JP2747130B2 - Schedule automatic creation processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a schedule for an electronic computer.
More particularly, the present invention relates to a schedule creation speeding-up method for efficiently allocating a plurality of jobs and a plurality of resources under certain constraints.

[0002]

2. Description of the Related Art Assigning a plurality of jobs and a plurality of resources under a certain constraint condition includes, for example, creating a shift work schedule for humans, a schedule for a production process in a factory, and AV equipment. This is a task that is routinely performed in a wide variety of fields, such as recording a video and creating a recording schedule.

Heretofore, such assignment work has been carried out exclusively by hand using expert know-how, but in recent years, with the development of computer technology, automation by machines has been actively promoted. As this automation method, there are a mathematical programming approach and a knowledge engineering approach. The former is effective only when the format of the constraint condition is established, such as when the constraint condition can be described by an equation or an inequality. However, in an actual problem, the constraints are often described in a more complicated format. Therefore, the heuristic approach based on the latter approach is more applicable. As a known example related to this, for example, JP-A-63-163669,
It describes a method of assigning jobs starting from the job with the most restrictive conditions.

[0004]

As a knowledge engineering approach to the schedule creation problem, the following processing procedure has conventionally been adopted.

[0005] (1) A plurality of jobs are expressed as frames serving as schedule units, stored in an ordered manner. (2) Take out the top frame. (3) One resource is extracted from a plurality of resources. (4) Check whether the taken out resource satisfies the allocation constraint. (A) If they match, the assignment for the frame is determined. (B) If not, the next resource is taken out and the constraint condition is checked. (C) When there are no more resources to be taken out, the processing of the previously processed frame is canceled and the processing is restarted from the processing of (3). That is, backtrack processing is performed. (5) Returning to (2), the next frame is allocated. In addition, until there are no unassigned pieces
The processes from (2) to (5) are repeated.

When a schedule is created by the above procedure, the time required for creation largely depends on the number of backtracking processes for re-assigning. Previous JP 63
The method described in JP-163669A employs the number of resources that can be allocated to each job as an index indicating the degree of severity, and allocates the jobs starting from the job with the smaller value. However, when this method is applied to schedule creation, there is a problem in that the number of redoing of resource allocation is not always reduced because the relationship between a plurality of constraint conditions is not considered.

An object of the present invention is to provide a schedule by an electronic computer.
An object of the present invention is to speed up schedule creation by reducing the number of backtracking processes in the automatic schedule creation process.

[0008]

To achieve the above object of the Invention The present invention is, by an electronic computer, to automatically create a schedule performed according to the allocation constraints between the plurality of jobs and a plurality of resources, the job Into a frame which is a schedule unit, the resource allocation order of each frame is determined from the constraint relationship of the constraints between the frames, and the resources are allocated to each frame under the constraints according to the determined order. It was done. Here, the frame
A binding relationship is one that is imposed on multiple pieces
Constraints bind each other when assigning resources
Relationship.

[0009]

In order to reduce the number of backtracking processes, efficient ordering in allocating resources to frames serving as schedule units in advance is searched for from the binding relationship of the constraints between the frames as follows. (1) Express the constraints imposed on the schedule in a graph. Here, a frame is a node, and when a constraint condition is imposed between two frames, a link between the nodes is established. (2) Weight each link indicating the importance of the constraint condition. (3) Assign a number to each node and perform ordering. (4) For each node, find the maximum value of the distance between adjacent nodes. This distance is defined using the difference between the weight on the link and the node number. (5) The maximum value of all the nodes of the distance obtained in (4) is obtained. (6) The maximum value obtained in (5) becomes the ordered feature amount given in (3). An order that minimizes this feature amount is searched for, and the final frame assignment order is determined.

[0010] The order of resource allocation to frames determined by the above processing is such that a plurality of frames that are mutually constrained by constraints are allocated at timings as close as possible. For this reason, resources to be allocated to a certain frame are lost, and even when performing backtracking processing, the number of frames going back to redoing allocation is as small as possible, making it possible to speed up schedule creation Becomes

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows a schedule of the present invention using an electronic computer .
1 is a block diagram of an embodiment of an information processing system for realizing a method for automatically creating a table . The system includes a job storage device 1, a resource storage device 2, a constraint condition storage device 3,
The job selection device 4 includes a job selection device 4, a resource selection device 5, an assignment determination device 6, and an assignment status storage device 7. The job selection device 4 further includes a schedule unit expansion unit 41, an assignment order determination unit 42, and an assignment order storage unit 43. .

