JPH0475836A - System and method for planning assembly order - Google Patents

Abstract

PURPOSE:To provide a satisfactory degree of level at a practicable speed by assembling re-arrangement by a vacant load replacing means for re-arrangement. CONSTITUTION:A vacant load replacing method is a method wherein the vacant load rear element of a notice option forms a moving candidature, an optimum distance position forms a replacement candidature, and leveling is progressed by effecting replacement. To improve the level of leveling with enough high speed property, a vacant load replacing method is employed as far as a vacant load is generated, and at a point of time when a vacant load is rendered ineffective, leveling is effected by a peak load replacing method once. Thereafter, when the vacant load is rendered effective, the vacant load replacing method is executed again. The two methods are employed through mixture of the ways.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an assembly sequence planning technique for planning the order of input models to an assembly line in a mixed production system that assembles product groups to which various options or additions are added. relates to methods and systems for implementing the technology.

[Conventional technology]

In the above-described production system, the daily production number for each model, that is, the production number for each added option, is determined a certain day before assembly, for example, the day before. -The task of determining the order in which 8-minute production models should be sent to the assembly line is called assembly sequence planning.The most basic idea in this assembly sequence planning is leveling, and the man-hour load in each process It is important to stabilize the process by reducing fluctuations as much as possible.

Conventionally, this plan creation has been done manually by manually determining the quality of the plan and revising defective areas that cause load concentration through trial and error. In order to avoid this, automation using computers is also being considered.

For example, in order to find the optimal solution for assembly sequence planning, there is an OR-like modeling method that calculates line balance using standard man-hours. However, the problem of assembly sequence planning has complex constraints, and the conditional expressions of the model can only be solved by calculating all sequence combinations and determining the optimal solution. This method is impractical because the number of combinations is increasing explosively and this calculation is impossible within a realistic amount of time even with a high-speed computer.

Therefore, instead of finding a solution with mathematical rigor, it may be possible to create an assembly order based on some kind of rule and try to obtain a satisfactory solution. In this case, in order to have a computer perform the sorting process at a speed sufficient for practical use, the sorting process was previously performed based on the personal know-how of staff in the manufacturing department. It is essential to establish a new sorting method derived from analysis.

[Invention or problem to be solved]

An object of the present invention is to realize a sorting process in which the obtained assembly order can be performed with practically sufficient speed and accuracy with respect to the attention option in a technology for planning the order of models to be introduced into an assembly line in a mixed production system. It is.

[Means to solve the problem]

In order to solve the above problems, the assembly order planning method according to the present invention includes a sorting step of rearranging the assembly order in order to evenly flow products to which attention-grabbing options are added onto the assembly line; There is a levelness evaluation step that calculates the levelness of the assembly order that has been rearranged with respect to It consists of a comparison step to determine whether the changed order is satisfactory, and the sorting step performs leveling by exchanging the peak load center and the widest bay cant load center regarding the attention option. The peak load replacement step to be continued,
The system includes a bay cant road replacement step in which the rear element of the bay cant road of the attention option is set as a movement candidate, the element at the optimal interval position is set as a replacement candidate, and leveling is promoted by performing the replacement.

Furthermore, the assembly order planning system according to the present invention includes a rearrangement processing unit that rearranges the assembly order in order to evenly distribute products to be added to the assembly line based on the options to be noticed; The levelness evaluation section calculates the levelness of the assembly order, and the levelness calculated by the levelness evaluation section is compared with the preset standard value of the corresponding option to determine if the rearranged order is satisfactory. It is composed of a comparison section that determines whether
The rearrangement processing unit is configured to replace the center of the peak load for the attention option with the center of the widest bay cant load, thereby promoting equalization, and the rear of the bay cant load of the attention option. It is composed of a bay cant load replacement means that uses an element as a candidate for movement, uses an optimal interval position as a candidate for replacement, and promotes leveling by performing replacement.

