JP2004287540A - Method and device for grouping facility - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and a device for grouping facilities whereby the optimum layout planning of facilities such as production facilities can be quickly made even in the case of a very large-scale and complex production system, etc. <P>SOLUTION: A predetermined number of individuals n having a group number g as genetic information are made to undergo generational changes by a predetermined number of generations by an operation using genetic algorithm to create the final generation (Nr +1). Based on the genetic information of the individuals n among the final generation who have the best goodness of fit to predetermined evaluation criteria, the facilities are allocated to different groups. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

【００５８】
ここに、
Ｔｉ：部品Ａ（ｉ）についてグループ間のワークの移動を要する回数
Ｎｉ：部品Ａの個数に応じた重み
とされる。
【００５９】
（ニ）下記式２により適合度ＦＶ（ｎ）を算出する。
【００６０】
ＦＶ（ｎ）＝１／ＴＧ（ｎ） （２）
【００６１】
すなわち、重み付け移動回数ＴＧ（ｎ）の逆数として適合度ＦＶ（ｎ）が算出される。
【００６２】
ステップＳ３：操作
Ｓ３−１：エリート保存
次世代の個体集団を発生させる前に、当該世代の個体集団の中で最も適合度ＦＶ（ｎ）の高い所定数、ここでは１体の個体ｎをエリートとして無条件に次世代に残す。
【００６３】
Ｓ３−２：交叉
交叉とは、比例した確率により選択された複数の親の遺伝子的情報を受け継ぐ新しい個体を生成させることをいう。つまり、両親の遺伝子を受け継いだ子を生成する。
【００６４】
Ｓ３−２−１：交叉させる個体（親個体）の選択法
ランキング選択とルーレット選択とを組み合わせた手法によって親個体を選択する。
【００６５】
（あ）評価が高い順、つまり適合度ＦＶ（ｎ）が高い順に個体ｎをランク（順位）付けし、順位が高いものほど選択確率が高くなるように下記式３を用いて各個体ｎの選択重みｆ（ｎ）を設定する。すなわち、ここまでの手順がランキング選択の手法に対応する。
【００６６】
ｆ（ｎ）＝Ｒｌａｓｔ−Ｒ（ｎ） （３）
ここに、
Ｒ（ｎ）：個体ｎの順位、Ｒｌａｓｔ：最下位の個体ｎの順位
とされる。
【００６７】
（い）下記式４を用いて各個体ｎの選択確率Ｐ（ｎ）を選択重みｆ（ｎ）に比例した値に設定する。
【００６８】
【数２】

【００６９】
（う）各個体ｎに順番ｉ（ｉ＝１、２、…）を付ける。乱数Ｒｎｄ（ただし、Ｒｎｄ： ０≦Ｒｎｄ＜１）を発生させ、発生させた乱数Ｒｎｄについて下記式５の条件を満足する個体ｎの順番ｉを求め、その順番ｉの個体ｎを親個体として選択する。前掲（い）およびこの（う）の手順がルーレット選択の手法に対応する。なお、各個体ｎの順番ｉの付け方は任意である。
【００７０】
【数３】

【００７１】
例えば、４つの個体ｎに個体１、個体２、個体３、個体４と順番ｉを付け、このとき、ｉ番目の個体ｎの選択確率Ｐ（ｉ）がそれぞれ次のようであるとする。
【００７２】
（Ｐ（１）、Ｐ（２）、Ｐ（３）、Ｐ（４））＝（０．２、０．１、０．４、０．３）
このとき、乱数Ｒｎｄの値と選択される個体ｎとの関係は下記のようになる。
【００７３】
（ａ）０≦Ｒｎｄ＜０．２であれば、個体１が選択される。
【００７４】
（ｂ）０．２≦Ｒｎｄ＜０．３（＝０．２＋０．１）であれば、個体２が選択される。
【００７５】
（ｃ）０．３≦Ｒｎｄ＜０．７（＝０．３＋０．４）であれば、個体３が選択される。
【００７６】
（ｄ）０．７≦Ｒｎｄ＜１．０（＝０．７＋０．３）であれば、個体４が選択される。
【００７７】
（え）以上の手順（あ）〜（う）を繰り返し、交叉させる２体の親個体を選択する。ただし、１体目の親として選択された個体は２体目の選択候補からは除外し、２体の親が同一の個体になることがないようにする。
【００７８】
Ｓ３−２−２：交叉箇所
親個体の遺伝子配列Ｕの１箇所をランダムに選んで交叉させる。具体的には、下記手順（か）〜（く）による。
【００７９】
（か）整数ｊ（ｊ：ｊ＝０、１、…、（Ｎｍ−１））をランダムに発生させ、各親個体の遺伝子配列Ｕのｊ番目の遺伝子Ｃ（ｊ）と（ｊ＋１）番目の遺伝子Ｃ（ｊ＋１）との間を交叉点とする。
【００８０】
図５に、ランダムに発生された整数ｊと、各親個体（親１、親２）の遺伝子配列Ｕ上の交叉点の位置との対応を示す。
【００８１】
（き）各親個体の遺伝子配列Ｕをこの交叉点において２つの部分に分離する。以下、図４および図５の例では、図の左側の遺伝子Ｃの番号の若い方の部分を前半部分、右側の番号の大きい方の部分を後半部分と呼ぶ。
【００８２】
（く）図６に示すように、前半部分と後半部分とに分離された（同図（ａ）参照）各親個体（親１、親２）の後半部分を入れ替える（同図（ｂ）参照）ようにして次世代の２個体（子１、子２）を生成する。なお、以上の説明より明らかなように整数ｊが数「０」であるときは遺伝子Ｃの入れ替えは行われず、交叉は行われない。このときは親の２個体がそのまま次世代に移る。
【００８３】
Ｓ３−２：突然変異
生成された子の個体（次世代の個体）を一定の確率で、遺伝子配列Ｕの１つの遺伝子Ｃをランダムに選択し、それをランダムに変更するようにして、突然変異させる。具体的には以下の手順による。
【００８４】
（さ）０以上１以下の乱数を発生させ、その乱数の値があらかじめ定められた確率（突然変異確率と称する）Ｐｍ以下であれば、下記手順（し）〜（せ）を実行する。突然変異確率Ｐｍは通常は微少な値に設定される。設例では、Ｐｍ＝０．０３に設定した。
【００８５】
（し）１以上Ｎｍ以下の整数ｋをランダムに発生させ、遺伝子配列Ｕのｋ番目の遺伝子Ｃ（ｋ）を選択する。
【００８６】
（す）１以上Ｎｇ以下の整数ｑをランダムに発生させる。
【００８７】
（せ）整数ｑを前掲手順（し）で選択した遺伝子Ｃ（ｋ）の値（グループ番号ｇ）として代入する。
【００８８】
以上のステップＳ３の手順（エリート保存、交叉、突然変異）によりＮｕ個の個体を生成し、これにより次世代の個体集団を生成する。
ステップＳ４：前掲ステップＳ２およびステップＳ３の手順による世代交代を回数Ｎｒ繰り返して最終世代を生成する。
【００８９】
ここで、回数Ｎｒは最優秀個体の適合度が向上しなくなるような回数を実験的に求めるようにして設定する。設例では回数Ｎｒは３００回とした。
