JPH0326672A - Group control device for elevator - Google Patents

Abstract

PURPOSE:To reduce a wait time by dicision-comparing an estimated number of sets of cages in a predetermined floor area with the first and second predetermined values and limiting an estimated vacant cage to be assigned to a riding spot call or excepting the cage from the assigned object. CONSTITUTION:In a group control device 10, information from each cage control unit 11 to 14 is input, and an estimated vacant cage, placed in a door-closed condition with no call provided, is detected by a detecting means 10H, while in a number of sets of cages-estimating arithmetic means 10E, being based on an estimated cage position and an estimated cage direction from an estimating means 10D, presence of and an number of sets of cages, estimated in a predetermined floor area, are estimation-calculated. This estimated number of sets is decision-compared with the first and second predetermined values, and the estimated vacant cage is limited to be assigned to a riding spot call or excepted from the assigned object by an assignment limiting means 10J. Thus, by considering a change of cage arrangement following the lapse of time and possibility of providing a vacant cage, a wait time for the riding spot call can be reduced from the present point of time over to the near future.

Description

【発明の詳細な説明】 ［産業上の利用分野］ この発明は，複数台のエレベータのかごの中から乗場呼
びにに対するサービスかごを選択し割り当てるエレベー
タの群管理装置に関するものである． ［従来の技術］ 複数台のエレベータが併設された場合は．通常群管理運
転が行われる．この群管理運転の一つに割当方式がある
が．これは乗場呼びが登録されると直ちに各かご毎に割
当評価値を演算し，この評価値が最良のかごをサービス
すべきかごとして選択して割り当て，上記乗場呼びには
割当かごだけを応答させるようにして．運行効率の向上
．および乗場待時間の短縮を計るものである．また，こ
のような割当方式の群管理エレベータにおいては，一般
に各階の乗場に各かごおよび各方向毎に到着予報灯を設
置し．これにより乗場待客に対して割当かごの予報表示
を行うようにしているので．待客は安心して予報かどの
前でかごを待つことができる． さて，上記のような乗場呼びの割当方式における割当評
価値は，現在の状況がそのまま進展するとしたらどのか
ごに乗場呼びを割り当てら最適かという観点に基づいて
演算されている．すなわち，現在のかご位置とかご方向
．および現在登録されている乗場呼びやかご呼びに基づ
いて，かごが上記呼びに順次応答して各階の乗場に到着
するまでに要する時間の予測値（以下，これを到着予想
時間という）と．乗場呼びが登録されてから経過した時
間（以下，これを継続時間という）を求め，さらに上記
到着予想時間と上記継続時間を加算して現在登録されて
いるすべての乗場呼びの予測待時間を演算する．そして
，これらの予測待時間の総和もしくは予測待時間の２乗
値の総和を割当評価値として設定し，この割当評価値が
最小となるかごに上記乗場呼びを割り当てる．このよう
