JP2001318711A - Scheduling device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a scheduling device capable of not only dissolving overloads at the time of scheduling the day's program but also dissolving load insufficiency so as not to fall into a production capacity excess and balancing loads. SOLUTION: This scheduling device is provided with an initial task stacking part for executing the initial stacking of tasks which are the works of respective processes required for manufacturing a product, a bottom search part for searching the minimum bottom which is the lowest load part within a schedule period stacked in the initial task stacking part, an adjacent peak search part for searching a peak adjacent to the minimum bottom, a load balancing processing part for moving the task present between the adjacent peak and the minimum bottom to the minimum bottom by a prescribed method and balancing the loads, a balancing degree evaluation part for evaluating a load balancing processing result in the load balancing processing part, an end judgment part for executing the end judgment of a processing based on the result of balancing evaluation and a schedule output part for outputting the load balancing processing result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a scheduling apparatus for leveling loads in a work process in small-lot production of many kinds.

[0002]

2. Description of the Related Art Conventionally, as an automatic scheduling apparatus which can cope with high-mix low-volume production, an apparatus incorporating a forward scheduling method and a backward scheduling method is generally known. The forward scheduling method is a method of stacking loads in accordance with the possible work start time of a process, and crushing the resulting work in an overloaded portion in the positive direction of the time axis (forward). The technique is a technique in which piles are performed starting from the completion delivery date of the process, and the work in the overload portion is collapsed in the minus direction (rearward) on the time axis to determine the start time of the process.

[0003]

All of the devices adopting the above-mentioned conventional scheduling method are intended for the device type industry, and basically assume the existence of the device.
This is a scheduling device whose main purpose is to adjust the overload portion of the process. Therefore, for each process, even if the apparatus is not operating due to a temporary shortage of load, there is basically no need to adjust the load unless the apparatus is overloaded.

[0004] On the other hand, in heavy industry that manufactures a wide variety of products based on order-based production, the load of the process is significantly different as the variety of products is increased. Furthermore, there are many worker-oriented processes in which a large number of workers are engaged, and a decrease in the operation rate due to a lack of load results in the generation of extra workers, which is larger than the ratio of the worker cost to production cost. Loss will occur. Therefore, in the scheduling method as well, not only the overload is eliminated but also how to eliminate the insufficient load and level the load so as not to cause an excessive production capacity is a major issue.

According to the present invention, there is provided a scheduling apparatus capable of solving the above-mentioned problems and eliminating a load shortage so as not to cause an excess of production capacity as well as eliminating an overload at the time of schedule planning and leveling the load. The purpose is to do.

[0006]

The above object is achieved by the following invention.

[1] An initial task stacking section for performing an initial stacking of tasks, which is an operation of each step necessary for manufacturing a product, and a portion having the lowest load within a planning period stacked by the initial task stacking section. A bottom search unit that searches for a certain minimum bottom, an adjacent peak search unit that searches for a peak adjacent to the minimum bottom, and a task existing between the adjacent peak and the minimum bottom is moved to the minimum bottom by a predetermined method. A load leveling processing unit that performs load leveling, a leveling degree evaluation unit that evaluates the load leveling processing result in the load leveling processing unit, and determines the end of the process based on the result of the leveling evaluation. A scheduling device comprising: an end determination unit to be implemented; and a plan output unit that outputs a load leveling process result.

[2] In the above [1], the load leveling processor selects only the tasks contributing to the increase in the minimum bottom load and generates a set of task lists for each of the left and right, and a minimum bottom. A landslide target setting section that sets a landslide target value for, a task list selection section that selects which of the left and right task lists to select, and a move task selection that selects one movable candidate task from the task list A task movement determining unit that determines whether the task can be moved and whether the task movement violates the constraint condition, and a task moving unit that moves the task for a predetermined time when the task can be moved. The task list updating unit that updates the task list according to the result of the movement, and the landslide to the bottom is satisfied as a result of the task movement. Scheduling apparatus; and a results mountains determines whether the host vehicle has reached the breaking determination unit.

[3] In the above item [2], the task list generation unit includes not only the tasks contributing to the increase in the minimum bottom load but also the tasks existing between the tasks and the minimum bottom that do not directly contribute to the load leveling. A scheduling device characterized in that by selecting and generating a set of task lists for each of the left and right sides, load leveling such that reversal of a work start order and / or reversal of a work completion order falls within a restricted range.

[4] In the above item [2], the task list generation section selects all the tasks existing between the minimum bottom and the adjacent peaks on the left and right sides, generates a set of task lists for each of the left and right sides, and sets a work start order. A scheduling apparatus characterized in that by performing load leveling such that the reversal and / or the reversal of the work completion order fall within the restricted range, it is possible to perform load leveling with the leveling target range expanded.

[5] In the above [1] to [4],
A max peak search unit that searches for the peak with the highest load within the planning period to be searched, an adjacent bottom search unit that searches for a bottom adjacent to the peak, and a task that exists between the adjacent bottom and the peak. A peak load leveling processing unit that moves to the adjacent bottom and performs load leveling by a predetermined method, a peak leveling degree evaluation unit that evaluates a load leveling process result in the peak load leveling processing unit, A scheduling device comprising: a peak leveling end determining unit that determines whether to end processing or to continue processing based on evaluation by a peak leveling degree evaluating unit.

[6] In the above [5], the peak load leveling processing section selects only the tasks contributing to the reduction of the load of the maximum peak, and generates a set of task lists for each of the right and left peaks. And a peak leveling landslide target setting section that sets the landslide target value for the Max Peak, a peak leveling task list selection section that selects which of the left and right task lists to select, and movable within the task list A peak leveling movement task selecting unit for selecting one of the candidate tasks, and a peak leveling task movement determining unit for determining whether the task can be moved and whether the task movement can violate the constraint condition. A peak leveling task moving unit that moves a task for a predetermined time when the task can be moved, and a task based on the result of the movement. A peak leveling task list updating unit for updating a list; and a peak leveling hillslide determination unit for determining whether or not a result of the task movement has satisfied the hillslide for the peak. Scheduling device.

FIG. 1 is a block diagram according to the present invention.
This will be described below with reference to the block diagram of FIG.

First, in the initial task stacking section, an initial stack of tasks, which is an operation of each step necessary for manufacturing a product, is performed. As a result of the initial pile, the tasks for each process are arranged in a sort order of work start or work completion time. The bottom search unit searches for the minimum bottom, which is the least loaded part of the process within the planning period to be searched, by a predetermined method based on the results of the initial pile of tasks.

The adjacent peak search section searches for an adjacent peak in which loads adjacent to the left and right sides of the minimum bottom searched by the bottom search section are local peaks.

In the load leveling processing unit, a task existing between the adjacent peak and the minimum bottom is moved to the minimum bottom by a predetermined method to perform load leveling.

The load leveling processor includes a task list generator for selecting only the tasks contributing to the increase in the minimum bottom load and generating a set of task lists for each of the left and right, and a hill-slip target value for the minimum bottom. A landslide target setting part, a task list selection part for selecting which of the left and right task lists is to be selected, a movement task selection part for selecting one movable candidate task in the task list, and a task movement A task movement determining unit that determines whether the task can be moved and whether the task can violate the constraint, a task moving unit that moves the task for a predetermined time when the task can be moved, and a result of the movement. As a result of the task list update unit and task movement, the task of updating the task list reaches Or it is preferred to have a whether mountain determines breaking determination unit.

Further, the task list generation unit selects not only tasks that contribute to the increase in the load of the minimum bottom but also tasks that exist between the tasks and the minimum bottom and that do not directly contribute to the load leveling. By generating a set of task lists, it is possible to perform load leveling such that the reversal of the work start order, the reversal of the work completion order, or the reversal of both work orders falls within a predetermined constraint range. .

Further, the task list generating section generates a set of task lists for each of the tasks existing between the minimum bottom and the adjacent peaks on the left and right, for each task. The load leveling can also be performed while moving, under the condition that the order reversal and / or the work completion order reversal fall within the constraint range.

The leveling degree evaluation section evaluates whether the load leveling is satisfied with respect to the bottom being processed by the load leveling processing section, and determines the end or continuation of the processing for the bottom.

The end determination unit determines whether to search for another bottom or to end the leveling process.

As described above, there is provided a scheduling apparatus capable of resolving a load shortage at the time of scheduling.

Further, the scheduling apparatus according to the present invention includes a maximum peak search unit for searching for a peak having the highest load within a planning period to be searched, and an adjacent bottom search unit for searching for a bottom adjacent to the peak. A peak load leveling processing unit that moves a task existing between the adjacent bottom and the peak to the adjacent bottom by a predetermined method to perform load leveling, and a load leveling processing result in the peak load leveling processing unit. It is possible to cope with an overload by having a peak leveling degree evaluating unit for performing the evaluation of the above, and a peak leveling end determining unit for determining the processing end or the processing continuation based on the evaluation by the peak leveling degree evaluating unit. It is a scheduling device.