[For production]

A peak load replacement means that rearranges elements in the most concentrated part with elements in the most empty part, and a peak load replacement means that rearranges elements in the most concentrated part with elements in the most empty part. Baycant load switching means for rearranging the options is selectively activated according to a predetermined rule, and the array of the attention options is leveled.

〔Effect of the invention〕

When sorting, if you simply use a peak load swapping method that replaces the elements in the most concentrated part with the elements in the least occupied part, it will be possible that when one part improves, the other parts become worse. Situations in which the solution oscillates due to unsatisfactory levelness often occur, making it impossible to find a satisfactory solution. However, by combining this sorting with the sorting using the bayant load replacement means, the oscillation of the solution as described above can be suppressed. A satisfactory levelness can be obtained at a practical speed.

[Other Features] As a preferred combination control of the peak load switching means and the bay cant load switching means, in a preferred embodiment of the present invention, as long as the bay cant load is generated, the bay cant load switching means is used to control the bay cant load. It is proposed that when the load disappears, leveling is performed once by the peak load replacement means, and then, if a bay cant load occurs, the bay cant load replacement means is executed again. With this kind of control, without losing the high speed of the peak load replacement method,
It becomes possible to improve the leveling level.

〔Example〕

The overall configuration of the assembly sequence planning system according to the present invention is shown in FIG. The input unit 1 receives the types of products to be assembled for one day, the presence or absence of various options, and their number from the host computer via communication means or magnetic or optical storage means, and sends these data to the I10 interface. Send to 2. This interface 2 has the function of converting this data into a format readable by the central processing unit of the system. Conversely, it also has a function of sending the results derived by the central processing unit to an output unit 3 such as a printer or CRT.

This central processing mechanism is composed of a rearrangement processing section 4, a level evaluation section 5, a comparison section 6, a reference value changing section 7, and a control section 8 that centrally controls these.
Basically, the assembly order is rearranged based on a rearrangement rule that is necessary for distributing products having noteworthy options evenly onto the assembly line without making them consecutive as much as possible. The level evaluation unit 5 calculates the level of the rearranged assembly order of the attention options based on rules that will be explained in detail later, and calculates a peak load evaluation value indicating the concentration of the attention options. a peak load evaluation section that calculates a dispersion evaluation value indicating the degree of dispersion of the option of interest; and a levelness calculation unit that determines a weighting coefficient for each and calculates the levelness of the option of interest based on the weighting coefficient. The comparison section 6 compares the levelness calculated by the levelness evaluation section with a preset reference value of the corresponding option and determines whether the rearranged order is satisfactory. Further, the reference value changing unit 7 has a function of lowering the reference value of a specific option when reordering is not successful due to mutual interference between options, which is used in special cases. The control section 8 controls the above-mentioned sorting processing section 4, levelness evaluation section 5, comparison section 6, and reference value changing section 7, and based on the daily production information input to this system, Deriving a satisfactory assembly order by paying attention to options in order of priority.

Next, the general flow of processing of the assembly order planning system according to the present invention will be explained using the flowchart shown in FIG.

In step 10 (hereinafter referred to as #lO), the input section 1
The data showing the products to be assembled for one day sent through the system is loaded into the work area of this system. #1
5, a reference value, that is, a sorting passing score is set for each option given to the input product to be assembled. In #20, it is determined whether there is a notable option for rearranging the ranking, and if YES, the process proceeds to #25. In #25, the sorting processing unit 4 sorts the options of interest. This sorting is carried out under sorting constraints such as not changing the predetermined order of preferred options.
This is done using the techniques described later. In #30, the level evaluation unit 5 evaluates the order sorted in #25.
calculates its levelness. In #35, the calculated degree of levelness is compared with the reference value, and if the degree of levelness exceeds the reference value, the process jumps to #20 and checks whether there is an option to be noted for next sorting. If the levelness is below the reference value, proceed to #40. In #40, we unfix the order of products that have higher-priority options that were noticed before the option we are currently paying attention to, that is, we relax the constraint on the upper-level options, and Reordering is performed as long as the evaluation value of the option exceeds the corresponding reference value. #45 and #50
Then, the levelness is calculated in the same way as #30 and #35, and the levelness is compared with the standard value. If the levelness exceeds the standard value, jump to #20. If the levelness is below the standard value, jump to #20. Proceed to #60 and inquire whether to change the reference point and rearrange again. If YES, jump to #15 and re-set the reference value. If no, #20
Jump to. If No in #20, that is, if there are no more noteworthy options, the final result is output in #65 and the process ends.

