package index part, and the transport index is modified accordingly. At this time, if there is a flight that does not carry any luggage in the transport flight index, it is deleted (step 2013). Next, the package of the package index stored in the array H is added to G (step 201).
Four). For this additional processing, the NI method shown in FIG. 13 or the multi-step NI method shown in FIGS. 14 and 15 is used. After that, the transportation route information represented by the newly created gene,
Check to see if you have violated transportation restrictions (load weight limit, load capacity limit, working hours, etc.) (step 2015). If they are protected, a new gene is registered (step 2016) and the process ends. If not protected, the process ends without registering the gene generated by this process.

FIG. 21 is a flow chart showing details of the result output process of step 1003 of FIG. First, the transit points of each transportation flight are displayed on the map (step 2101). Next, the route obtained by the route generation process (step 1002) and its evaluation value can be compared with the purpose or target,
Display as a table or graph (step 2102). Next, if there is a print instruction, the graph or table is printed (step 210
3).

FIG. 22 is a flow chart showing details of the solution group evaluation process in step 1203 of FIG.

In this process, first, the total traveling distance is calculated from the transportation route information (216) indicated by the gene to be evaluated (step
2201), calculating the capacity margin of each truck (step 2202), calculating the deviation of the operating time of each truck (step 2204), and other necessary for the evaluation specified in the objective information (214) A feature amount is calculated (step 2203). afterwards,
The feature information calculated in steps 2201 to 2204 is multiplied by the weighting coefficient specified in the target information, all are added, and the result is used as the evaluation value of the gene (step 2205).

In the above-described embodiment, in the case of the transportation truck, the case has been described as an example in which the truck loads the parcel at the distribution center and creates the delivery route for circling the delivery destination base to deliver the parcel. However, the present invention is not limited to this. The same can be applied to the case where the parcel is delivered from each location to the distribution center.

[0093]

As described above, according to the present invention, for creating a logistics plan at a strategic level, various conditions (purpose information) are changed, and a highly accurate transportation plan can be executed at high speed. Can be created.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a delivery route and an example of inserting a base in a multi-step NI.

FIG. 2 is an overall configuration diagram showing a system of the present embodiment.

FIG. 3 is a diagram showing an example of a configuration of base information input by a user and used for creating a delivery route in the present embodiment.

FIG. 4 is a diagram showing an example of a configuration of package information input by a user and used for creating a delivery route in the present embodiment.

FIG. 5 is a diagram showing an example of a configuration of transport flight information input by a user and used for creating a delivery route in the present embodiment.

FIG. 6 is a diagram showing an example of a distance table between bases used for creating a delivery route in the present embodiment.

FIG. 7 is a diagram showing an example of gene data used for a genetic process used for creating a delivery route in the present embodiment.

FIG. 8 is a diagram showing an example of transport truck information created in the present embodiment.

FIG. 9 is a diagram showing an example of purpose information used for creating a delivery route in the present embodiment.

FIG. 10 is a flowchart showing the overall processing of this embodiment.

FIG. 11 is a flowchart showing details of input processing according to the present embodiment.

FIG. 12 is a flowchart showing a route generation process of this embodiment.

FIG. 13 is an example of an NI used during the route generation processing of this embodiment.
It is a flowchart figure which shows the detail of the process by the method.

FIG. 14 is a flowchart showing the processing by the multi-step NI method used during the route generation processing of this embodiment.

FIG. 15 is a flowchart showing details of insertion processing by the multi-step NI method used during the route generation processing of the present embodiment.

FIG. 16 is a flowchart showing the details of the initial population generation process of the genetic process using the random NI method used during the route generation process of this embodiment.

FIG. 17 is a flowchart showing the details of the initial population generation process of the genetic process using the multi-step NI method used during the route generation process of this embodiment.

FIG. 18 is a flowchart showing the genetic processing of this embodiment.

FIG. 19 is a flowchart showing details of the crossover process of the genetic process used in this embodiment.

FIG. 20 is a flowchart showing the details of mutation processing used during the genetic processing of this embodiment.

FIG. 21 is a flowchart showing details of route output processing of the present embodiment.

FIG. 22 is a flowchart showing details of solution evaluation processing used during the route generation processing of the present embodiment.