【００９０】
ステップＳ５：解導出
前掲ステップＳ２およびステップＳ３の手順を回数Ｎｒ繰り返すことによって第（Ｎｒ＋１）世代（最終世代）の集団を発生させた後、その中で最も適合度ＦＶ（ｎ）の高い個体ｎを解として選択する。
【００９１】
例えば、選択された個体ｎの遺伝子配列Ｕが（Ｃ１、Ｃ２、Ｃ３、Ｃ４、Ｃ５、Ｃ６）＝（１、３、２、２、１、３）であれば、遺伝子Ｃ１、Ｃ５に対応する生産設備Ｍ１、Ｍ５は、グループ番号１のグループに配置され、遺伝子Ｃ３、Ｃ４に対応する生産設備Ｍ３、Ｍ４はグループ番号２のグループに配置され、遺伝子Ｃ２、Ｃ６に対応する産設備Ｍ２、Ｍ６は、グループ番号３のグループに配置される。
【００９２】
図７に、前掲図２の設例に対して実施形態の生産設備グルーピング処理を実施した結果である最適解を表形式で示す。表中の値「１」および値「０」は、それぞれの部品Ａ（Ａ１、Ａ２、…、Ａ２９）について生産設備Ｍ（Ｍ１、Ｍ２、…、Ｍ２９）で加工が行われるか否かを示している。例えば部品Ａ１については生産設備Ｍ２、Ｍ３、Ｍ４の欄が値「１」であり、これらの生産設備Ｍ２、Ｍ３、Ｍ４のみで加工されることが分かる。
【００９３】
また、第１グループ、第２グループ、…はそれぞれグループ番号１、グループ番号２、…の各グループＧと対応している。
【００９４】
したがって、値「１」の生産設備Ｍが複数のグループＧに跨って存在している部品Ａ、図７では部品Ａ１３、Ａ２７、Ａ２８については、ワークＷをグループ間で移動させる必要があることになる。
【００９５】
しかしながら、この解における重み付け移動回数ＴＧは、部品Ａ１３について２（回）×１（個）、部品Ａ２７について１（回）×２（個）および部品Ａ２８について１（回）×２（個）の総計６（回）でしかなく、優秀な結果が得られているものといえる。
【００９６】
このように、実施形態の生産設備のグルーピング処理は、
【００９７】
（１）個々の生産設備Ｍに対応する遺伝子的情報、つまり遺伝子Ｃを順序付けて１列に並べたものを遺伝子配列Ｕとして設定し、
【００９８】
（２）各遺伝子Ｃにその値として生産設備Ｍの各グループＧのグループ番号ｇを所定の原則の下でランダムに格納するようにして数Ｎｕの初期解（個体の集団）を生成し、
【００９９】
（３）グループ間において各部品Ａのワークを移動する移動回数に基づいて、各個体ｎの適合度ＦＶ（ｎ）を計算するとともに、これを利用しつつ所定の遺伝的アルゴリズムによる操作（エリート保存、交叉、突然変異）を実施するようにして次世代の集団を生成し、
【０１００】
（４）この操作を所定回数Ｎｒ繰り返し実施することによって適合度の高い個体ｎを選択するようにして、ワークＷのグループＧ間の移動回数（重み付け移動回数）の少ない最適な解を導出する
ものとされる。
【０１０１】
したがって、製造される部品の種類が数百から数千種類存在するような製品の生産システム、例えばガスタービンエンジンの生産システムにおいても、各生産設備のグルーピングを最適にするといった、従来、熟練した設計者であっても解決が非常に困難であった大規模・複雑な最適化問題の解を短時間の内に導出することが可能となる。
【０１０２】
以上、本発明を実施形態に基づいて説明してきたが、本発明はかかる実施形態のみに限定されるものではなく、種々改変が可能である。例えば、実施形態では適合度をワークの設備間の移動回数を基準に算出されているが、適合度を算出する基準はワークの設備間の移動回数に限定されるものではなく、グルーピングをなす設備に応じて適宜設定でき、例えば設備の類似に基づいて適合度を算出するようにしてもよい。
【０１０３】
また、実施形態では設備は生産設備とされていが、本発明が適用される設備は生産設備に限定されるものではなく、各種設備とすることができる。
【０１０４】
【発明の効果】
以上詳述したように、本発明によれば多数の設備から構成されるシステムにおける設備を最適なグループへの振り分けが迅速になし得るという優れた効果が得られる。
【図面の簡単な説明】
【図１】本発明の一実施形態に係る生産設備のグルーピング方法が適用される生産設備のグルーピング装置の概略構成を示すブロック図である。
【図２】本生産設備のグルーピング方法の理解を容易にするために用いた設例における計画基礎情報を示すテーブル図である。
【図３】本生産設備のグルーピング方法に用いた遺伝的アルゴリズムの所要手順を示す概略フローチャートである。
【図４】同遺伝的アルゴリズムにおける遺伝子配列の構造を示した模式図である。
【図５】同遺伝的アルゴリズムによる操作で交叉点を設定するための原理を示す模式図である。
【図６】同遺伝的アルゴリズムによる操作としての交叉の態様を示す模式図である。
【図７】図２の設例に本生産設備のグルーピング方法を実施して得られた最適解を示すテーブル図である。
【符号の説明】
Ａ 部品
Ｃ 遺伝子
Ｋ 作成装置
Ｍ 生産設備
Ｎｍ 生産設備総数
Ｎｕ 個体数
Ｕ 遺伝子配列
１ 処理部
２ 入力部
３ 出力部
４ データベース[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a facility grouping method and a grouping apparatus. More specifically, the present invention relates to an arrangement plan for appropriately arranging facilities such as production facilities, and particularly to a grouping method and a grouping apparatus for facilities suitable for creating an arrangement plan of facilities such as a production facility for a product composed of a huge number of parts. .
[0002]
[Prior art]
In recent years, in various manufacturing industries, the transition from the conventional low-mix high-volume production type to the high-mix low-volume production type has progressed, and depending on the product, there are cases in which hundreds to thousands of parts need to be manufactured.
[0003]