The peak load leveling processing section selects a task that contributes to the reduction of the load of the maximum peak and generates a set of task lists for the left and right sides, and a peak leveling task list generating section. Peak leveling landslide target setting part to set target value, peak leveling task list selection part to select left or right task list to be selected, and select one movable candidate task from task list A peak leveling move task selecting unit, and a peak leveling task move determining unit that determines whether the task can be moved and whether the task can violate the constraint, and a task that can be moved. A peak leveling task moving unit for moving a task for a predetermined time, and a peak leveling task for updating a task list based on a result of the movement. Scheduling device capable of coping with overload by having a generalized task list updating unit and a peak leveling landslide determination unit for determining whether or not a result of task movement has reached a satisfactory result of the landslide for the peak It has become.

As a result, in addition to resolving the load shortage at the time of scheduling, the overload portion can be leveled, and a scheduling apparatus capable of creating a schedule with more leveled load can be provided. Note that a series of processing in the max peak search unit, the adjacent bottom search unit, the peak load leveling processing unit, the peak leveling degree evaluation unit, and the peak leveling end determination unit includes a series of processes described in claims 1 to 4. The processing may be performed subsequently to the load shortage resolving process or may be performed before the load shortage resolving process. In addition, it may be performed independently or not.

[0026]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in more detail with reference to the block diagram of FIG.
Note that the bottom search section, the adjacent peak search section, the load leveling processing section, the leveling degree evaluation section, and the end determination section are collectively called a bottom leveling section, and the max peak search section, the adjacent bottom search section, the peak load leveling section. The processing section, peak leveling degree evaluation section, and peak leveling end determination section are collectively referred to as a peak leveling section.

[Initial Task Stacking Section] The initial task stacking section includes an initial setting section, an information reading section, a process master information reading section, a standard lead time calculating section, an earliest and latest schedule calculating section, a process load stacking section, and a leveling target. It has a setting unit.

The initial setting section initializes the non-target bottom list and the number of times of bottom search, and initializes various parameters. Table 1 shows an example of parameter setting items in the initial task stacking section and the bottom leveling section, and Table 2 shows an example of parameter setting items in the peak leveling section.

[0029]

[Table 1]

[0030]

[Table 2]

The information reading unit includes production plan information recording the scheduled production start date and time and production completion date and time of all products to be produced, process design information recording all work process procedures relating to production for each product, It reads and stores the management amount information in which the management amount information serving as the basis for calculating the work lead time in each work process is recorded for each product.

Here, the management amount information is, for example, an attachment length obtained by summing up the lengths of the attachment portions of the members to be attached, or a welding length obtained by summing up the lengths of the welding portions of the members to be welded.

The process master information reading unit reads and stores process master information, which is a basis for calculating the processing capacity of each work process.

Here, the process master information is a database in which the work constraints, the equipment capacity, the work efficiency and the like for each process are described. The work constraints for each process are as follows:
Equipment constraints such that the required number of workers is naturally stipulated from work constraints such as cranes in the work process, and joint work of multiple workers is efficient based on past work experience etc., and the optimal number of workers Are naturally divided into work posture constraints that are naturally defined. As an example of such a restriction, in assembly work, one worker operates a crane and another worker performs positioning of a mounting member,
Such a work posture of a pair of two persons corresponds.

The standard lead time calculation unit calculates a standard lead time representing the processing time in each work process for each product from the information read by the information reading unit and the process master reading unit.

For example, assuming that the working efficiency in the member mounting work process is 10 M / hour / person, a value obtained by dividing the managed amount by the working efficiency, that is, a standard lead time of 10 hours is required for mounting a member having a controlled amount of 100 M. Is calculated. The same applies to a processing machine. For a drilling device with a drilling capability of 1 minute / hole drilling, the standard lead time for a process with 200 holes is a control amount, which is 200 minutes. Becomes

The earliest / latest schedule calculation unit calculates the earliest work start date and time and the latest work completion date and time in each work process for each product based on the manufacturing plan information read by the information reading unit.

The earliest work start date and time is calculated by adding the standard lead time obtained for each process in order from the first process, starting from the scheduled production start date and time for each product described in the manufacturing plan information. . Further, the latest work completion date and time is calculated by subtracting the standard lead time obtained for each process in reverse order from the last process, starting from the planned manufacturing completion date and time described in the manufacturing plan information.

The process load stacking section performs the initial stacking of work for each work process in the following procedure. (1) The reference work start date and time obtained by adding the minimum required inter-process stock time between processes to the corresponding process to the earliest work start date and time obtained by the earliest and latest schedule calculation unit, A reference work completion date and time is determined by subtracting the stock time between processes in each process after the corresponding process from the work completion date and time. Here, the inter-process stock time is a process waiting time, and when a trouble occurs in a previous process, a predetermined amount of an intermediate product (assembled part) is stocked so that a subsequent process is not affected. This is a margin time that occurs due to the setting, and usually, a predetermined time is set as an inter-process stock time when a manufacturing plan is drafted.

(2) A period between the reference work start date and time and the reference work completion date and time is set as a reference work possible period in each process.

FIG. 2 shows, as an example, a reference workable period for a product having four work processes. A,
The periods indicated by B, C, and D are the reference workable periods in steps 1, 2, 3, and 4. That is, each work process is arranged according to the predetermined setting condition within the reference workable period.

(3) Pile of work (tasks) consisting of the standard lead time in each work process during the reference workable period. Here, it is a prerequisite for making a process plan that the standard lead time is shorter than the reference workable period.

When the standard lead time is longer than the reference workable time, a method of changing the processing result in the standard lead time calculation unit to shorten the lead time, or reducing the inter-process stock time is used. Corresponding.

As a stacking method, a method of stacking tasks consisting of a standard lead time according to the reference work start date and time, a method of stacking tasks according to the reference work completion date and time,
Alternatively, the stacking is performed by any of the methods of stacking so that the margins from the reference work start date and time and the reference work completion date and time are equal. This selection is performed in “selection of a stacking method” in the initial setting section shown in Table 1. It should be noted that which stacking method is selected can be appropriately selected according to the production method or the like. For example, in the case of a production system that emphasizes minimizing the stock of finished products, a method of stacking in accordance with the reference work completion date and time is fundamental. On the other hand, when the production is started when the material is received, the pile is piled up in accordance with the reference work start date and time. In this case, you will have stock as a finished product.

(4) Steps (1) to (3) are repeated for all work steps for all products. The unit of time for stacking tasks depends on the lead time of the process and the accuracy (fineness) of the schedule for performing load leveling. For example, the unit of time is one day, one half day, one hour, and 30 days. Minutes, 1
The setting can be made in units of 5 minutes, 10 minutes, 5 minutes, 1 minute, and the like. This time unit is hereinafter referred to as a management time unit (step).

The leveling target setting unit calculates the average load amount within the planning period to be load leveled for each work process from the task stacking result in the process load stacking unit. Note that the average load amount obtained here means a load amount when all loads within the plan period can be processed without individual time restrictions. For the average load, a leveling target value is calculated based on the parameters set by the initial setting unit. As an example, the leveling target in the initial pile of FIG.
38 points, and the average value is 138/23 = 6 points, so this value is set as the leveling target value. As another method, an arbitrary target value may be set in addition to the setting of the target value by the evaluation parameter calculated by the statistical method.

Processing in the following steps is performed based on the initial stacking result created in the initial task stacking section (process load stacking section) and each parameter set in the initial setting section. In this case, the processing method differs depending on the task list generation type of the load leveling processing unit.

The movable step in the left-right direction of each task based on the initial stacking result in the initial task stacking unit is the number of units in the management time unit that can be moved within the reference workable period shown in FIG. However, in the following description of the embodiments, for simplicity, each task is described as being able to move left and right by one step.
If a task that can be moved left and right one step at a time moves one step to the right, for example, the rightward moveable step for this task is zero, the rightward movement impossible flag is set, and the Therefore, the number of movable steps is 2.

[In the case of TYPE-A] FIG. 3 is an explanatory diagram for explaining the processing when TYPE-A is selected as the task list generation type. FIG. 3 (a) to (h)
Shows a piled-up graph of the tasks in the planning period (B to X) in the middle of the load leveling process, and the tasks (a to ii) constituting the respective peaks. The shaded portions of the tasks in the figure show the results of the task movement by the mountain crushing process.

Note that TYPE-A is a method of selecting only tasks that contribute to minimum-bottom load leveling.