な従来の方式では，乗場呼びの割当を行う場合，現在の
状況の延長線上で最適か否かを判断しているためにその
割当の後に新たに登録された乗場呼びが長持ちになると
いう不具合が発生していた．この不具合発生の例を第１
２図〜第１５図によって説明する．第１２図において，
ＡおよびＢは．それぞれ１号機および２号機のかこて，
１号機は６階のかこ呼び（６Ｃ）に応答すべく上昇中で
あり，２号機は５１１１の上り呼び（５ｕ）に割り当て
られており，上り割当（５ｕＢ）に応答すべく上昇中で
ある．このような状況において．第１３図のように９階
に下り呼び（９ｄ）が登録されたとする．上記従来の割
当方式の割当評価値に従うと．全体として待時間が最小
になるようにかこＡに９階の下り呼び（９ｄ）を割り当
てるため．１号機は９階の下り割当（９ｄＡ）を持ち，
その結果２台とも上方に向かって走行することになる． もし，この９階の下り呼び（９ｄ）の割当後しばらくし
て下方階に，例えば１階に上り呼び（１ｕ）が登録され
たとすると，この１階の上り呼び（１ｕ）はかごＡおよ
びかごＢの背後呼びとなり，いずれのかごに割り当てら
れたとしても応答されるまでに時間がかかり長待ちにな
ってしまうことになる．一方．９階の下り呼び（９ｄ）
をかとＢに割り当てた場合．約１５秒後には第１４図の
ようにかごＡは６階で空かごになって待機状態になり，
かごＢは５ＪＩｌでサービス中（この階で１０階のかご
呼び（１０ｃ）が登録されたとする）と予想される．し
たがって．その後にＩｌｌｉｆの上り呼び（１ｕ》が登
録されたとしても．第１５図に示すように６階で待機し
ているかごＡが直行サービスするので長待ちになること
はない．このように長持ちを防止するには，近い将来の
かご配置がどうなるのか，空かごが生じる可能性はどう
かを考慮し，一時的に待時間が長くなる割当を行ってで
も，かごが１か所に集まらないように乗場呼びを割り当
てる必要がある． ［発明が解決しようとする課題］ 従来，かごが１か所に集まらないように乗場呼びを割り
当てる方式が次に述べるようにいろいろ提案されてきた
が，十分な効果が得られていない． 特公昭５５−３２６２５号公報に示されているエレベー
タの群管理装置は，上記ゾーン割当方式と同様に．かご
が１か所に集まることを防ぎ運転効率の向上を計るため
に，乗場呼びが登録されるとその呼びの近くの階床に停
止する予定のあるかごを割り当てるという割当方式であ
る．この割当方式においても．近接階への停止予定かご
の有無に注目しているのみで，停止予定かごがその階に
到着するまでにどのくらいの時間を要するのか．他の乗
場呼びがどのように分布して登録されていていつ頃応答
されそうであるのか，近い将来空がごになりそうなかご
はないのか，他のかごはどの階にいてどの方向に運行し
ようとしているのか，など時間経過に伴うかご配置の変
化などを適確にとらえた判断を行っていないので，やは
り長持ち呼びが発生するという問題点があった．さらに
また，特公昭６２−５６０７６号公報に示されたエレベ
ータの群管理制御方法は，乗り捨て位置にかごを待機さ
せるものにおいて，新たに乗場呼びが発生するとこの乗
場呼びを順次各かごに仮に割り当てて仮割当かごの乗り
捨て位置を予想し，仮割当かごの予想乗り捨て位置とそ
の他のかごの位置からかごの分散度を演算し．少なくと
も上記分散度を各割当かごの評価値として分散度が大き
いほど割り当てられやすくなるようにして，各かごの上
記評価値から割当かごを決定するようにした割当方式で
ある．これにより．乗場呼びにサービス終了後も分散配
置された状態となり，分散待機による空かごの無駄運転
を防止して省エネルギーに大きな効果を発揮すると共に
ビル居住者の不審感をなくすことができるという効果を
有するものである．しかし，この割当方式はその目的か
ら明らかなように，夜間などの閑散時を対象とするもの
で，かごが全て空かごて待機している状態で乗場呼びが
一つ登録された場合を前提としている．そのため，乗場
呼びが次々に登録され各かごが呼びに応答しながらそれ
ぞれ運行しているというような交通状態における乗場呼
び割当にはこの割当方式を適用できず．長待ちが発生す
るという問題点があった．すなわち，このような問題が
生じるのは，空かごの配置をバランスさせることを目的
としているため，仮割当かご以外のかごに対し時間経過
に伴うかご位置の変化を考慮する構成になっていない（
その前提からして他のがごのかご位置変化を考慮する必
要がない）こと．および上記仮割当かごが乗り捨てられ
る時点のがご位置（その時点には全てのかごが空がごと
なり待機状態となる〉にのみ着目して乗場呼び割当の判
断をしていることが原因である． さらにまた，昭１１６３年電気・情報間連学会（昭和６
３年１０月３日〜５日開催．会場：新潟大学工学部）に
おいて発表された「エレベーターの群管理」　（予稿集
第２分冊Ｐ２−１１７〜１２０）では．ファジィ理論を