The production system for producing such products has a very large scale. For example, as in a production system for a gas turbine engine, a production system may be formed from a group of parts processing factories. . Production equipment such as a processing machine for manufacturing and processing each part is not provided in one-to-one correspondence with each part, but one production equipment is used for manufacturing a plurality of parts, and Usually, one part is processed in accordance with a predetermined processing procedure (processing step) using a plurality of production facilities.
[0004]
For this reason, if the grouping of production equipment and the allocation of production equipment to each factory in the above example are not appropriate, during the process of manufacturing each part in accordance with the processing procedure, the work that is the material is transferred to each factory. , The production time is prolonged, and the production cost is increased. Therefore, it is very important to properly arrange production equipment.
[0005]
However, as the types of parts that need to be manufactured increase and the production system becomes larger and more complex, the difficulty of creating an optimal arrangement plan increases exponentially. For this reason, in the grouping method of production equipment that has been performed based on the experience and intuition of experienced designers, it is no longer possible to take appropriate measures due to changes in production forms in various manufacturing fields as described above. .
[0006]
[Problems to be solved by the invention]
SUMMARY OF THE INVENTION The present invention has been made in view of the problems of the related art, and is an equipment capable of quickly making an optimal arrangement plan of equipment such as production equipment even in a system such as a very large-scale and complicated production system. It is an object of the present invention to provide a grouping method and a grouping device.
[0007]
[Means for Solving the Problems]
A first embodiment of the present invention is a facility grouping method for allocating a required number of facilities to a required number of groups in accordance with a processing procedure of a material to be processed, wherein the group number is a numerical data string in a one-dimensional array; A predetermined number of generations are changed by a genetic algorithm for a predetermined number of individuals having the length of the numerical data sequence as genetic information having a length corresponding to the total number of facilities to generate a final generation, The present invention relates to a grouping method characterized by allocating each facility to each group on the basis of genetic information of an individual having an optimum degree of conformity to a predetermined evaluation criterion in the last generation.
[0008]
In the first embodiment of the present invention, it is preferable that the initial value of the numerical data sequence is set at random.
[0009]
Further, in the first embodiment of the present invention, the operation by the genetic algorithm is an elite preservation in which individuals having the highest fitness in the generation are retained in a predetermined number of next generations, and the genetic information is inherited from a plurality of parents. It preferably contains crossovers that generate individuals and mutations that change some of the genetic information.