FIG. 3A shows an initial stacking result in the initial task stacking unit. As a stacking method, stacking is performed in the order of task start time. However, even when stacking is performed in the order of task completion time, the same processing may be performed below.

[Bottom Search Section] In the pile graph of FIG. 3A, when the load is less than the above-mentioned leveling target value,
And, among the bottoms that are not listed in the bottom list that is not targeted or that are not listed in the bottom list that is not targeted but that have not yet reached the “maximum number of times of processing for the same bottom” defined in the initial setting unit shown in Table 1. Search for the bottom (minimum bottom) with the minimum load.

Here, in the search for the minimum bottom according to the present embodiment, low load portions at both ends of the planning period, for example, portions B and X in FIG. Only the concave bottom having a peak at is selected. This is because, for example, the left end of the planning period may have work already in progress before the start of the planning period (it does not appear as a task requiring adjustment), and the right end may be added in the future, This is because it is considered that there are many cases where correction is required even at the both ends of the planning period even if the load is adjusted. Needless to say, the low load portions at both ends of the planning period may be included in the selection. In this case, the task list created is only the right list at the left end, and the left list is at the right end. Only the processing method of load leveling is the same.

In the first bottom search, since the non-target bottom list has been initialized at the initial setting stage, all bottoms are targeted. Here, as a result of the search, a minimum bottom R that is lower than the leveling target value of 6 points is searched. If there is a bottom at the same level as a result of the search, processing is performed according to the parameters set by “selection of priority when bottom at the same level exists” in the initial setting unit. in this case,~
A method of selecting one of the above, and
Alternatively, there is a method of selecting a combination.

[Adjacent Peak Search Unit] The adjacent peaks on the left and right of the minimum bottom R searched by the bottom search unit are searched. Here, as a result of the search, adjacent peaks N and T are searched. In the case where a plurality of peaks are formed (in this case, the right peak is formed of S and T), a peak far from the minimum bottom is selected. For the purpose of reducing the target range, a peak close to the minimum bottom may be selected. ("Priority when there is a peak of the same level" in the initial setting section)

[Load leveling processing unit] The task list generation unit generates a task list by TYPE-A. Here, TYPE-A is a set of task lists for each of the left and right sides that selects only tasks that contribute to an increase in the minimum bottom load by moving tasks among all movable tasks existing between the bottom and peak. It is a method of generating. The left list arranges tasks in which tasks are completed before the minimum bottom R and the work start time is later, and the right list arranges tasks in which the task work starts after the bottom R and the work start time is earlier. Table 3 shows the results of task list creation.

[0057]

[Table 3]

In the description of the embodiment of TYPE-A, the task list does not include the task forming the bottom in the "determination of whether or not the bottom task can be selected" of the initial setting section shown in Table 1. It is assumed that it is selected. The task list is configured so that not only the task name but also a processing pointer indicating the task to be processed, the number of steps that can be moved, and a non-movable flag in the case of non-movable can be set.

The landslide target setting section sets a landslide target value for the minimum bottom. In the present embodiment, the landslide target value is set to the above-described leveling target value, but a local landslide target is set as the landslide target value, and a leveling target value that is an index of the leveling degree of the entire plan is set. Can also be set separately.

The task list selecting section selects a task list according to the "task list selecting method" set in the initial setting section. The method of prioritizing the task list with the higher peak enables the hills with higher peaks. In the method of giving priority to the task list on the left side, the task is shifted to the delivery date side due to a landslide, so that the stock between processes with the next process can be reduced. When the size of the intermediate product gradually increases as in the assembling work process, there is an advantage that the number of stock locations can be reduced. The method of giving priority to the task list on the right side can reduce the stock between processes between the previous process and the previous process. The method of selecting a list alternately from left and right can be achieved by evenly sloping from the left and right, so that the range of tasks where changes occur can be suppressed to near the minimum bottom. That is, a change from the initial plan can be reduced.

Here, the left list is selected by a method of giving priority to the task list with the higher peak. In any of the methods, the target load leveling is performed. Therefore, an appropriate task selection method may be determined based on target product characteristics, process characteristics, equipment characteristics, and the like.

The moving task selecting section selects a task indicated by the processing pointer. At the time of generating the list, the processing pointer is set to the highest task excluding the immovable task from the task list generated based on the list selected by the task list selecting unit. Here, the cc at the top of the left list is selected.

The task movement determining section determines whether or not the task selected by the moving task selecting section can be moved based on the parameters set by the initial setting section.
Can be moved. The task movement determination includes, in addition to the determination of whether or not movement is possible based on the movement impossible flag, determination of whether or not the task order is reversed, whether the task start or task end time is reversed by moving the task, and whether the bottom is equal to or less than the minimum bottom. Is determined, and whether a bottom below the leveling target value is generated is determined.

The task moving section moves the task determined to be movable by the task moving determining section. Here, the task cc is moved to the right by one step. At this time, the number of steps that can be moved in one task movement is set in the initial setting, and the task is gradually and repeatedly moved from the upper part to the lower part of the task list. This prevents the task start order from being reversed. In this description, only one step is moved at a time.

The task list updating unit moves the processing pointer one step down with respect to the moved task. When the pointer reaches the bottom of the list, move it back to the top. At the same time, if the moved task becomes immovable, a non-movable flag is set in the task list. Here, since the task cc moves one step to the right and cannot move in the right direction, a rightward movement impossible flag is set, and the processing pointer is moved from the task cc to aa.

The landslide determination section compares the minimum bottom R after the movement of the task cc with the landslide target value. Here, it is assumed that the landslide target value is 6 points which is the average value of the pile amount within the planning period.

Since the minimum bottom R after the movement of the task cc has not reached the target at 2 points, the processing pointer is confirmed. Since the processing pointer is not at the top, the processing returns to the moving task selection unit.

[Load leveling processor] The task aa is selected in the moving task selector. The task aa is moved to the right by one step in the task movement unit according to the determination of whether movement is possible in the task movement determination unit. In the task list update section,
Set a rightward immovable flag for task aa,
The processing pointer is returned to the head of the list, that is, the task cc.

The landslide determination unit compares the minimum bottom R after the movement of the task aa with the landslide target value (6 points). Since the minimum bottom R after the movement of the task aa has not reached the target at 3 points, the processing pointer is confirmed. Since the processing pointer is at the top, the process returns to the task list selection unit. FIG. 3B shows this state. The task list selection unit can also switch the selected task list based on the initial setting conditions.

[Load leveling processing unit] The task list selection unit selects the right list because there are no more tasks that can be processed in the left task. At this point, if the immovable flag is not set in the left task list, that is, if there is a task having one or more movable steps,
After the tasks are repeatedly selected by the moving task selection unit, the process moves to the right task list.

The task dd is selected in the moving task selecting section,
The task dd is moved to the left by one step in the task moving unit according to the determination of whether movement is possible by the task movement determining unit. The task list updating unit sets a leftward movement impossible flag for the task dd, and moves the processing pointer from the task dd to the task ee.

The landslide determination section compares the minimum bottom R after the movement of the task dd with the landslide target value (6 points). Since the minimum bottom R after the movement of the task dd has not reached the target at 4 points, the processing pointer is confirmed. Since the processing pointer is not at the top, the processing returns to the moving task selection unit.

[Load leveling processing unit] The task ee is selected in the moving task selection unit. The task ee moves the task ee to the left by one step in the task movement determining unit based on the determination of the movement possibility. In the task list update section,
Set a leftward movement impossible flag for task ee,
The processing pointer is moved from task ee to ff.

The landslide judgment unit compares the minimum bottom R after the movement of the task ee with the landslide target value (6 points). Since the minimum bottom R after the movement of the task ee has not reached the target at 5 points, the processing pointer is confirmed. Since the processing pointer is not at the top, the processing returns to the moving task selection unit.

[Load leveling processing unit] The moving task selection unit selects the task ff. The task ff is moved to the left by one step in the task moving unit according to the determination of whether movement is possible by the task movement determining unit. In the task list update section,
Set a leftward movement impossible flag for task ff,
The processing pointer is moved from task ff to gg.

The landslide determination unit compares the minimum bottom R after the movement of the task ff with the landslide target value (6 points). The minimum bottom R after the movement of the task ff reaches the target at 6 points, and the processing for the minimum bottom R ends. FIG. 3C shows this state. R is stored in the non-target bottom list, and the number of processing times 1 is also stored.

[Equalization degree evaluation unit] The equalization degree evaluation unit
As a result of the load leveling process on the minimum bottom R, the degree of load leveling within the planning period is calculated. Here, as a simple calculation method, a minimum bottom load value within the planning period is determined. In addition, a statistical evaluation index may be used. In FIG. 3C, there is a bottom (bottom Q) that has not reached the landslide target value.