応用した新群管理方式が提案されている．これには．フ
ァジィルールの例として下記のものが示されている． （Ｒｕｌｅ　　Ｒｍ） ＩＦ （上方階に乗場呼び発生） ａｎｄ　（あるかご（Ａ）に割り当てると上方階にかご
が集中する） ＴＨＥＮ 《上記の性質（Ａ）を持つかごを除いて割当候補かごと
する） （割当候補かごの中から評価値最小の かごを割り当てる） さらに．シミュレーション例として第１６図に示すよう
な状況でＩＯＦの下り呼びが登録されたときには．空か
ごてある２号機と４号機を温存し，上方階にかご呼びを
持つ１号機と３号機の中から評価値に最も良いかご（３
号構）を割当かごに選択することが良いとしている．こ
れに従えば近い将来の呼び発生を考慮した割当が行え長
待ちの発生を防止することができる．しかしながら，「
上方階にかごが集中する」という判定が単に「上方階に
かご呼びを持つ」，または「上方階に割当呼びを持つ」
という条件だけで行われており．時間経過に伴うかご相
互の位置関係の変化を考慮する構成になっていない．そ
のため，いずれは上方階に達するかもしれないが他のか
ごとの位置関係が明確に規定されていないため，実際に
かごが「上方附に集中』しない場合も十分考えられ，そ
の場合には空かごを温存したがためにかえって待時間を
長くするという問題点があった．また，さらに．近い将
来に呼びに答え終わったあと戸閉状態で待機するであろ
うというかごく＝予測空かご〉の発生予測を全く考慮し
ていないため．近い将来の長待ちの発生の防止効果が十
分でないという問題点があった． この発明は．かかる問題点を解決するためになされたも
ので，時間経過に伴ったかご配置の変化と空かごになる
可能性とを考慮し，現時点から近い将来にわたって乗場
呼びの待時間を短縮することのできるエレベータの群管
理装置を得ることを目的とする． ［課題を解決するための手段］ この発明に係るエレベータの群管理装置は，乗場釦が操
作されると乗場呼びを登録する乗場呼゜び登録手段と，
前記乗場呼びに対して複数のかごの中からサービスすべ
きかごを選択して割り当てる割当手段と．かごの運行方
向決定．出発，停止．および戸開閉等の運転制御を行い
，かごをかご呼びと前記割当乗場呼びに応答させるかご
制御手段と，かごがすべての呼びに答え終わると答え終
わった階床で待機させるか．もしくは所定の階床へ走行
させて待機させる待機手段とを有するエレベータの群管
理装置において．現時点で応答すべき呼びを持っている
かごが呼びに順次応答して所定時間経過した後に応答す
べき呼びを持たす戸閉状態になっていると予測されるか
ごを予測空かごとして検出する予測空かご検出手段と．
かごが現時点から前記かご呼びと前記割当乗場呼びに順
次応答して所定時間経過後のかご位置とかご方向とをそ
れぞれ予測演算するかご位置予測手段と，前記予測かご
位置と前記予測かご方向に基づいて，前記所定時間経過
後に所定階もしくは所定階床域にいるであろうかごの有
無または台数を予測演算するかご台数予測手段と，前記
所定時間経過後に前記所定階もしくは前記所定階床域内
にいると予測されるかごの予測台数を第１の所定値と比
較判定する第１の判定手段と予測かごの台数を第２の所
定値と比較判定する第２の判定手段並びに前記第１の判
定手段と前記第２の判定手段の判定結果に基づいて前記
予測空かごが前記乗場呼びに割り当てるのを制限するか
もしくは割当対象から除外する指令を出力する第３の判
定手段とを有する割当制限手段とを備えたものである． ［作用］ この発明においては，乗場呼びに対する割当動作におい
て，所定時間経過後に所定階もしくは所定階床域にいる
であろうかごの台数の予測値と予測空かごの台数を使用
して所定の動作を行う．［実施例］ 第１図〜第１０図はこの発明の一実施例を示す図である
．なお，この実施例では１２階建ての建物に４台のかご
が設置されているものとする．第１図は全体構成図で．
群管理装置（１０）とこれによって嗣御される１号機〜
４号機用かご制御装置（１１）〜《１４）から構成され
ている．（１０Ａ）は乗場呼び登録手段で，各階の乗場
呼び（上り呼び．および下り呼び）の登録・解消を行う
と共に乗場呼びが登録されてからの経過時間，すなわち
継続時間を演算する．（ＩＯＢ）は到着予想時間演算手
段で，各かごが各階の乗場（方向別）に到着するまでに
要する時間の予測値．すなわち到着予想時間を演算する
．（Ｉ　ＯＣ）は乗場呼びにサービスするのに最良のか
ごを１台選択して割り当てる割当手段で，乗場呼びの予
測待時間と後述する割当制限手段による判定結果とに基
づいて割当演算を行う．（ＩＯＤ）はかご位置予測手段
で，かごが現時点から所定時間Ｔ経過後のかご位置とか
ご方向とを予測演算する．（ＩＯＥ）はかご台数予測手
段で，前記予測かご位置と予測かご方向に基づいて所定
時間Ｔ経過後に所定階床域にいるであろうかご台数を予
測演算する．　（１　０　Ｆ）は待機手段で，かごが全