The storage device 1 stores information on each job as shown in FIG. For example, as job number 1, information such as job name A, work time is the morning of the day, three workers, required qualifications are B level, work content is C and location is A are stored. Resource storage device 2
Stores information on each resource as shown in FIG. For example, as the resource number 1, the resource name A, the qualifications B and C held, the possible work A, B,
D, no time conditions, no place conditions, A, E, F,
And other information are recorded. The constraint condition storage device 3 stores information on the relationship between jobs and resources as shown in FIG. For example, as the constraint condition number 1,
Information such as "content for night work requires at least one person with qualification B or higher" is recorded.

In the job selecting device 4, first, the schedule unit developing unit 41 reads out a job from the job storage device 1, develops each job and converts it into a frame format. For example, frame A expands to the first staff member who works the morning of job 1 on the first day, frame B the second staff member similarly, frame C expands to the first staff member who performs the afternoon job of job 2 on the first day, and so on. . A frame is a unit for allocating each resource. Next, the allocation order determining unit 42 determines the order of resource allocation to each frame. At this time, the allocation order determining unit 42 calculates an allocation order that allows the schedule to be created most efficiently, and stores it in the allocation order storage unit 43. I do. This assignment order determination unit 4
2 is the feature of the present invention.

The resource selection device 5 reads resources from the resource storage device 2 and selects them one by one. The assignment determination device 6 checks whether the job (frame unit) and the resource selected by the job selection device 4 and the resource selection device 5 do not violate the constraint conditions stored in the constraint condition storage device 3. If all the constraints are satisfied, the assignment is determined, and the assignment status storage device 7
To be stored. If even one constraint is not satisfied, another assignment is examined and an assignment that satisfies all the constraints is searched.

FIG. 5 is a flowchart showing a processing procedure in the job selection device 4. First, the schedule unit expansion unit 41 expands the job read from the job storage device 1 into frames serving as schedule units, and refers to the constraint conditions of the constraint condition storage device 3 to restrict the constraints imposed as a relationship between a plurality of frames. Is represented by a graph together with the data (steps 501 and 502). Where the top is a node,
Constraints are represented by links. Further, the constraint condition represented by the link is weighted, and the constraint condition is given strength (step 503). Next, in the assignment order determining unit 42,
A number is assigned to each frame represented as a node in the graph, and one assignment order (resource assignment order candidate) is determined (step 505). For this ordering, a feature amount reflecting the binding relationship between the frames is calculated (step 506). An example of a method of calculating the feature amount will be described in detail with reference to FIG. The obtained current feature amount is compared with the previous feature amount (the initial value is assumed to be the maximum possible feature amount) (step 507). If the current feature amount is larger, the next feature value is immediately obtained. The process proceeds to the assignment. If the current feature amount is smaller, the current order and the feature amount are stored (step 508), and then the process proceeds to the next assignment. In this way, the feature amount is calculated for each ordering of all the frames, the ordering in the smallest order is searched for, and the result is stored in the assignment order storage unit 43 (steps 504 and 509).

FIG. 6 is a detailed flowchart of the feature value calculation in step 506. First, one frame is determined, the distance to each frame adjacent thereto is measured, and the maximum value is obtained (steps 602 and 603). The distance between the frames is defined based on the difference between the numbers assigned to the frames, the value of the weight assigned to the link connecting the frames, and the like. By comparing the current calculation result with the previous calculation result and storing the larger one (steps 604 and 605), the maximum value of these distances is calculated for each frame, and the largest value is adopted as the feature quantity of this ordering. (Steps 601, 60
6).

The assignment order determination unit 42 of the job selection device 4
In the above, the assignment order obtained according to the above-described processing procedure minimizes the number of backtracking processes performed as a re-execution in normal schedule creation.

For example, FIGS. 7 (a) and 7 (b) are graphs showing the constraints on six frames A, B, C, D, E and F. Here, it is assumed that the weights of the constraint conditions are all equal. In FIG. 7A, the distance between the frames is defined as the absolute value of the difference between the numbers, and the ordering is performed so that the feature amount becomes as small as possible. on the other hand,
FIG. 7B shows a random order for each frame. The numbers indicate the assignment order.