[0010]
Further, in the first embodiment of the present invention, it is preferable that the location where the genetic information of the parent crosses is set at random.
[0011]
Further, in the first embodiment of the present invention, the predetermined evaluation criterion is obtained by integrating the product of the number of times each of the workpieces moves between groups and the number of the workpieces with respect to all the workpieces. It is preferable that the number of times of weight movement is set.
[0012]
Furthermore, in the first embodiment of the present invention, it is preferable that an upper limit value and / or a lower limit value are provided for the number of facilities allocated to each group.
[0013]
Furthermore, in the first embodiment of the present invention, it is preferable that the number of groups is limited.
[0014]
A second embodiment of the present invention is a grouping device for equipment that divides a required number of equipment into a required number of groups in accordance with a processing procedure of a material to be processed, wherein the group number is a one-dimensional array of numerical data strings, and A predetermined number of generations are changed by a genetic algorithm for a predetermined number of individuals having the length of the numerical data sequence as genetic information having a length corresponding to the total number of facilities to generate a final generation, The present invention relates to a facility grouping apparatus configured to sort each facility into each group based on genetic information of individuals whose suitability to a predetermined evaluation criterion in the last generation is optimal.
[0015]
In the second embodiment of the present invention, it is preferable that the initial value of the numerical data sequence is set at random.
[0016]
Further, in the second embodiment of the present invention, the operation by the genetic algorithm is an elite preservation that leaves the highest-fit individual of the generation in a predetermined number of next generations, and inherits the genetic information from a plurality of parents. It preferably comprises crossovers that produce individuals and mutations that change part of the genetic information.
[0017]
Furthermore, in the second embodiment of the present invention, it is preferable that the location where the genetic information of the parent crosses is set at random.
[0018]
Further, in the second embodiment of the present invention, the predetermined evaluation criterion is obtained by accumulating the product of the number of times each workpiece moves between groups and the number of the workpieces for all the workpieces. It is preferable that the number of times of weight movement is set.
[0019]
Furthermore, in the second embodiment of the present invention, it is preferable that an upper limit value and / or a lower limit value are provided for the number of facilities allocated to each group.
[0020]
Furthermore, in the second embodiment of the present invention, it is preferable that the number of groups is limited.
[0021]
A third embodiment of the present invention is a computer-readable program for allocating a required number of facilities to a required number of groups in accordance with a processing procedure of a material to be processed, wherein the group number is a one-dimensional array of numerical data strings. Generating a predetermined number of first-generation individuals having genetic data of each individual, the number of which is set to be a length corresponding to the total number of facilities, and the length of the numerical data sequence is set to a length corresponding to the total number of facilities; A procedure for calculating the degree of conformity of the obtained individual to a predetermined evaluation criterion, a procedure for generating a final generation by performing a generation alternation on the individual who has reached the predetermined criterion by an operation using a genetic algorithm, A procedure of calculating the fitness of the generation individual for a predetermined evaluation criterion, and a procedure of selecting an individual having an optimal fitness for the predetermined evaluation criterion among the final generation individuals, Performing a grouping of equipment based on the genetic information of the selected individual.
[0022]
In the third embodiment of the present invention, the operation by the genetic algorithm includes elite preservation in which the highest number of individuals having the highest fitness level in the relevant generation are left for a predetermined number of generations, and individuals inheriting the genetic information from a plurality of parents. It preferably contains the resulting crossovers and mutations that change part of the genetic information.
[0023]
In the third embodiment of the present invention, it is preferable to include a procedure for randomly setting a place where the parent's genetic information is crossed.
[0024]
Further, in the third embodiment of the present invention, the predetermined evaluation criterion is obtained by integrating the product of the number of times each workpiece moves between groups and the number of the workpieces with respect to all the workpieces. It is preferable that the number of times of weight movement is set.
[0025]
Furthermore, in the third embodiment of the present invention, it is preferable that an upper limit value and / or a lower limit value are provided for the number of facilities allocated to each group.
[0026]
Further, in the third embodiment of the present invention, it is preferable that the number of groups is limited.
[0027]
[Action]
Since the present invention is configured as described above, even in a system including a large number of facilities, each facility can be assigned to an optimal group according to a processing procedure.
[0028]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, the present invention will be described based on embodiments with reference to the accompanying drawings, but the present invention is not limited to only such embodiments.
[0029]
FIG. 1 shows a schematic configuration of a facility grouping apparatus (hereinafter simply referred to as an apparatus) applied to a facility grouping method according to an embodiment of the present invention. In this embodiment, the facility is a production facility for producing goods.
[0030]
The device K includes an electronic computer (computer) such as a personal computer, and includes a processing unit 1, an input unit 2, an output unit 3, and a database 4.
[0031]
The processing unit 1 includes a CPU (central processing unit; central processing unit) and a storage device, and executes a grouping program for performing grouping of production equipment described later. Here, the grouping of the production facilities means that all production facilities are allocated to a required number of groups.
[0032]
The input unit 2 includes information input devices such as a keyboard, a mouse, and a flexible disk reader, and various information serving as a basis for grouping (hereinafter, referred to as planning basic information), for example, a processing procedure for each part, equipment required for the processing, The characteristics and processing capacity of various facilities are input to the processing unit 1.