Next, we will explain the sorting method and level calculation method in the sorting process, but first we will define a few words and phrases used for this purpose.

When products with options are evenly distributed, the number of units in the spacing condition is called the "optimal spacing." In other words, optimal spacing = (total number of units produced - number of units with options) /
Number of units with options However, when the number of units produced of products with options is 0,
The optimum interval is set to 0, and when the number of products with options produced exceeds 50% of the total number, the optimum interval is set to the number of consecutive units even if they are optimally distributed. If the optimal interval is not an integer, the value rounded up is called the optimal wide interval, and the value rounded down is called the optimal narrow interval.

When focusing on a predetermined option in the assembly order of products for one day, if there is continuity, there is an option that includes the part with the largest number of consecutive units, and if there is no continuity, there is an option that includes the part with the smallest number of consecutive units. Each part is called "peak load".

If there are multiple peak loads, that portion is called a "peak load set."

If the interval with a predetermined option is larger than the optimal wide interval in the assembly order of products for one day, that part is called a "bay canto load" and the set is called a "bay can load set."

In the present invention, the replacement processing method uses either a peak load replacement method or a bay cant load replacement method. This is a method that promotes leveling by replacing the elements.The Bay Cant Road replacement method uses the Bay Cant Road rear element of the noted option as a candidate for movement, and optimally considers the interval position as a candidate for replacement. This is a method that promotes leveling up as the situation progresses.

In order to improve the leveling level with sufficient speed, use the bay cant load replacement method as long as the bay cant load is occurring, and once the bay cant load disappears, use the peak load replacement method once. After leveling, if a bay cant load occurs, it is desirable to use a mixture of both methods, such as executing the bay cant load replacement method again.

The level evaluation unit calculates the level of the noteworthy option for a given assembly order using the following formula: Level = (W*G+(1-W)*H)*10.

Here, G is the dispersion evaluation value, When y=<x, (X: set of actual intervals, Y.

When G=Σ(Y/X)/number of production units with option Y>X, G=Σ(X/Y)/number of production units with option H is the peak load evaluation value, When is the interval, H = Peak load interval number of units / Optimum narrow interval (all cases where the value exceeds 1) When the peak load is continuous, H = O W is the weight of the dispersion evaluation value, S When =>1, (S: optimal spacing or optimal wide spacing) W=l/S When S<1, W=1.

Next, the actual sorting and calculation of its levelness are shown in Figure 3 (
This will be explained using the examples shown in a) to (r).

In each figure, the horizontal number row represents one product, and each number, ie, "1" or "0", represents whether it has each option. For example, the product indicated by "0100" does not have the first option, has the second option, and does not have the second or third option. As is clear from the figure, 10 products are planned in this example,
The number of options is 5. Note that when determining this assembly order, the previous assembly order must be taken into account, but here it is assumed that there were three optional intervals in the previous time. In addition, the standard value of each option,
In other words, the passing score for sorting is 60 starting from the first option.
,50140,30,20 points.

First step of sorting process...Fig. 3(a) to Fig. 3(b) First, attention is paid to the first option. As mentioned above, there are three gaps from the previous publication, and there is also one gap at the beginning of the current data, making a total of four intervals. Since the optimal interval for the first option is 3, a bay cant load relocation is performed to fill in the excess space.

In other words, the first option has no upper-level constraints, so the first option (the product indicated by the horizontal row of numbers at the top) and the second option (
(products indicated by the horizontal row of numbers in the second row).

2nd step...Figures 3(b) to 3(c) Regarding the first option, the interval between the 3rd and 8th levels is 4.
Therefore, the Beycant Road is further rearranged, and the 7th and 8th stages are swapped.