[End Judgment Unit] As the end judgment condition, the leveling value obtained by the leveling degree evaluation unit is set to the leveling target allowable range value defined by the leveling target setting unit which is the initial setting condition. A comparison is made to determine whether or not it has been reached. If the load leveling is within the target allowable value range, the process ends. On the other hand, in FIG. 3 (c), since it is calculated that there is a bottom (bottom Q) that has not reached the landslide target value, the process returns to the bottom search unit and the same processing is repeated.

[Bottom Search Section] Searches for the minimum bottom Q.

[Adjacent peak search section] Adjacent peaks O and S
Search.

[Load leveling processing unit] Table 4 shows the result of task list creation by the task list generation unit.

[0082]

[Table 4]

The left list is selected in the task list selection section. The task x is selected by the moving task selecting unit, and the task moving unit moves the task x to the right by one step according to the determination of the movement possibility by the task moving determining unit. The task list updating unit sets a rightward movement impossible flag for task x, and moves the processing pointer from task x to w.

The landslide judgment unit compares the minimum bottom Q after the movement of the task x with the landslide target value (6 points). Since the minimum bottom Q after the movement of the task x has not reached the target at 4 points, the processing pointer is confirmed. Since the processing pointer is not at the top, the processing returns to the moving task selection unit.

[Load leveling processing unit] The task w is selected in the moving task selection unit. Based on the determination of whether movement is possible by the task movement determination unit, the task movement unit moves the task w to the right by one step. The task list updating unit sets a leftward movement impossible flag for the task w, and returns the processing pointer to the head of the list.

The landslide determination section compares the minimum bottom Q after the movement of the task w with the landslide target value (6 points). Since the minimum bottom Q after the movement of the task w has not reached the target at 5 points, the processing pointer is confirmed. Since the processing pointer is at the top, the process returns to the task list selection unit. At this time, there are no more tasks that can be processed in the left list, so the right list is selected. However, as shown in Table 4, there is no task that corresponds to the right list, and the processing for the minimum bottom Q ends. FIG. 3D shows this state. Q is stored in the non-target bottom list, and the processing number 1 is also stored.

[Leveling Evaluator] In the leveling degree evaluator,
As a result of the load leveling process for the minimum bottom Q, the degree of load leveling within the planning period is calculated. In FIG. 3D, there is a bottom (bottom M) that has not reached the landslide target value.

[End Judgment Unit] Since the end judgment unit has calculated that there is a bottom (bottom Q) that has not reached the landslide target value, the process returns to the bottom search unit and repeats the same processing.

[Bottom Search Section] Searches for the minimum bottom M.

[Adjacent peak search section] Adjacent peaks I and O
Search.

[Load leveling processing unit] Table 5 shows the result of task list creation by the task list generation unit.

[0092]

[Table 5]

The left list is selected in the task list selection section. The task t is selected by the moving task selecting unit, and the task moving unit moves the task t to the right by one step according to the determination of whether or not the task can be moved. The task list updating unit sets a rightward movement impossible flag for task t, and moves the processing pointer from task t to p.

The landslide determination unit compares the minimum bottom M after the movement of the task t with the landslide target value (6 points). Since the minimum bottom M after the movement of the task t has not reached the target at 5 points, the processing pointer is confirmed. Since the processing pointer is not at the top, the processing returns to the moving task selection unit.

[Load leveling processing unit] The moving task selecting unit selects the task p. The task p is moved one step to the right in the task moving section by the task movement judging section judging that movement is possible. The task list updating unit sets a leftward movement impossible flag for the task p, and moves the processing pointer from the task p to o.

The landslide determination unit compares the minimum bottom M after the movement of the task p with the landslide target value (6 points). The minimum bottom Q after the movement of task w reaches the target with 6 points, and the minimum bottom M
The processing for is ended. FIG. 3E shows this state. M is stored in the non-target bottom list, and the processing count 1 is also stored.

[Equalization degree evaluation unit] In the equalization degree evaluation unit,
As a result of the load leveling process for the minimum bottom M, the degree of load leveling within the planning period is calculated. In FIG. 3E, there is a bottom (bottom L, P, Q) that has not reached the landslide target value.

[End judging unit] Since the end judging unit calculates that there is a bottom (bottom Q) that does not reach the landslide target value, the process returns to the bottom searching unit and repeats the same processing.

[Bottom search section] Here, L, P, and Q are searched as the minimum bottom, and are set by "selection of priority when bottom of same level is found as a result of search" in the initial setting section. The processing is performed according to the set parameters. Hereinafter, a description will be given of the case of the selection condition of a parameter having a small number of selections and a parameter close to the previous bottom, but the same processing is performed in another selection method.

According to the selection conditions, Q has already been selected once, and the minimum bottom L is selected as the bottom close to the previous bottom M.

[Adjacent peak search section] Adjacent peaks I and O
Search.

[Load leveling processing unit] Table 6 shows the result of task list creation by the task list generation unit.

[0103]

[Table 6]

The left list is selected in the task list selection section. The task s is selected by the moving task selecting unit, and the task moving unit moves the task s to the right by one step based on the determination of the movement possibility by the task moving determining unit. The task list updating unit sets a rightward movement impossible flag for the task s, and moves the processing pointer from the task s to q.

The landslide determination section compares the minimum bottom L after the movement of the task s with the landslide target value (6 points). The minimum bottom L after the movement of task s reaches the target with 6 points, and the minimum bottom L
The processing for is ended. FIG. 3F illustrates this state.

L is stored in the non-target bottom list, and the processing number 1 is also stored.

The similar processing in the leveling degree evaluation section and the end determination section after processing for the minimum bottom L returns to the bottom search section, and the same processing is repeated.

[Bottom Search Section] Here, P and Q are searched as the minimum bottom, but since Q has already been selected once, the minimum bottom P is selected.

[Adjacent peak search section] Adjacent peaks O and S
Search.

[Load leveling processing unit] Table 7 shows the results of task list creation by the task list generation unit.

[0111]

[Table 7]

The left list is selected in the task list selection section. The task z is selected by the moving task selecting unit, and the task moving unit moves the task z to the right by one step according to the determination of the movement possibility by the task moving determining unit. Since the initial setting condition of the task movement determination unit is that “the generation of a bottom smaller than the minimum bottom is allowed”, the task list update unit sets a rightward movement impossible flag for task z, and sets the processing pointer to task z. To y.

The landslide determination unit compares the minimum bottom P after the movement of the task z with the landslide target value (6 points). The minimum bottom P after the movement of task z reaches the target with 6 points, and the minimum bottom P
The processing for is ended. FIG. 3G illustrates this state. P is stored in the non-target bottom list, and the processing number 1 is also stored.

The similar processing in the leveling degree evaluation unit and the end determination unit after processing for the minimum bottom P returns to the bottom search unit and repeats the same processing.

[Bottom Search Section] Here, M and Q are searched as the minimum bottom, and the number of selections is the same for both, and the minimum bottom Q is selected as the bottom close to the previous bottom P.

[Adjacent peak search section] Adjacent peaks O and S
Search.

[Load leveling processing unit] As a result of creating the task list in the task list generation unit, there is no movable task.

A task list selecting section, a moving task selecting section,
In the task movement judgment unit, task movement unit, and task list update unit, the task is not selected, so it passes without processing. The landslide determination unit stores the number of processes 2 for Q.

The similar processing in the leveling degree evaluation section and the end determination section after processing for the minimum bottom Q returns to the bottom search section, and the same processing is repeated.

[Bottom Search Section] Searches for the minimum bottom M.

[Adjacent peak search section] Adjacent peaks I and O
Search.

[Load leveling processing unit] Table 8 shows the result of task list creation by the task list generation unit.

[0123]

[Table 8]

The left list is selected in the task list selection section. The task o is selected by the moving task selecting unit, and the task moving unit moves the task o to the right by one step according to the determination of the movement possibility by the task moving determining unit. The task list updating unit sets a rightward movement impossible flag for the task o, and returns the processing pointer to the head of the list.

The landslide determination section compares the minimum bottom M after the movement of the task o with the landslide target value (6 points). The minimum bottom M after the movement of task o reaches the target with 6 points, and the minimum bottom M
The processing for is ended. FIG. 3H shows this state. The number of processes 2 is stored for M in the bottom list not to be processed.

The similar processing in the leveling degree evaluation unit and the end determination unit after processing for the minimum bottom M returns to the bottom search unit and repeats the same processing.