[Explanation of symbols]

201 ... Input device, 202 ... Output device, 203 ... Display device, 204 ... Processing device, 209 ... Storage device, 206 ... Input processing unit, 207 ... Route generation unit, 208 ... Result output unit, 205 ... Program, 210 ... Location Information, 211 ...
Package information, 212 ... Truck information, 213 ... Distance table, 214 ... Objective information, 215 ... Road map, 216 ... Transportation route information.

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G06N 3/00 550 G06N 3/00 550C (72) Inventor Takashi Onoyama 6-chome, Onoue-cho, Naka-ku, Yokohama-shi, Kanagawa Address 81 Hitachi Software Engineering Co., Ltd. Internal F-term (reference) 5H180 AA15 BB15 BB20 EE02 FF01 FF11 FF22 FF32

Claims (12)

[Claims] 1. A transportation plan creation method for creating a transportation plan by using a system having at least an input / output device, a processing device, and a storage device, which is provided with base information, which is position information of a base to which a package is delivered. Base information registration step to input, baggage information registration step to input baggage information including the baggage amount of each baggage and a delivery destination base to which the baggage should be delivered, and transportation including a limit of the load capacity of a vehicle to be a transportation flight The transportation flight information registration step for inputting flight information, the objective information registration step for inputting the objective information indicating the evaluation criteria for the transportation plan together with the priority order thereof, and the objective information registration step While evaluating the plan according to the objective information obtained and searching for the optimal plan,
A transportation plan creation method comprising: a plan creation step for creating travel plans for a plurality of transportation flights at once based on other input information; and a plan output step for outputting the created travel plans. 2. The transportation planning method according to claim 1, wherein the planning step includes an initial solution group generation step of creating an initial solution group composed of a plurality of initial solutions, and the created initial solution group genetically. A transportation plan creation method characterized in that a transportation plan is created using a genetic algorithm including a solution group improvement step of improving by processing. 3. A transportation plan creating method for creating a transportation plan by using a system having at least an input / output device, a processing device, and a storage device, the method comprising: Base information registration step to input, baggage information registration step to input baggage information including baggage amount of each baggage and delivery destination base to which the baggage should be delivered, transportation including limit of load capacity of vehicle to be transportation flight A transportation information registration step of inputting flight information, a purpose information registration step of inputting purpose information indicating what kind of evaluation criteria a transportation plan is to be evaluated together with its priority, the base information, the package information, A table in which a travel plan that expresses an initial solution group of travel plans of a plurality of transport flights using the transport flight information and the purpose information includes travel plans of a plurality of transport flights. In a format, and performing a genetic process on each solution of the initial solution group, which is a group of solutions expressed in the expression format, to improve the group of solutions more suitable for the target information. And a plan output step for outputting a travel plan of the solution of the created solution group. 4. The transportation plan creating method according to claim 2, wherein one solution of the travel plan is to arrange a baggage index specifying a baggage to be delivered by the first transportation flight in the delivery order, and A plurality of transportation flights are arranged such that the parcel indexes identifying the parcels to be delivered by the second transportation flight are arranged in the delivery order, the parcel indexes identifying the parcels to be delivered by the third transportation flight are arranged in the delivery order, and so on. The parcel index part in which the parcel indexes of parcels to be delivered in the order are arranged in the order of delivery, an identifier for identifying the transport flight, an identifier for identifying the type of the transport flight, and the start of the package to be delivered by the transport flight in the parcel index region. A table showing the position and the ending position of the package to be delivered by the transportation flight in the luggage index site in the form of an expression including the transportation flight index site arranged for each of the plurality of transportation flights. It has been transportation schedule creation method, which is a data. 5. The transportation plan creation method according to claim 2, wherein the initial solution group generation step of the transportation plan creation step uses the transportation based on the amount of cargo to be transported and conditions. The number and type of flights are determined, and a randomly selected destination for one package is assigned to each of the determined number of transport flights, and other packages are either transported using the NI (Nearest Insertion) method. A transportation plan creating method characterized in that one initial solution is generated by inserting it as a delivery destination of a certain number of flights, and this is repeated a predetermined number of times to generate an initial solution group including a plurality of initial solutions. 