[0033]
The output unit 3 includes information output devices such as a display device, a printer device, and a flexible disk writing device, and outputs a grouping result of the production facilities obtained by the processing unit 1 executing the grouping of the production facilities.
[0034]
The database 4 stores the basic plan information and the grouping result in a predetermined format.
[0035]
FIG. 2 shows an example of the basic plan information. The basic plan information includes the quantity of each part (A1, A2,..., A29; also collectively referred to as part A) used for the product, the processing procedure (process), and the processing procedure (10, 20,. ., M29, which are used in the following order).
[0036]
Hereinafter, the grouping of the production facilities will be described with reference to the plan basic information of FIG. 2 as an example.
[0037]
Here, the grouping of the production equipment is planned so that the work, which is the material of each part A, is not moved between the groups of the production equipment as much as possible. That is, the number of movements of the work between the groups is used as an evaluation index (evaluation criterion), and the grouping of the production equipment is performed so as to optimize the index.
[0038]
In the embodiment, the grouping of the production facilities is synonymous with deciding in which of the factories constituting the production system the respective production facilities M are arranged.
[0039]
Here, the principle of grouping the production facilities M is as follows.
[0040]
(B) The total number of groups Ng is set in consideration of the restriction on the number of factory buildings. In the example, the group total number Ng is set to “5”.
[0041]
(B) The upper and lower limits of the number of facilities are set and the number of facilities in each group is set within the range of the upper and lower limits in which the number of facilities in each group is set so that a group with an extremely small number or a large number of production facilities M to be arranged does not appear. Make it fit. In the example, the lower limit of the number of production facilities included in one group is “2” and the upper limit is “9”. Depending on the system, only one of the upper limit and the lower limit may be provided.
[0042]
In the grouping of the production equipment, a genetic algorithm that has a reputation as a method for solving a large-scale and complex optimization problem is used for a solution for appropriately grouping the production equipment. A genetic algorithm is a method of finding the optimal solution to a problem that imitates the laws of genetics that exist in the living world, and a method of finding a better solution while genetically changing many solutions. Say.
[0043]
Hereinafter, a grouping process of production equipment using a genetic algorithm will be described with reference to the drawings.
[0044]
As shown in the flowchart of FIG. 3, the present process generates a group of individuals (chromosomes) having, as genetic information, a group number g (g = 1, 2,...) Assigned to each group of the production equipment ( Step S1), the fitness of the individual is calculated based on the number of movements of the work between the groups G of the production equipment M (Step S2), and the operation by the genetic algorithm is performed in consideration of the calculated fitness (elite storage). , Crossover, mutation, etc.) to generate a population of next-generation individuals (step S3), and perform generation alternation a sufficient number of times to generate individuals with a high degree of fitness to generate the final generation (step S4). The optimal solution, that is, the group to which each production facility belongs, is derived based on the genetic information of the individual having the optimal fitness in the final generation.
[0045]
Hereinafter, each procedure of FIG. 3 will be described in more detail. In the figure, reference numerals S1 to S5 indicate step numbers.
[0046]
Step S1: Individual occurrence
S1-1: Setting of initial value of genetic information
Genetic information of one individual (corresponding to a chromosome in a cell, which is referred to as a gene sequence U in the embodiment) has a total length Nm of production facilities (M1, M2,...) To be grouped. It is represented by a one-dimensional array of numerical data strings.
[0047]
That is, as shown in FIG. 4A, the gene sequence U is composed of elements (C1, C2,..., C (Nm) corresponding to each of the production facilities (M1, M2,... M (Nm)); Gene C) are arranged in a line in order.
[0048]
Then, as shown in FIG. 4 (b), the values 1, 2,..., Ng of the group number g are randomly assigned to the genes C1, C2,. , And generate a predetermined number Nu of initial solutions. That is, a group of individuals of the total number Nu of the first generation is generated. Also, the number of individuals in the population after the next generation is set to several Nu. However, FIG. 4B shows a case where the total group number Ng is a value “5”. That is, the group number g is one of the values “1”, “2”,..., “5”.
[0049]
When an initial solution is actually generated by a computer, an integer is randomly generated, and a value obtained by adding 1 to the remainder obtained by dividing the integer by the total group number Ng is used as a value for each of the genes C1, C2,. Each individual is generated in such a manner as to be substituted into ()).
[0050]
S1-2: Number of individuals generated
The number Nu of individuals of each generation is determined by experimentally obtaining the number of individuals that is likely to generate a solution that is considered to be the optimal solution and that does not significantly increase the processing time. In the example (in the case of 29 parts in total), the number of individuals Nu is set to 200.
[0051]
Step S2: Calculation of fitness (fitness value)
The fitness FV (n) of the individual n is calculated based on the number of times that the work needs to be moved between the groups (referred to as the number of times of movement). The detailed procedure for calculating the fitness FV (n) is as follows.
[0052]
(B) The numerical values (group number g) stored in the genes C1, C2,..., C (Nm) of the gene sequence U of the individual n for which the fitness FV (n) is to be calculated are referred to. The group to which the production facility M1, M2,..., M (Nm) corresponding to the gene C belongs is checked.