[Bottom Search Section] In the initial setting conditions in Table 1, the maximum number of times of processing for the same bottom is defined as an end determination condition. In this embodiment, a maximum of 2
Times. As a result of the above-described series of load leveling processes, the bottom is not stored in the non-target bottom list or is stored in the non-target bottom list but the number of times of processing has not reached the maximum value. Since there is no bottom, the processing after the adjacent peak search is passed without being processed, and the end determination unit determines whether the leveling index calculated by the leveling degree evaluation unit has reached the target or the number of times of bottom search. Has reached the default maximum number of searches,
Alternatively, when there is no target bottom, the bottom leveling process ends. In FIG. 3 (h), only the bottom which has not reached the average value of 6 points, which is the landslide target value, is Q, and the number of leveling processes for the minimum bottom Q has reached the specified value. The end determination condition that no bottom exists is met, and the bottom leveling process ends.

[Case of TYPE-B] FIG. 4 is an explanatory diagram for explaining processing when TYPE-B is selected as the task list generation type. 4 (a) to 4 (e)
Shows a piled-up graph of the tasks in the planning period (B to X) in the middle of the load leveling process, and the tasks (a to ii) constituting the respective peaks. The hatched portions of the tasks in the figure show the results of the task movement due to the landslide.

TYPE-B is a method for observing the task start order by moving not only tasks contributing to load leveling but also tasks existing in the middle of tasks between peaks adjacent to the minimum bottom and left and right. It is. In this case, when the task for forming the minimum bottom is not moved (TYPE-B) and when it is moved (TYPE-B ')
The method of generating the task list and the method of moving the tasks are slightly different, which will be described later.

FIG. 4A shows an initial stacking result in the initial task stacking section. As a stacking method, stacking is performed in the order of task start time. However, even when stacking is performed in the order of task completion time, the same processing may be performed below.

The processing when TYPE-B is selected as the task list generation type will be described below with reference to FIG. Note that the description of portions overlapping with TYPE-A is simplified.

[Bottom Search Section] In the piled-up graph of FIG. 4A, the load not described in the non-target bottom list or described in the non-target bottom list but not reaching the specified number of times of processing is the minimum. Search for the bottom (minimum bottom). In the first bottom search, since the non-target bottom list is initialized at the initial setting stage, all bottoms are targeted. Here, as a result of the search, the minimum bottom R is searched.

[Adjacent Peak Search Unit] The adjacent peaks on the left and right of the minimum bottom R searched by the bottom search unit are searched. Here, as a result of the search, adjacent peaks N and T are searched.

[Load leveling processing section] The task list generation section differs from TYPE-A in that the task start order is obeyed. Therefore, the internal load leveling of the task between the minimum bottom and the left and right adjacent peaks is performed. In addition to the tasks contributing to the conversion, the tasks existing between the load contributing tasks and the minimum bottom are selected. The left list arranges tasks starting from those whose task work is completed and the work start time is later than the minimum bottom R, and the right list arranges tasks starting from the task start and the work start time earlier than the bottom R. .

The task list includes TYPE-B and TYPE-B.
-B ', and in TYPE-B, Table 9-1
Does not include the task bb, which is the bottom task. On the other hand, in TYPE-B ', as shown in Table 9-2,
Bottom task bb is included.

[0136]

[Table 9]

In both TYPE-B and TYPE-B ', the task list selecting section selects the left list by giving priority to the task list having the higher peak. The task cc is selected by the moving task selecting unit, and the task moving unit moves the task cc one step to the right in the task moving determining unit based on the determination that the task can be moved. In TYPE-B, the task list updating unit sets a rightward non-movable flag for task cc, and moves the processing pointer from task cc to aa. In TYPE-B ', task cc
, A rightward movement impossible flag is set, and the processing pointer is moved from task cc to bb.

The landslide determination section compares the minimum bottom R after the movement of the task cc with the landslide target value. Here, it is assumed that the landslide target value is 6 points which is the average value of the pile amount within the planning period.

Since the minimum bottom R after the movement of the task cc has not reached the target at 2 points, the processing pointer is confirmed. Since the processing pointer is not at the top, the processing returns to the moving task selection unit.

[Load leveling processing unit] In the moving task selection unit, the task aa is selected in TYPE-B.
Here, a description will be given assuming that a constraint condition that the inversion of the start time of (1) of the task movement determination unit, which is one of the initial setting conditions shown in Table 1, is not allowed. In this case, the task aa does not move due to the improper movement determination by the task movement determination unit. This is because the start time of aa is reversed with respect to bb for the task bb (because the task bb cannot be moved) by moving aa by one step, which is contrary to the above-described constraint.

On the other hand, in the case of TYPE-B ', task b
b is selected. As the task bb moves one step, the starting order does not reverse, so the task moving unit executes the task movement. In the following description, the case where TYPE-B is selected will be described. However, the case where TYPE-B 'is selected can be considered in the same manner, and the description is omitted.

The task list updating unit sets a rightward immovable flag for the task aa, and returns the processing pointer to the head of the list. Since the task aa cannot be moved, the next task is selected, but since the processing pointer is at the highest position,
The process returns to the task list selection section.

In the task list selection unit, the unprocessed tasks for the left task are confirmed. At this point, there are no more tasks that can be processed in the left list, so the right list is selected. For the tasks dd, ee, ff, and gg, the processes in the sequentially moving task selecting unit, the task moving judging unit, the task moving unit, the task list updating unit, and the landslide judging unit are repeated.

The minimum bottom R after the movement of the task gg reaches the target at 6 points, and FIG. 4B shows this state. R is stored in the non-target bottom list, and the number of processing times 1 is also stored. As a result of the leveling degree evaluation and the end determination after the leveling process for the minimum bottom R, the process returns to the bottom search unit, and the same process is repeated.

[Bottom search section] Searches for the minimum bottom Q.

[Adjacent peak search section] Adjacent peaks N and S
Search.

[Load leveling processor] Table 10 shows the results of task list creation by the task list generator.

[0148]

[Table 10]

Here, since the task v is a task constituting the left peak N, it may correspond to the left list. However, even if v is moved, it does not contribute to the increase in the load on the minimum bottom Q and does not move further. Does not cause the task start time to be reversed, and is therefore removed from the list. In addition,
Z and y included in the left list do not contribute to the increase in the load of the minimum bottom Q, but are added to the list in order to prevent the task start time from being reversed.

The left list is selected in the task list selection section. For the tasks z, y, x, and w, the processing in the moving task selecting unit, the task moving determining unit, the task moving unit, the task list updating unit, and the landslide determining unit are repeated.

As a result of moving tasks z, y, x, and w, the load on M decreases from 6 points to 2 points, and falls below the leveling target value of 6 points.
In addition, a bottom that is lower than the minimum bottom currently being processed has occurred. In this description, such a movement is allowed in the initial setting. As a matter of course, if the task movement is set to be restricted in the initial condition, the task movement determination unit determines that the task cannot be moved.

The minimum bottom Q after the movement of the task w is 5
Since the target has not been reached at the point, the processing pointer is confirmed. Since the processing pointer is at the top, the process returns to the task list selection unit. The left task list is checked for unprocessed tasks. At this point, there are no more tasks that can be processed in the left list, and the process moves to the right list. However, as shown in Table 10, there is no task corresponding to the right list. The processing for the bottom Q ends. FIG. 4C shows this state. Q is stored in the non-target bottom list, and the processing number 1 is also stored.

As a result of the leveling degree evaluation and the end determination after the leveling processing for the minimum bottom Q, the process returns to the bottom search section and the same processing is repeated.

[Bottom search section] Searches for the minimum bottom M.

[Adjacent Peak Search Section] Adjacent Peaks I and P
Search.

[Load leveling processor] Table 11 shows the result of task list creation by the task list generator.

[0157]

[Table 11]

Here, s, r, q included in the left list
Is not added to the load on the minimum bottom M, but is added to the list to prevent the task start time from being reversed.
Since the tasks w, x, y, z, and cc in the right list have already moved one step to the right, they can move two steps to the left.

The left list is selected in the task list selection section. For tasks t, s, r, q, p, o,
The processes in the task selecting unit, the task moving determining unit, the task moving unit, the task list updating unit, and the landslide determining unit are repeated.

The minimum bottom M after the movement of the task o is 5
Since the target has not been reached at the point, the processing pointer is confirmed. Since the processing pointer is at the top, the process returns to the task list selection unit. An unprocessed task in the left task list is checked. At this point, there are no more tasks that can be processed in the left list, and the right list is selected in the task list selection unit.

The task w is selected by the moving task selecting section, and the task moving section moves the task w to the left by one step according to the judgment of the moving possibility by the task moving determining section. The task list updating unit moves the processing pointer from task w to x.

The minimum bottom M after the movement of the task w is 6
At the point, the target is achieved, and the processing for the minimum bottom M is completed. FIG. 4D shows this state.