Third step...Figure 3 (C) to Figure 3 (cl)
For the first option, peak load relocation is performed because three or more intervals of the optimal interval are gone.

First, the part with the highest density is searched, but here, since the top row and the third row are one interval apart, it is moved to the top row or the part with the lowest density. At this point, the top row will be moved, but if we assume that this is the only row left, there will be 5 intervals from the previous time. It would be a good idea to replace it with one of these, but the previous day's data has already been published and cannot be replaced. For this reason, 8
- Aim at the 9th and 10th rows at 3 intervals, and these will be candidates for replacement, but replacing them with the 8th or 9th row will not improve the situation, so the 10th row will be replaced with the top row.

4th step...Figures 3(d) to 3(e) Considering the previous data, there are 5 intervals between the 3rd and the 3rd, so bayant load rearrangement is performed and the The upper and third rows are swapped, and the interval between them is 3.

5th step...Figures 3(e) to 3(f) Since there is a gap of 5 between the top tier and the 7th tier, bay cant load rearrangement is performed, and the 5th and 7th tier The stages can be swapped.

6th step...Figures 3(f) to 3(g) Since there are 4 gaps between the 5th and 10th stages, the bay cant load is rearranged, and the 9th stage and The 10th stage can be replaced.

By the above sorting processing from the first step to the sixth step, the sorting regarding the first option is completed. Here, the levelness of this order is calculated as follows using the above-mentioned formula.

As is clear from Figure 4, at this stage, considering the results from the previous day, the interval for option 1 is 3.
The set of intervals is (3, 3, 3). Since the last interval cannot be determined until the order of the next day is determined, it is excluded from the object of calculating the degree of levelness here.

The number of production units is 10, and 3 of them have the first option, so the set of optimal intervals is (3, 2, 2). However, in this set, the numerical values of its elements are arranged in descending order.

First, the variation evaluation value is G = (3/3 + 2/3 + 2/3) / 3 = 0.778 The peak load evaluation value is, since the peak load interval is 3, H = 3/2 = 1.5 Since it is greater than 1 H=1.

Furthermore, the weight W of the dispersion degree evaluation value is such that the optimal wide interval is 3.
Therefore, W=1/3=0.333 Therefore, levelness=(0,333*0.778+(1-0,333
)*1)*100=93...The passing score for the first option is 60 points, rounding up the decimal point, so we will start the replacement process for the next second option, but first we will change the number of points for the second option at this stage. The degree of levelness is determined with reference to FIG. 4 as follows.

G= (3/3+1/2+1/2)/3=0.667 H=1/2=0.5 W=1/3=0.333 Levelness=(0,333*0.667+(1-0 ,333
) * 0.5) * 100 = 56 Therefore, the passing score of the second option is higher than 60 points, but since the second option has not been rearranged yet, in order to find a better order, here We will perform the replacement processing routine once.

7th step...Figure 3 (g) to Figure 3 (h) Since there are 4 gaps between the 2nd and 7th stages, bay cant load rearrangement is performed, and the 6th stage and The 7th stage can be replaced.

Eighth step...FIG. 3(h) to FIG. 3(i) Regarding the second option, since there is no longer an interval with an optimum interval of 3 or more, peak load relocation is performed. Since the top row and the second row are continuous, we will eliminate this continuity.

There are three intervals between the top row and the previous issue and also below the second row, but here we will move the top row. There is a gap of 4 between the 7th row and the 10th row, but even if you replace the 7th row, a continuity will occur again, so the 8th, 9th, and 10th are candidates for replacement, so the 9th row is the center. or will be replaced. In other words, the top row and the ninth row are swapped. As for the second option, due to the restrictions of the first option, which is a higher-order option, the sorting for the second option is completed by the sorting processing in the seventh and eighth steps. The levelness of this order is 94 when calculated as described above.

Since the passing score for the second option is 50 points, replacement processing for the next option begins. Next, let's focus on the third option.