[0053]
(B) Referring to the basic plan information (FIG. 2), when the work is moved between the corresponding production equipment M in accordance with the processing order (process) for all the parts A1, A2,. Is used to calculate the number of times the workpiece is moved.
[0054]
In the example of FIG. 2, since the workpiece is moved in the order of the production equipment M3, M9, M10, M3, and M11 for the part A12, the production equipment M3, The group numbers g assigned to the genes C3, C9, C10, and C11 corresponding to M9, M10, and M11 are checked, and the group numbers g are arranged in the order of the genes C3, C9, C10, C3, and C11. The number of times of movement is calculated by checking how many times it is necessary to move between times. This procedure is executed for all parts A1, A2,..., A29.
[0055]
(C) In the individual n whose fitness is to be calculated, the weight according to the number of parts A is multiplied by the number of movements of the part A obtained in the procedure (b) described above, and the sum of the multiplication results for each part A ( TG (n) will be calculated below.
[0056]
That is, the weighted number of movements TG (n) is calculated according to the following equation 1.
[0057]
(Equation 1)

[0058]
here,
Ti: The number of times the part A (i) needs to move the work between groups
Ni: weight according to the number of parts A
It is said.
[0059]
(D) Calculate the fitness FV (n) according to the following equation (2).
[0060]
FV (n) = 1 / TG (n) (2)
[0061]
That is, the fitness FV (n) is calculated as the reciprocal of the weighted number of movements TG (n).
[0062]
Step S3: Operation
S3-1: Elite preservation
Before generating a next-generation individual population, a predetermined number having the highest fitness FV (n) in the individual population of the relevant generation, here, one individual n is unconditionally left in the next generation as an elite.
[0063]
S3-2: Crossover
Crossover refers to generating a new individual that inherits the genetic information of a plurality of parents selected with a proportional probability. In other words, they produce offspring that inherit the genes of their parents.
[0064]
S3-2-1: Method of selecting individuals to be crossed (parent individuals)
A parent individual is selected by a method combining ranking selection and roulette selection.
[0065]
(A) The individuals n are ranked (ranked) in the order of higher evaluation, that is, in the order of higher fitness FV (n), and the higher the rank, the higher the selection probability is. Set the selection weight f (n). That is, the procedures up to this point correspond to the ranking selection method.
[0066]
f (n) = Rlast-R (n) (3)
here,
R (n): rank of individual n, Rlast: rank of lowest individual n
It is said.
[0067]
(I) The selection probability P (n) of each individual n is set to a value proportional to the selection weight f (n) using Expression 4 below.
[0068]
(Equation 2)

[0069]
(C) Assign an order i (i = 1, 2,...) To each individual n. A random number Rnd (where Rnd: 0 ≦ Rnd <1) is generated, an order i of the individual n that satisfies the condition of the following Expression 5 is obtained for the generated random number Rnd, and the individual n in the order i is selected as a parent individual. I do. The steps (i) and (u) correspond to the roulette selection method. The order i of each individual n is arbitrary.
[0070]
[Equation 3]

[0071]
For example, four individuals n are given an order i in the order of individual 1, individual 2, individual 3, and individual 4. At this time, it is assumed that the selection probability P (i) of the i-th individual n is as follows.
[0072]
(P (1), P (2), P (3), P (4)) = (0.2, 0.1, 0.4, 0.3)
At this time, the relationship between the value of the random number Rnd and the selected individual n is as follows.
[0073]
(A) If 0 ≦ Rnd <0.2, the individual 1 is selected.
[0074]
(B) If 0.2 ≦ Rnd <0.3 (= 0.2 + 0.1), the individual 2 is selected.
[0075]
(C) If 0.3 ≦ Rnd <0.7 (= 0.3 + 0.4), the individual 3 is selected.
[0076]
(D) If 0.7 ≦ Rnd <1.0 (= 0.7 + 0.3), the individual 4 is selected.
[0077]
(E) The above procedures (a) to (u) are repeated to select two parent individuals to be crossed. However, the individual selected as the first parent is excluded from the second selection candidate so that the two parents do not become the same individual.
[0078]
S3-2-2: Intersection
One position in the gene sequence U of the parent individual is randomly selected and crossed over. Specifically, the following procedures (K) to (K) are performed.
[0079]
(K) Integer j (j: j = 0, 1,..., (Nm-1)) is randomly generated, and the j-th gene C (j) and the (j + 1) -th gene in the gene sequence U of each parent individual are generated. The intersection with the gene C (j + 1) is defined as an intersection.
[0080]
FIG. 5 shows the correspondence between the randomly generated integer j and the position of the intersection on the gene sequence U of each parent individual (parent 1, parent 2).
[0081]
(G) The gene sequence U of each parent individual is separated into two parts at the intersection. Hereinafter, in the examples of FIGS. 4 and 5, the lower part of the gene C on the left side of the figure is referred to as the first half, and the higher part of the right side of the figure is referred to as the second half.
[0082]
(C) As shown in FIG. 6, the latter part of each parent individual (parent 1, parent 2) separated into a first half part and a second half part (see FIG. 6 (a)) is replaced (see FIG. 6 (b)). ), Two individuals of the next generation (child 1, child 2) are generated. As is clear from the above description, when the integer j is the number “0”, the gene C is not replaced and no crossover is performed. At this time, the two parents move to the next generation.