M is stored in the non-target bottom list, and the number of times of processing 1 is also stored. As a result of the leveling degree evaluation and the end determination after the leveling process for the minimum bottom M, the process returns to the bottom search unit and repeats the same process.

[Bottom search section] Searches for the minimum bottom Q. The bottom Q is selected for the second time, but is selected again because the maximum number of times of processing for the same bottom is set to 2 in the initial setting.

[Adjacent peak search section] Adjacent peaks N and S
Search.

[Load leveling processor] Table 12 shows the result of task list creation by the task list generator.

[0167]

[Table 12]

Here, the task x in the left list is a task forming the minimum bottom Q, and is not included in the initial condition setting.

The left list is selected in the task list selection section. For the task w, the processing is sequentially performed by the moving task selecting unit, the task moving determining unit, the task moving unit, the task list updating unit, and the landslide determining unit.

The minimum bottom Q after the movement of the task w is 5
Since the target has not been reached at the point, the processing pointer is confirmed. Since the processing pointer is at the top, the process returns to the task list selection unit. The left task list is checked for unprocessed tasks. At this point, there are no more tasks that can be processed in the left list, and the process moves to the right list. However, as shown in Table 12, there is no task corresponding to the right list. The processing for the bottom Q ends. FIG. 4E shows this state. It is assumed that the number of times of bottom Q processing of the non-target bottom list is two.

As a result of the leveling degree evaluation and the end determination after the leveling process for the minimum bottom Q, the process returns to the bottom search section and the same process is repeated.

Similarly, the minimum bottom M is searched (second time), and as a result of the landslide, there is no bottom to be processed, and the bottom leveling process ends.

[Case of TYPE-C] FIG. 5 is an explanatory diagram for explaining processing when TYPE-C is selected as the task list generation type. FIG. 5 (a) to (d)
Shows a piled-up graph of the tasks in the planning period (B to X) in the middle of the load leveling process, and the tasks (a to ii) constituting the respective peaks. The hatched portions of the tasks in the figure show the results of the task movement due to the landslide.

TYPE-C is a method of moving all tasks existing between the minimum bottom and the peak.

In this method, since the affected task range is large, the entire mountain landslide can be performed. Note that whether or not the task forming the bottom is included in the task list is determined by the initial setting condition, so that the processing of TYPE-C ′ including the bottom task is naturally possible. Therefore, the description is omitted.

In the following description, the processing content in setting the initial condition that the task forming the bottom is not included in the task list will be described. Further, in this case, the task is moved even if the minimum bottom load exceeds the mountain collapse target value.

FIG. 5A shows an initial stacking result in the initial task stacking section. As a stacking method, stacking is performed in the order of task start time. However, even when stacking is performed in the order of task completion time, the same processing may be performed below.

The processing when TYPE-C is selected as the task list generation type will be described below with reference to FIG. Note that the part different from TYPE-A, B is the task list generation part, and the processing of the other parts is the same.
The description of the overlapping parts is simplified.

[Bottom Search Section] In the pile graph of FIG. 5 (a), a bottom (minimum bottom) in which the load is equal to or less than the target value and the load is minimum is searched. As a result of the search, the minimum bottom R is searched.

[Adjacent Peak Search Unit] The adjacent peaks on the left and right of the minimum bottom R searched by the bottom search unit are searched. Here, as a result of the search, adjacent peaks N and T are searched.

[Load leveling processor] The task list generator generates a task list by TYPE-C. Here, in task list creation in TYPE-C, all tasks existing between the minimum bottom and the peak are selected except for the task forming the bottom. Table 13 shows the results of task list creation.

[0182]

[Table 13]

In the task movement judging section, even when the load on the minimum bottom R exceeds the target value due to the landslide,
All corresponding tasks are moved (set as initial conditions). Further, it is assumed that “replacement of task start order” is permitted. FIG. 5B illustrates a state after the left and right tasks are moved by the task moving unit.

After the movement, the task list updating unit sets a rightward movement impossible flag for the tasks on the left list, and sets a leftward movement impossible flag for the tasks on the right list. R is stored in the bottom list that is not targeted,
The processing number 1 is also stored.

After the leveling degree evaluation and the end determination for the minimum bottom R searched first, the process returns to the bottom search section and the same processing is repeated.

[Bottom search section] Searches for the minimum bottom M.

[Adjacent peak search section] Adjacent peaks I and O
Search.

[Load leveling processing unit] Table 14 shows the results of task list creation by the task list generation unit.

[0189]

[Table 14]

Here, w, x, y, x, a in the right list
Since a and cc have moved one step to the right, the number of steps that can be moved to the left is two.

In the task transfer section, for all tasks on the left and right lists, the task on the left list is one step to the right, the task on the right list is one step to the left for w, x, y, z, bb, and aa , Cc are moved to the left by two steps. FIG. 5C shows this state.

After the movement, the task list updating unit sets a movement impossible flag for the task that cannot be moved. M is stored in the non-target bottom list, and the processing count 1 is also stored.

After the leveling degree evaluation and the end determination for the minimum bottom M, the process returns to the bottom search section and the same processing is repeated.

[Bottom Search Section] Search for the minimum bottom Q (second selection).

[Adjacent peak search section] Adjacent peaks M and S
Search.

[Load leveling processing unit] Table 15 shows the result of task list creation by the task list generation unit.

[0197]

[Table 15]

In the task moving section, all the tasks cc, aa, z, y, x, w, u in the left list are moved one step to the right. FIG. 5D shows this state.

The number of times of processing 1 is stored for Q. Upon the determination of the end of the minimum bottom Q, the process returns to the bottom search unit and repeats the same processing.

[Bottom Search Section] Searches for the minimum bottom Q.

[Adjacent peak search section] Adjacent peaks N and S
Search.

[Load leveling processing unit] As a result of creating the task list in the task list generation unit, there are no tasks that can be moved in both the left list and the right list. This is because z, y, and v that can correspond to the left list have the rightward movement impossible flag set, and dd, ee, ff, gg, and h that can correspond to the right list.
h and ii are because the leftward movement impossible flag is set.

Since there is no corresponding task, the hill-slide processing for the minimum bottom Q ends. The number of processing 2 is stored for Q. After the leveling evaluation and the end determination of the minimum bottom Q, the process returns to the bottom search unit and repeats the same processing.

[Bottom Search Unit] Since there is no target bottom, the process proceeds to the end determination unit.

[End Determination Unit] The end determination unit determines whether the number of bottom searches has reached the initially set number, if there is no bottom that is lower than the average value of the bottom search results, or if all bottoms have been processed as a result of the bottom search. Has reached the initially set number of times, the leveling process in the bottom leveling unit ends.

After the processing in the bottom leveling section is completed, it is determined whether or not a series of processing in the peak leveling section is to be performed. In addition, when ending the leveling process, the process proceeds to the plan output unit.

An example in which a series of processing is performed in the peak leveling unit will be described below. FIG. 6 is an explanatory diagram for explaining the series of processes. FIG.
(G) is within the planned period during the peak load leveling process (B
7A to 7C illustrate task pile graphs and tasks (a to ii) constituting each mountain. The hatched portions of the tasks in the figure show the results of the task movement due to the landslide.

FIG. 6A is an example showing a processing result in the above-mentioned bottom leveling unit. Note that the processing results in the bottom leveling unit include TYPE-A, TYPE-B, and TYPE-B.
It is also possible to use the results of PE-B ', TYPE-C, TYPE-C'.

In the description of the present embodiment, it is assumed that the processing is performed on the following assumptions. However, the present invention is not limited to this, and the processing may be performed by another method based on the initial setting in the initial setting unit. It is possible. Focusing on the max peak, the landslide collapses to the adjacent bottom. Only tasks that directly contribute to Max Peak load leveling are moved. If a load equal to or greater than Max Peak occurs due to task movement, the task is not moved. The task movement from the peak to the bottom is performed sequentially from the task closer to the bottom toward the peak. If there are a plurality of max peaks, a peak close to the previous peak is selected. The landslide target value is the average value of the pile load between the peak and the left and right adjacent bottoms.

[Max Peak Search Section] Searches for Max Peak J.

[Adjacent Bottom Search Unit] Adjacent Bottom G, M
Search. In addition, in the case of the same level, the far bottom is selected, but it may be close.

[Peak load leveling processor] Table 1 shows the result of task list creation by the peak leveling task list generator.
6 is shown.

[0213]

[Table 16]

The reason why task j is not included in the task list is that it has been calculated in advance that within the movable range of task J, movement that directly contributes to peak J load leveling cannot be performed.