9th step...Figures 3 (i) to 3 (j) Considering the previous issue, there are 5 intervals up to the 3rd stage, but the 1st and 2nd stages have upper options. Since there is no replacement, peak load relocation is performed from the beginning. The third and fourth stages are continuous, and the interval above the third stage is wider than the interval below the fourth stage, so the third stage will be moved. There are 4 intervals between the 7th and 10th rows, but if you swap the 7th rows, they will be continuous, so the 8th, 9th,
The 10th candidate will be replaced, and the 9th candidate will be replaced. That is, the third and ninth
The numbers are swapped.

10th step...Fig. 3(j) to Fig. 3(k) Considering the previous issue, 6 intervals have occurred up to the 4th stage, and bay cant load rearrangement is performed. But the I
Since the 3rd and 2nd options cannot be swapped because they have higher-level options, the 3rd and 4th options are swapped.

Eleventh step: Since there is no longer an interval part with an optimum interval of 3 or more for the third option from FIG. 3(k) to FIG. 3, peak load rearrangement is performed. Since there is an interval between the 6th and 8th options and the 6th option has a higher option, we will move the 8th option. If the 8th row has no option, there will be 4 intervals from the 7th row to the 10th row, but the 10th row will be replaced due to the restriction of the upper option. That is, the 8th and 10th
The numbers are swapped.

Regarding the third option, the first option, which is the upper option,
Due to the restriction of two options, the sorting for the third option is completed by the sorting processing from the ninth step to the eleventh step. The levelness of this order is 92.

Since the passing score for the second option is 40 points, replacement processing regarding the next option begins. Next, let's focus on the fourth option.

12th step...Figure 3 (1) to Figure 3 (m) There is a gap of 7 between the 1st stage and the 9th stage, but due to upper option restrictions, the bay cant road is rearranged. can't do it. For this reason, peak load relocation is performed. Since the 9th and 10th are consecutive and the 9th has a higher option, the 9th will be moved. In the interval from the second to the eighth option, the third option that is not subject to the constraints of the upper option is selected as the replacement target. Therefore, the 10th and 3rd positions are exchanged.

Regarding the 4th option, the 1st option, which is the upper option,
Due to the constraints of the 2nd and 3rd options, the sorting process in this step completes the sorting regarding the fourth option. The levelness of this order is 53.

Since the passing score for the fourth option is 30 points, replacement processing regarding the next option begins. Next, let's focus on the fifth option.

13th step...Figure 3(m) to Figure 3(n) Since there are 9 intervals from the previous issue to the 7th stage, bay cant load rearrangement is performed.

Since the first, second, and third stages cannot be swapped due to upper-level option restrictions, the fourth and seventh stages are swapped.

Regarding the fifth option, due to the restrictions of the upper option, it is not possible to perform further processing to improve the order in addition to the sorting processing in this step, but the levelness of this order is 12 points, which is lower than the passing score of 20 points. ing. For this reason,
As shown below, the sorting process is performed by relaxing the constraints on the upper options.

14th step... FIG. 3(n) to FIG. 3(o) Since six intervals have occurred from the previous issue to the fourth stage, constraint relaxation Baycant load rearrangement is performed. If you replace the 4th with the top row and set the interval from the previous issue to 3, the 1st option will be continuous and the average score or passing score for the 1st option will be divided, so the 4th and 2nd row Eyes can be replaced. In this case, the order of the second option will be disrupted, but the average score will be 50 points or more, which is the passing score.

15th step...Figure 3 (0) to Figure 3 (p) The 9th position is further moved, but if you select the 3rd to 7th items as replacement targets, you will receive a passing score in any case. Since it does not meet the criteria, the 8th option, which satisfies all the passing scores for the upper option, is selected, and the 9th option and the 8th option are swapped. Its levelness is 52.

The sorting process regarding the fifth option is now complete,
Since there is no next option, the sorting process is finally completed.

In the example just described, in the sorting process that focuses on each option, the average value is calculated after the bay cant load or peak load rearrangement is carried out to the end. It is also possible to calculate the average value each time the option is selected, and when the average value exceeds the passing score, perform the sorting process for the immediately next option.