[0083]
S3-2: Mutation
The generated offspring individual (next-generation individual) is mutated with a certain probability by randomly selecting one gene C of the gene sequence U and changing it randomly. Specifically, the following procedure is used.
[0084]
(S) A random number from 0 to 1 is generated, and if the value of the random number is equal to or less than a predetermined probability (referred to as mutation probability) Pm, the following steps (S) to (S) are executed. The mutation probability Pm is usually set to a very small value. In the example, Pm = 0.03 was set.
[0085]
(I) An integer k of 1 or more and Nm or less is randomly generated, and a k-th gene C (k) of the gene sequence U is selected.
[0086]
(S) An integer q of 1 or more and Ng or less is randomly generated.
[0087]
(Se) The integer q is substituted as the value (group number g) of the gene C (k) selected in the above procedure (i).
[0088]
Nu individuals are generated by the procedure of the above step S3 (elite preservation, crossover, mutation), thereby generating a next-generation individual population.
Step S4: The last generation is generated by repeating the number of generations Nr by the procedure of steps S2 and S3 described above.
[0089]
Here, the number of times Nr is set such that the number of times that the fitness of the best individual does not improve is experimentally obtained. In the example, the number Nr is set to 300 times.
[0090]
Step S5: Solution derivation
After generating the (Nr + 1) th generation (final generation) group by repeating the procedure of the above steps S2 and S3 Nr times, an individual n having the highest fitness FV (n) is selected as a solution. .
[0091]
For example, if the gene sequence U of the selected individual n is (C1, C2, C3, C4, C5, C6) = (1, 3, 2, 2, 1, 3), it corresponds to the genes C1, C5. The production facilities M1 and M5 are arranged in the group of group number 1, and the production facilities M3 and M4 corresponding to the genes C3 and C4 are arranged in the group of group number 2 and the production facilities M2 and M6 corresponding to the genes C2 and C6. Are arranged in the group of group number 3.
[0092]
FIG. 7 shows, in the form of a table, optimal solutions as a result of performing the production equipment grouping process of the embodiment on the example of FIG. The value “1” and the value “0” in the table indicate whether or not each part A (A1, A2,..., A29) is processed by the production facility M (M1, M2,..., M29). ing. For example, regarding the component A1, the values of the columns of the production facilities M2, M3, and M4 are “1”, and it can be seen that the processing is performed only by the production facilities M2, M3, and M4.
[0093]
Also, the first group, the second group,... Correspond to the respective groups G of group number 1, group number 2,.
[0094]
Therefore, for the part A in which the production facility M having the value “1” exists over a plurality of groups G, and in FIG. 7, the parts A13, A27, and A28, it is necessary to move the work W between the groups. Become.
[0095]
However, the number of times of weighting movement TG in this solution is 2 (times) × 1 (pieces) for component A13, 1 (times) × 2 (pieces) for component A27, and 1 (times) × 2 (pieces) for component A28. The total is only 6 (times), and it can be said that excellent results have been obtained.
[0096]
As described above, the grouping process of the production equipment according to the embodiment includes:
[0097]
(1) Genetic information corresponding to each production facility M, that is, a sequence in which genes C are arranged in a line is set as a gene sequence U,
[0098]
(2) A number Nu of initial solutions (groups of individuals) is generated by randomly storing the group number g of each group G of the production facility M as a value for each gene C under a predetermined principle,
[0099]
(3) The fitness FV (n) of each individual n is calculated based on the number of times the workpiece of each part A is moved between the groups, and an operation is performed using a predetermined genetic algorithm (elite storage). , Crossover, mutation) to generate the next generation of populations,
[0100]
(4) By repeating this operation a predetermined number of times Nr, an individual n having a high degree of matching is selected, and an optimal solution with a small number of times of movement (weighted number of times of movement) between the groups G of the work W is derived.
It is assumed.
[0101]
Therefore, even in a production system of a product in which there are hundreds to thousands of types of parts to be manufactured, for example, in a production system of a gas turbine engine, a conventionally skilled design such as optimizing the grouping of each production equipment is required. It is possible to derive a solution to a large-scale and complex optimization problem, which was very difficult for even a person to solve, in a short time.
[0102]
As described above, the present invention has been described based on the embodiments, but the present invention is not limited to only such embodiments, and various modifications are possible. For example, in the embodiment, the degree of conformity is calculated based on the number of movements of the work between the facilities. However, the basis for calculating the degree of conformity is not limited to the number of movements of the work between the facilities. May be set as appropriate, and the fitness may be calculated based on, for example, the similarity of the equipment.
[0103]
Further, in the embodiment, the facility is a production facility, but the facility to which the present invention is applied is not limited to the production facility, and may be various facilities.
[0104]
【The invention's effect】
As described in detail above, according to the present invention, an excellent effect is obtained that equipment in a system including a large number of equipment can be quickly allocated to an optimum group.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of a production facility grouping apparatus to which a production facility grouping method according to an embodiment of the present invention is applied.
FIG. 2 is a table diagram showing basic plan information in an example used to facilitate understanding of a grouping method of the production equipment.
FIG. 3 is a schematic flowchart showing a required procedure of a genetic algorithm used in a grouping method of the production equipment.