In the task list generation unit, whether the task forming the peak is included in the left or right task list is determined by the position of the center of gravity of each task. For example, the position of the center of gravity of the task (tasks g, h, k, l,
If (s, t black circle position) is on the left side of the peak, it is put in the left task list. On the other hand, if the position of the center of gravity of the task is on the right side of the peak, it is put in the right task list. As another method, distribution may be performed according to the center position of the reference workable period described in FIG. Additionally, as an alternative,
From the earliest arrival procedure, 1/2 may be assigned to the left task list, and the remaining 1/2 to the right task list. Similarly, when a plurality of load center-of-gravity positions exist at the same position as the peak, the half of the arrival procedure is sorted to the left task list, and the remaining half to the right task list. Here, only the tasks that contribute to the reduction of the peak load in the movable range are selected. However, a method of selecting all the tasks between the bottom peaks and a selection method that does not reverse the task start time can be used.

The peak leveling landslide target setting section obtains the average value of the load in the period between the left and right bottoms, and sets it as the landslide target value. Here, the total load during the period from the bottom G to the bottom M is 59 points, and since the period is a unit of 7 management hours, the average value is 8.4. Become.

In the peak leveling task list selecting section, whether to select the left list or the right list is determined for the same reason as in the bottom leveling section. here,
Select the left list.

The task g in the peak leveling moving task selecting section
Is selected, and the peak leveling task movement determining unit moves task g to the left by one step in the peak leveling task moving unit. The peak leveling task list updating unit sets a leftward non-movable flag for task g, and moves the processing pointer from task g to h.

[0219] The peak leveling landslide determination section compares the maximum peak J after the movement of the task g with the landslide target value. Max peak J after moving task g
Since the target is not reached at 12 points, the processing pointer is confirmed. Since the processing pointer is not at the highest position, the processing returns to the peak leveling movement task selecting unit.

[Peak load leveling processing unit] The task h is selected in the peak leveling moving task selecting unit. The task h is moved one step to the left in the peak leveling task moving section by the peak leveling task moving determining section determining that movement is possible. In the peak leveling task list update section,
A rightward movement impossible flag is set for task h, and the processing pointer is moved from task h to k.

The peak leveling landslide judging section compares the maximum peak J after the movement of the task h with the landslide target value (9 points). Since the maximum peak J after the movement of the task h has not reached the target at 11 points, the processing pointer is confirmed. Since the processing pointer is not at the highest position, the processing returns to the peak leveling movement task selecting unit.

[Peak load leveling processing section] The task k is selected in the peak leveling moving task selecting section. The task k is moved one step to the left in the peak leveling task moving section by the peak leveling task moving determining section determining that movement is possible. In the peak leveling task list update section,
A rightward movement impossible flag is set for task k, and the processing pointer is moved from task k to l.

The peak leveling landslide determination section compares the maximum peak J after the movement of the task k with the landslide target value (9 points). Since the maximum peak J after the movement of the task k has not reached the target at 10 points, the processing pointer is confirmed. Since the processing pointer is not at the highest position, the processing returns to the peak leveling movement task selecting unit.

[Peak load leveling processing unit] The peak leveling moving task selecting unit selects the task l. The peak leveling task moving unit moves task 1 to the left by one step in the peak leveling task moving unit according to the determination of whether or not movement is possible. In the peak leveling task list update section,
A rightward movement impossible flag is set for task l, and the processing pointer is returned to the top of the list.

The peak leveling landslide determination section compares the maximum peak J after the movement of the task 1 with the landslide target value (9 points). The maximum peak J after the movement of the task 1 reaches the target at 9 points, and the processing for the maximum peak J ends. FIG. 6 (b)
Represents this state. Here, the task on the left list is not moved since the target value has been reached by the movement of the task on the right list. J is stored in the non-target peak list, and the processing count 1 is also stored.

Upon evaluation of the degree of peak leveling for the maximum peak J and determination of the end, the process returns to the maximum peak search section and the same processing is repeated.

[Max Peak Search Section] Searches for Max Peak I.

[Adjacent Bottom Search Unit] Adjacent Bottom G, J
Search. In the case of the same level, the far side is selected.

[Peak load leveling processor] Table 1 shows the result of task list creation by the peak leveling task list generator.
FIG.

[0230]

[Table 17]

The peak leveling landslide target setting section calculates the landslide target value in the same manner as described above. Here, the bottom G
The total load during the period from to the bottom J is 32 points, and since the period is in units of four management hours, the average value is 8, and the landslide target value is 8 points.

In the peak leveling task list selecting section, the right list is selected because there is no movable task in the left list.

The task r in the peak leveling moving task selecting section
Is selected, and the task r is moved to the right by one step in the peak leveling task moving section by the peak leveling task moving determining section determining that movement is possible. The peak leveling task list updating unit sets a rightward non-movable flag for task r, and moves the processing pointer from task r to q.

The peak leveling landslide judging section compares the maximum peak I after the movement of the task r with the landslide target value. The landslide target value here is 8 points, and the maximum peak J after the movement of the task r is 10 points, which is below the target, so that the processing pointer is confirmed.
Since the processing pointer is not at the highest position, the processing returns to the peak leveling movement task selecting unit.

[Peak load leveling processing unit] The peak leveling moving task selecting unit selects the task q. The peak leveling task moving unit moves the task q right by one step in the peak leveling task moving unit according to the determination of whether or not the movement is possible. In the peak leveling task list update section,
The rightward movement impossible flag is set for task q, and the processing pointer is moved from task q to p.

The peak leveling landslide determination section compares the peak value I after the movement of the task q with the landslide target value (8 points). Since the maximum peak J after the movement of the task has not reached the target at 9 points, the processing pointer is confirmed. Since the processing pointer is not at the highest position, the processing returns to the peak leveling movement task selecting unit.

[Peak load leveling processing unit] The task p is selected in the peak leveling moving task selecting unit. The task p is moved one step to the right in the peak leveling task moving section by the peak leveling task moving determining section determining that movement is possible. In the peak leveling task list update section,
A rightward movement impossible flag is set for task p, and the processing pointer is moved from task p to o.

The peak leveling landslide judging section compares the maximum peak I after the movement of the task p with the landslide target value (8 points). After the task is moved, Max Peak J achieves the target at 8 points, and the processing for Max Peak I ends. FIG. 6 (c)
Represents this state. I is stored in the non-target peak list, and the number of times of processing 1 is also stored.

Evaluation of peak leveling degree of Max Peak I,
When the end is determined, the process returns to the maximum peak search unit and the same processing is repeated.

[Max Peak Search Section] Searches for Max Peak K.

[Adjacent Bottom Search Unit] Adjacent Bottom G, M
Search. In the case of the same level, the far side is selected.

[Peak load leveling processor] Table 1 shows the result of task list creation by the peak leveling task list generator.
FIG.

[0243]

[Table 18]

The peak leveling landslide target setting section calculates the landslide target value in the same manner as described above. Here, the bottom G
The total load during the period from to the bottom M is 54 points, and the average value is 7.7 since the period is in units of 7 management hours.
And the landslide target value is 8 points.

In the peak leveling task list selecting section, the left list is selected. In the peak leveling task selection, tasks n and j are selected and moved sequentially in the same manner as described above.

In the peak leveling landslide judgment section, the landslide target value (8 points) is reached for the max peak K by the movement of n and j, and the processing for the max peak K ends. FIG. 6D shows this state. K is stored in the non-target peak list, and the processing count 1 is also stored.

Evaluation of peak leveling degree of Max Peak K,
When the end is determined, the process returns to the maximum peak search unit and the same processing is repeated.

[Max Peak Search Section] Searches for Max Peaks E, J and N. If there are multiple max peaks, select the one near the previous peak,
Select J.

[Adjacent Bottom Search Unit] Adjacent bottom G, M
Search.

[Peak load leveling processor] Table 1 shows the result of task list creation by the peak leveling task list generator.
It is shown in FIG.

[0251]

[Table 19]

In the peak leveling landslide target setting section, a landslide target value is calculated in the same manner as described above. Here, the bottom G
The total load during the period from to the bottom M is 53 points, and the average value is 7.6 since the period is a unit of 7 management hours.
And the landslide target value is 8 points.

The peak leveling task list selecting section selects the right list because there is no target task in the left list.

In the peak leveling movement task selection section, the task t
choose. It does not move due to the improper movement determination in the peak leveling task movement determination unit. This is because, by moving t, a load equal to or greater than the maximum peak is generated in N. Since t cannot be moved, the next task s is selected. The peak leveling task moving unit moves the task s one step to the right in the peak leveling task moving unit according to the determination of the movement possibility. The peak leveling task list update unit sets a rightward movement impossible flag for the task s, and returns the processing pointer to the top of the list.