Further, in the example explained using FIG. 3, the number of machines produced is 10, but in an actual factory, the number of machines produced may range from 100 to 1000 or more. In this type of production order planning system, as the number of units increases, the processing time increases exponentially. In order to avoid this, in the present invention, the number of production units for one day is divided into several blocks, an independent assembly order is determined for each block, and the blocks are connected together to assemble one day's worth of units. Decide on the order.

In addition, when distributing products with options evenly into blocks, a simple and effective method for dispersing the options of the product as much as possible is to change the presence or absence of options for each product as an option as shown in Figure 3. A sequence of numbers expressed in order of priority, for example "10010"
It is proposed to consider `` as a binary number and allocate it to each block in the order of its numerical value.Also, in this case, products with priority delivery dates can be forcibly allocated to the first block.

[Brief explanation of the drawing]

The drawings show an embodiment of the assembly order planning system and assembly order planning method according to the present invention, in which Fig. 1 is an overall configuration diagram, Fig. 2 is a flowchart showing the processing procedure, and Figs. 3 (a) to (p).
) is an explanatory diagram showing the state of sorting, and FIG. 4 is an explanatory diagram for calculating levelness. (1)...Input section, (3)...Output section, (4)...Sort processing section, (5)...
...Evenness evaluation section, (51) ...Peak load evaluation section, (52) ...Dispersion degree evaluation section, (53) ...Evenness calculation section , ...Comparison section, ...Reference value change section

Claims (1)

[Claims] 1. An assembly sequence planning system for planning the order of input models to an assembly line in a mixed production system for assembling product groups to which various options are added, the system comprising: a sorting processing unit 4 that rearranges the assembly order in order to distribute the products equally on the assembly line, and a levelness evaluation unit 5 that calculates the levelness of the rearranged assembly order regarding the noted option. a comparison unit 6 that compares the levelness calculated by the levelness evaluation unit with a preset reference value of the corresponding option and determines whether the sorted order is satisfactory; and the above-mentioned sorting processing unit 4;
In the system, which is composed of a level evaluation section 5 and a control section 8 that controls a comparison section 6, the sorting processing section 4 sorts the center of the peak load and the center of the widest bay cant load regarding the attention option. By replacing the peak load with a peak load replacement means that levels the load by replacing the peak load, with the Bay Cant Road backward element of the featured option as a candidate for movement, and with the element at the optimal spacing position as a candidate for replacement, and by performing replacement. An assembly sequence planning system characterized by comprising: means for exchanging bay cant loads to promote leveling. 2. The control unit 8 activates the bay cant load replacement means as long as the bay cant load is generated;
When there is no bay cant load, the peak load replacement means is leveled once, and if a bay cant load has occurred, the rearrangement processing unit 4 is controlled to activate the bay cant load replacement means again. The assembly order planning system according to claim 1, characterized in that: 3. If the levelness calculated after the end of the sorting process focusing on lower-order options is lower than the reference value, the control unit 8 selects products with higher-order options as long as the levelness of the higher-order options exceeds the corresponding reference value. 3. The assembly sequence planning system according to claim 1, further comprising controlling the sorting processing unit 4 to perform sorting processing again as sortable objects. 4. An assembly sequence planning method for planning the order of input models to an assembly line in a mixed production system that assembles a group of products to which various options are added; A sorting step in which the assembly order is rearranged in order to distribute it evenly on the assembly line, a levelness evaluation step in which the levelness is calculated for the rearranged assembly order regarding the attention option, and a levelness evaluation unit calculates the levelness. and a comparing step of comparing the obtained levelness level with a preset reference value of the relevant option to determine whether the sorted order is satisfactory, wherein the sorting step includes a A peak load replacement step that promotes leveling by exchanging the center of the road and the center of the widest bay cant road, and the element at the optimal spacing position by using the rear element of the bay cant road of the featured option as a movement candidate. An assembly sequence planning method comprising: a bay cant load replacement step in which a candidate is used as a replacement candidate, and leveling is promoted through replacement.