FIG. 4 is a schematic diagram showing a structure of a gene sequence in the genetic algorithm.
FIG. 5 is a schematic diagram showing a principle for setting a crossing point by an operation according to the genetic algorithm.
FIG. 6 is a schematic diagram showing a mode of crossover as an operation by the genetic algorithm.
FIG. 7 is a table showing optimal solutions obtained by executing the grouping method of the present production equipment in the example of FIG. 2;
[Explanation of symbols]
A parts
C gene
K making device
M production equipment
Nm Total production equipment
Nu population
U gene sequence
1 Processing unit
2 Input section
3 Output section
4 Database

Claims (20)

A grouping method of equipment in which a required number of facilities are allocated to a required number of groups in accordance with a processing procedure of a material to be processed,
A predetermined number of individuals having a group number as a one-dimensional array of numerical data strings and having a length corresponding to the total number of equipment as a genetic data having a length corresponding to the total number of equipment, by a genetic algorithm operation Generate a final generation by changing generations to a predetermined generation,
A grouping method characterized by allocating each facility to each group based on genetic information of an individual having an optimum degree of conformity to a predetermined evaluation criterion in the last generation. 2. The equipment grouping method according to claim 1, wherein an initial value of the numerical data sequence is set at random. Operation by genetic algorithm,
Elite preservation that leaves the highest number of individuals with the highest degree of fitness in the next generation for a given number of generations,
Crossover to generate individuals that inherit their genetic information from multiple parents,
2. The method of grouping equipment according to claim 1, further comprising a mutation that changes a part of the genetic information. 4. The grouping method for equipment according to claim 3, wherein the positions where the genetic information of the parents are crossed are set at random. The predetermined evaluation criterion is characterized in that the weighted number of movements is obtained by integrating the product of the number of times each material to be processed moves between groups and the number of the materials to be processed with respect to all the materials to be processed. The method of grouping equipment according to claim 1. 2. The method of grouping equipment according to claim 1, wherein an upper limit value and / or a lower limit value are provided for the number of facilities allocated to each group. 2. The equipment grouping method according to claim 1, wherein the number of groups is limited. A grouping device for equipment that divides a required number of facilities into a required number of groups in accordance with a processing procedure of a material to be treated,
A predetermined number of individuals having a group number as a one-dimensional array of numerical data strings and having a length corresponding to the total number of equipment as a genetic data having a length corresponding to the total number of equipment, by a genetic algorithm operation The system is configured to generate a final generation by changing generations to a predetermined generation, and to allocate each facility to each group based on genetic information of individuals whose fitness to the predetermined evaluation criteria is optimal in the final generation. A facility grouping device. 9. The equipment grouping apparatus according to claim 8, wherein an initial value of the numerical data sequence is set at random. Operation by genetic algorithm,
Elite preservation that leaves the highest number of individuals with the highest degree of fitness in the next generation for a given number of generations,
Crossover to generate individuals that inherit their genetic information from multiple parents,
9. The equipment grouping device according to claim 8, further comprising a mutation that changes a part of the genetic information. 11. The equipment grouping device according to claim 10, wherein the locations where the parent's genetic information is crossed are set at random. The predetermined evaluation criterion is characterized in that the weighted number of movements is obtained by integrating the product of the number of times each material to be processed moves between groups and the number of the materials to be processed with respect to all the materials to be processed. The equipment grouping device according to claim 8. 9. The equipment grouping device according to claim 8, wherein an upper limit value and / or a lower limit value are provided for the number of facilities allocated to each group. 9. The equipment grouping apparatus according to claim 8, wherein the number of groups is limited. A computer-readable program for assigning a required number of facilities to a required number of groups in accordance with a processing procedure of a material to be processed,
First-generation individuals having a group number as a one-dimensional array of numerical data strings and having the length of the numerical data string as a length corresponding to the total number of equipment as first-generation individuals are referred to as genetic information of each individual. A procedure of randomly generating and generating a predetermined number,
Procedure for calculating the degree of conformity of the obtained individual to a predetermined evaluation criterion,
A procedure for generating a final generation by causing a generation alternation to a predetermined generation by an operation according to a genetic algorithm for an individual who has reached a predetermined standard,
A procedure for calculating the fitness of the final generation individual with respect to a predetermined evaluation criterion,
A procedure for selecting an individual having an optimum fitness for a predetermined evaluation criterion among the final generation individuals,
Performing a grouping of facilities based on the genetic information of the selected individual. Operation by genetic algorithm,
Elite preservation that leaves the highest number of individuals with the highest degree of fitness in the next generation for a given number of generations,
Crossover to generate individuals that inherit their genetic information from multiple parents,
16. The computer readable program according to claim 15, further comprising a mutation that changes a part of genetic information. 17. The computer-readable program according to claim 16, further comprising a step of randomly setting a position where the parent genetic information element is crossed. The predetermined evaluation criterion is characterized in that the weighted movement number is obtained by integrating the product of the number of times each material to be processed moves between groups and the number of the materials to be processed with respect to all the materials to be processed. 16. The computer-readable program according to claim 15, wherein: 16. The computer readable program according to claim 15, wherein an upper limit value and / or a lower limit value are provided for the number of facilities allocated to each group. 16. The computer readable program according to claim 15, wherein the number of groups is limited.

Images