The peak leveling landslide judging section compares the maximum peak J after the movement of the task s with the landslide target value. The target value here is 8 points,
The maximum peak J after the movement of the task s reaches the target at 8 points, and the processing for the maximum peak J ends. FIG. 6E shows this state.
J is stored in the non-target peak list, and the processing count 2 is also stored.

Evaluation of peak leveling degree of Max Peak J,
When the end is determined, the process returns to the maximum peak search unit and the same processing is repeated.

[Max Peak Search Section] Searches for Max Peaks E and N. If there are multiple max peaks, select N to select the one near the previous peak.

[Adjacent bottom search section] Adjacent bottom G, Q
Search.

[Peak load leveling processor] Table 2 shows the result of task list creation by the peak leveling task list generator.
0 is shown.

[0260]

[Table 20]

In the peak leveling landslide target setting section, a landslide target value is calculated in the same manner as described above. Here, the bottom G
The total sum of the loads from the period from to to the bottom Q is 84 points, and since the period is 11 management time units, the average value is 7.
6 and the landslide target value is 8 points. In the peak leveling task list selection section, the left list is selected.

The task p in the peak leveling movement task selecting section
choose. It does not move due to the improper movement determination in the peak leveling task movement determination unit. This is because a load equal to or greater than the maximum peak is generated in I by moving p. Since there are no more tasks that can be processed in the left list, the processing returns to the task list selection unit.

[Peak load leveling processor] The right list is selected in the task list selector. The task cc is selected by the moving task selecting unit, and the task moving determining unit determines that the task cc can be moved.
Move step. The task list updating unit sets a leftward movement impossible flag for the task cc, and moves the processing pointer from the task cc to aa.

The peak leveling landslide determination section compares the maximum peak N after the movement of the task cc with the landslide target value. The target value here is 8 points, and the maximum peak J after the movement of the task cc reaches the target at 8 points, and the processing for the maximum peak N ends. FIG. 6F shows this state. N is stored in the non-target peak list, and
And memorize it.

Evaluation of peak leveling degree of Max Peak N,
When the end is determined, the process returns to the maximum peak search unit and the same processing is repeated.

[Max Peak Search Section] Searches for Max Peak E.

[Adjacent Bottom Search Unit] Adjacent Bottom B, G
Search.

[Peak load leveling processing unit] Table 2 shows the results of task list creation by the peak leveling task list generation unit.
It is shown in FIG.

[0269]

[Table 21]

In the peak leveling landslide target setting section, a landslide target value is calculated in the same manner as described above. Here, bottom B
The total of the loads during the period from to the bottom G is 40 points, and the average value is 6.7 because the period is a unit of 6 management time units.
And the landslide target value is 7 points.

In the peak leveling task list selecting section, the left list is selected. The task d is selected by the peak leveling moving task selecting unit, and the peak leveling task moving unit moves the task d one step to the left by the peak leveling task moving determining unit determining that the task d can be moved. The peak leveling task list updating unit sets a leftward movement impossible flag for the task d, and moves the processing pointer to the top.

The peak leveling landslide determination section compares the peak value E after the movement of the task d with the landslide target value. The target value here is 7 points,
Since the maximum peak J after the movement of the task d has not reached the target at 8 points, the processing pointer is confirmed. Since there are no more tasks that can be processed in the left list, the processing returns to the task list selection unit.

[Peak load leveling processing unit] The right list is selected in the task list selection unit. All tasks on the right list cannot be moved because moving them causes a load greater than Max Peak.

Since there is no corresponding task, the hill-slide processing for the maximum peak E ends. FIG. 6 (g)
Represents this state. E is stored in the non-target peak list, and the processing count 1 is also stored.

Evaluation of peak leveling degree of Max Peak E,
When the end is determined, the process returns to the maximum peak search unit and the same processing is repeated.

By repeating the above processing, even when the peak leveling degree evaluation unit does not reach the peak leveling target, the target peak gradually exists under the constraint conditions set by the initial setting unit. And the process proceeds to the peak leveling end determination unit.

[Peak leveling end determination section] The peak leveling end determination section determines whether not only the peak leveling target has been achieved, but also the peak search frequency has reached the initially set frequency, or all the results of the peak search. If the number of peak processing times reaches the initially set number, the peak leveling process in the peak leveling unit ends. In addition, when ending the leveling process, the process proceeds to the plan output unit.

[Plan output unit] The plan output unit outputs the leveling schedule within the planning period as a process plan.
In this case, the process plan to be output may be output for all processes of all products within the planning period, or may be output for an arbitrary period for each specific product or each specific process.

[0279]

As described above, according to the present invention, even in heavy industry that manufactures a wide variety of products based on order-based production, not only the overload at the time of schedule planning but also the shortage of the load so as not to cause an excessive production capacity can be solved. In addition, a scheduling device capable of leveling the load is provided.

[Brief description of the drawings]

FIG. 1 is a block diagram according to the present invention.

FIG. 2 is a diagram showing a reference workable period for a product including four work processes.

FIG. 3 is an explanatory diagram for describing processing when TYPE-A is selected as a task list generation type.

FIG. 4 is an explanatory diagram illustrating a process when TYPE-B is selected as a task list generation type.

FIG. 5 is an explanatory diagram for describing processing when TYPE-C is selected as a task list generation type.

FIG. 6 is an explanatory diagram for describing a series of processes in a peak leveling unit.

Claims (6)

[Claims] 1. An initial task stacking section for performing an initial stacking of tasks, which are operations of each step necessary for manufacturing a product, and a portion having the lowest load within a planning period stacked by the initial task stacking section. A bottom search unit for searching for a minimum bottom; an adjacent peak search unit for searching for a peak adjacent to the minimum bottom; A load leveling processing unit that performs leveling, a leveling degree evaluation unit that evaluates the load leveling processing result in the load leveling processing unit, and performs a process termination determination based on the result of the leveling evaluation A scheduling device comprising: an end determination unit that performs the operation; and a plan output unit that outputs a load leveling process result. 2. A task level generator for selecting only tasks contributing to an increase in the minimum bottom load and generating a set of task lists for each of the right and left sides, and setting a hill-slip target value for the minimum bottom. A landslide target setting unit, a task list selection unit for selecting which of the left and right task lists is to be selected, a moving task selection unit for selecting one of the movable candidate tasks in the task list, and task moving. A task movement determining unit that determines whether the task can be violated and whether the task movement violates the constraint, a task moving unit that moves the task for a predetermined time when the task can be moved, and a task based on a result of the movement. A task list updating unit for updating the list, and determining whether or not the result of the task movement has reached a satisfactory result of the landslide on the bottom. Scheduling device according to claim 1, characterized in that it comprises a pile breaking judging unit for. 3. A task list generation unit selects not only tasks that contribute to the increase in the load of the minimum bottom but also tasks that do not directly contribute to load leveling and exist between the tasks and the minimum bottom, and a set of tasks for each of the left and right. 3. The scheduling apparatus according to claim 2, wherein the generation of the list enables load leveling such that the inversion of the work start order and / or the inversion of the work completion order falls within the constraint range. 4. A task list generating section selects all tasks existing between the minimum bottom and adjacent peaks on the left and right sides, generates a set of task lists for each of the left and right sides, and reverses the work start order and / or completes the work. 3. The scheduling apparatus according to claim 2, wherein the load leveling is performed such that the reversal of the order falls within the restricted range, thereby enabling the load leveling to expand the leveling target range. 5. A maximum peak search section for searching for a peak having the highest load within a planning period to be searched, an adjacent bottom search section for searching for a bottom adjacent to the peak, and a section between the adjacent bottom and the peak. A peak load leveling processor for moving a task existing in the above to the adjacent bottom by a predetermined method to perform load leveling, and a peak leveling level for evaluating a load leveling process result in the peak load leveling processor. 5. The scheduling according to claim 1, further comprising: an evaluation unit; and a peak leveling end determination unit that determines whether to end or continue processing based on the evaluation performed by the peak leveling degree evaluation unit. 6. apparatus. 6. A peak leveling task list generating section for selecting a task contributing to a load reduction of a peak peak and generating a set of task lists for each of right and left, a peak leveling task list generating section; Peak leveling landslide target setting part to set the value, peak leveling task list selecting part to select left or right task list to be selected, and peak to select one movable candidate task in task list A leveling movement task selection unit, a peak leveling task movement determination unit that determines whether the task can be moved and whether the task movement can violate the constraint condition, and a case where the task can be moved. A peak leveling task moving unit that moves tasks for a predetermined time, and a peak leveling task that updates the task list based on the result of the movement The scheduling according to claim 5, further comprising a skrist update unit and a peak leveling landslide determination unit that determines whether a result of the task movement as a result of the landslide for the peak has reached a satisfactory result. apparatus.