FIGS. 8A and 8B show the backtracking process accompanying the resource allocation in the allocation determining device 6 for the frames ordered as shown in FIGS. 7A and 7B, respectively. ) And (b). Here, it is assumed that there are two resources for each frame. The numbers indicate the allocation order, and "x" indicates that the allocation has failed. The bold line indicates that the assignment has actually been performed, the thin line indicates that the assignment has not been performed, and the dotted line indicates that the assignment has not yet been considered.

In the example of FIGS. 8A and 8B, since there is a constraint between the frame C and the frame D, the resources to be allocated to the frame D are affected by the allocation of the resources to the frame C. This shows a situation in which the backtracking process has occurred and backtracking has occurred. In this case, in FIG. 8A, three backtracking processes are completed before redistribution to the frame C is performed. However, in FIG. Processing is increasing.

[0022]

As it is apparent from the foregoing description, according to the present invention, in the automatic Sukeji <br/> Yuringu performed in accordance with the assignment constraints between the plurality of jobs and a plurality of resources by computer, scheduling When allocating resources to a unit frame, consider the relationship between multiple constraints on multiple frames, and allocate resources to a frame at a timing as close as possible to adjacent frames that are strongly linked by the constraints. By calculating the order to be performed and adopting an order that reduces the number of reassignments, the number of backtracking processes is significantly reduced, and the schedule can be created at a higher speed.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an embodiment of an information processing system for realizing a schedule creation method of the present invention.

FIG. 2 is a diagram illustrating a specific example of a job.

FIG. 3 is a diagram illustrating a specific example of a resource.

FIG. 4 is a diagram showing a specific example of a constraint condition.

FIG. 5 is a flowchart of an assignment order determination processing procedure in the job selection device.

FIG. 6 is a detailed flowchart of a feature amount calculation procedure in an assignment order determination unit.

FIG. 7 is a diagram showing a specific example of a graph expression of a constraint condition.
(A) assigns the assignment order according to the present invention to each node, and (b) assigns a random order to each node.

FIGS. 8A and 8B are diagrams showing a backtrack processing occurrence state at the time of resource allocation, wherein FIGS.
This is when assignment is performed in the order shown in (a) and (b).

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 job storage device 2 resource storage device 3 constraint condition storage device 4 job selection device 41 schedule unit expansion unit 42 allocation order determination unit 43 allocation order storage unit 5 resource selection device 6 allocation determination device 7 allocation status storage device

Continued on the front page (72) Inventor Kazuhiro Okamoto 6-81 Onoe-cho, Naka-ku, Yokohama-shi, Kanagawa Prefecture Inside Hitachi Software Engineering Co., Ltd. (72) Inventor Mariko Kondo 6-81-Oue-cho, Naka-ku, Yokohama-shi, Hitachi Software Engineering Co., Ltd. (56) References JP-A-63-214874 (JP, A) JP-A-63-163667 (JP, A) JP-A-2-29867 (JP, A) JP-A-3-113669 (JP, A) JP-A-3-228175 (JP, A) RICHARD A. SPRAGUE ET AL, "Backtracking in Scheduling" Information Processing Society of Japan, VOL. 35, NO. 2 (1987), PP. 1649-1650 (58) Field surveyed (Int. Cl. ⁶ , DB name) G06F 17/60 G06F 17/00 JICST file (JOIS)

Claims (1)

(57) [Claims] In a processing method for automatically creating a schedule by allocating a plurality of jobs and a plurality of resources according to constraints using an electronic computer, a step of expanding the jobs into frames serving as schedule units Generating a graph that expresses the relationship between the constraint conditions between the frames by linking between the nodes when constraints are imposed between the two frames, and All combinations of the resource allocation order with respect to are calculated (hereinafter, each combination is referred to as a resource allocation order candidate), and for each resource allocation order candidate, the distance between each frame is calculated based on the graph, and the maximum value is calculated. Is the feature amount of the resource assignment order candidate, and the resource assignment order candidate with the smallest feature amount is determined as the resource assignment order. Step and, schedule automatic creation processing method characterized by comprising the steps of: allocating resources to each frame according to the resource allocation order in which the determined.