6. The transportation planning method according to any one of claims 2 to 4, wherein the genetic process is a cross process with a solution of the solution population as a gene. In (2), two genes are randomly selected, their crossover positions are determined, and the delivery route recorded before the crossover position of one gene and the delivery destination not included in the delivery route are A method for creating a transportation plan, characterized in that the NIs are inserted into the previous delivery route according to the order in the gene. 7. The transportation planning method according to any one of claims 2 to 4, wherein the genetic treatment is a mutation treatment in which a solution of the population is a gene. In the mutation process, one delivery destination recorded in a randomly selected gene is selected, and one randomly determined distance is selected, and the selected delivery destination and within the distance from the delivery destination are selected. A method for creating a transportation plan, which comprises deleting a certain destination from the selected gene and inserting it again by the NI method. 8. The transportation plan creating method according to claim 5, wherein instead of inserting the delivery destination by the NI method, when the delivery destination of one package is inserted, the insertion is performed. The route length between the delivery destinations before and after the previous insertion position is compared with the route length between the delivery destinations after the insertion, and the increment value of the route length due to the insertion is greater than the route length between the delivery destinations before the insertion. If the ratio is more than a certain ratio, the insertion will not be performed, the delivery destination of the next package will be inserted first, and only the delivery destinations that have not been inserted will be reordered randomly and the increment value of the route length Creating a transportation plan characterized by using the multi-step NI method, in which the NI method is inserted stepwise using a larger value as the allowable ratio from the original route length, and thus all delivery destinations are inserted Method. 9. A transportation plan creation system having at least an input / output device, a processing device, and a storage device, and site information registration means for inputting site information which is position information of a site to which a package is delivered, Baggage information registration means for inputting baggage information including the baggage amount of the baggage and the delivery destination base to which the baggage is to be delivered, and transportation flight information registration for entering the transportation flight information including the limit of the loading capacity of the vehicle to be a transportation flight A means and a purpose information registering means for inputting purpose information indicating what kind of evaluation criteria the transportation plan is to be evaluated together with its priority, and a plan is evaluated according to the purpose information input by the means for registering purpose information. It is equipped with a plan creation means that creates a travel plan for multiple transportation flights at once based on other input information while searching for an optimal plan, and a plan output means that outputs the created travel plan. Transportation planning system, characterized in that was. 10. The transportation plan creation system according to claim 9, wherein the plan creating means genetically generates the initial solution group creating means for creating an initial solution group consisting of a plurality of initial solutions. A transportation plan creating method characterized in that a transportation plan is created by using a genetic algorithm using a solution group improving means for improving by processing. 11. A transportation plan creation system having at least an input / output device, a processing device, and a storage device, and site information registration means for inputting site information that is position information of a site to which a package is delivered, Baggage information registration means for inputting baggage information including the baggage amount of the baggage and the delivery destination base to which the baggage is to be delivered, and transportation flight information registration for entering the transportation flight information including the limit of the loading capacity of the vehicle to be a transportation flight Means, purpose information registration means for inputting purpose information indicating with which evaluation criteria the transportation plan is evaluated together with its priority, the base information, the package information, the transportation flight information, and the purpose information And means for generating an initial solution group of travel plans for a plurality of transportation flights in an expression format such that a travel plan represented by one solution includes travel plans for a plurality of transportation flights, and Expression Solution for improving the target information by applying genetic processing to each solution of the initial solution group, which is a group of the generated solution, and the solution of the created solution group. And a plan output unit that outputs the travel plan of the transportation plan creation system. 12. The transportation plan creation system according to claim 10 or 11, wherein one solution of the travel plan is to arrange a baggage index specifying a baggage to be delivered by a first transportation flight in the delivery order, and A plurality of transportation flights are arranged such that the parcel indexes identifying the parcels to be delivered by the second transportation flight are arranged in the delivery order, the parcel indexes identifying the parcels to be delivered by the third transportation flight are arranged in the delivery order, and so on. The parcel index part in which the parcel indexes of parcels to be delivered according to the above are arranged in the order of delivery, an identifier for identifying the transport flight, an identifier for identifying the type of the transport flight, and the start of the parcel to be delivered by the transport flight in the parcel index region. A table including a position and an end position of a package to be delivered by the transportation flight in the luggage index site, arranged for each of the plurality of transportation flights. Transportation planning system, characterized in that the data represented in the form.