ての呼びに答え終わると答え終わった階床もしくは特定
階でかごを待機させる．（ＩＯＧ）は空かご検出手段で
．応答すべき呼びを持っていず戸閉状態のかごを空かご
として検出する．《１０Ｈ）は予測空かご検出手段で，
現時点で応答すべき呼びを持っているかご（呼びに応答
中のかごも含む）が所定時間Ｔ経過後に空かごになって
いると予想されるかごを予測空かごとして検出する．（
ＩＯＪ）は割当制限手段で，前記予測空かごに対して乗
場呼びの割当を制限するもしくは割当対象から除外する
かどうかを判定する． （１１Ａ）は周知の乗場呼び打消手段で．１号機用かご
制御装置（１１）に設けられ，各階の乗場呼びに対する
乗場呼び打消信号を出力する．（１１Ｂ）は周知のかご
呼び登録手段で．同じく各階のかご呼びを登録する．（
１１Ｃ）は周知の到着予報灯制御手段で．同じく各階の
到着予報灯（図示しない）の点灯を制御する，　（１　
１　Ｄ）は周知の運行方向制御手段で，かごの運行方向
を決定する．（１１Ｅ》は周知の運転制御手段で，かご
呼びゃ割り当てられた乗場呼びに応答させるためにかご
の走行および停止を制御する．　（１　１　Ｆ）は周知
の戸制御手段で，戸の開閉を制御する．なお．２号機〜
４号機用かご制御装置（１２）〜（１４）も１号機用か
ご制御装置（１１）と同様に構成されている．第２図は
，群管理装置（１０）の構戒ブロック図で，群管理装置
（１０）はマイクロコンピュータ（以下．マイコンとい
う〉で構威され，ＭＰＵ（マイクロプロセシングユニッ
ト）（１　０　１　）．　ＲＯＭ（１０２）．ＲＡＭ（
１　０３），入力回路（１０４），および出力回路（１
０５）を有している．入力回路（１０４）には．各階の
乗場釦からの乗場釦信号（１９）．および１号機〜４号
機用かご制御装置（１１）〜（１４）からの１号機〜４
号機の状態信号が入力され．出力回路（１０５）から各
乗場釦に内蔵された乗場釦灯への信号（２０＞．および
１号機〜４号機用かご制御装置（１１）〜（１４）への
指令信号が出力される． 次に，この実施例の動作を第３図〜第８図を参照しなが
ら説明する．第３図は群管理装置（１ｏ）を構成するマ
イコンのＲＯＭ（１　０２）に記憶された群管理プログ
ラムを示すフローチャート．第４図はその空かご検出プ
ログラム（３３）を表すフローチャート，第５図は同じ
くかご位置予測プログラム（３５）を表すフローチャー
ト，第６図は同じくかご台数予測プログラム（３６）を
表すフローチャート．第７図は同じく割当制限濱箪プロ
グラム〈３７〉を表すフローチャート，第８図は建物を
複数の階床域（ゾーン）に分割した状態を表す図である
．まず．第３図の群管理動作の概要を説明する．ステッ
プ（３１）の入力プログラムは，乗場釦信号（１９），
１号機〜４号機用かご制御装置（１１）〜（１４）から
の状態信号《かご位置，方向．停止．走行，戸開閉状態
，かご負荷．かご呼び，乗場呼び打消信号など）を入力
するもので周知のものである． ステップ（３３）の空かご検出プログラムでは，第４図
に示すように応答すべき呼びを持っていず戸閉状態のか
ごを空かごとして検出する．第４図において，ステップ
（４６）．すなわちステップ（４２）〜（４５）は，１
号機が現在空かごてあるかどうかを検出する手順を表す
．１号機がかご呼びまたは剖当乗場呼びを持っていす，
無方向で戸閉状態であれば，ステップ（４２）→ステッ
プ（４４〉→ステップ（４５〉へと進み，ここで空かご
フラグＡＶ１を「１』にセットする．上記以外は．ステ
ップ（４３）で空かごフラグＡＶＩを「０』にリセット
する．２号機〜４号機に対しても同様にステップ（４７
〉〜（４９）でかごの検出を行い，空かごフラグＡＶ２
〜ＡＶ．を設定する． ステップ（３４）の到着予想時間演算プログラムでは．
１号機〜４号機に対して各乗場ｉ（１＝１．２，３，・
・・，１１は，それぞれＢ２，Ｂｌ．１，・・・．９階
の上り乗場．ｉ＝１２．１３，　　・・・２１．２２は
．それぞれ１０．９．　・・・１，８１階の下り方向乗
場を表す）への到着予想時間Ａｊ〈１）をかごＪ（Ｊ＝
１．２．３．４）毎に演算する．到着予想時間は．例え
ばかごが１階床進むのに２秒，１停止するのに１０秒を
要するものとして，かごが全乗場を順次一周運転するも
のとして演算される．なお，到着予想時間の演算は周知
のものである． ステップ（３５〉のかご位置予測プログラムでは，１号
機〜４号機の所定時間Ｔ経過後の予測かご位置Ｆｌ（Ｔ
）〜Ｆ４（Ｔ）と予測かご方向Ｄ＋（Ｔ）〜Ｄ４（Ｔ）
の各かごについてそれぞれ予測演算する．これを第５図
によって詳細に説明する．第１！Ｉのかご位置予測プロ
グラム（３５）において，ステップ（６５）．すなわち
ステップ（５１）〜（６４）は．１号機の所定時間Ｔｆ
＆の予測がご位置ｐ＋（’ｒ）と予測かご方向Ｄｔ（Ｔ
）を演算する手順を表す．１号機に割り当てられた乗場
呼びがある時は，ステップ（５１）→ステップ（５３）
へと進み，ここで最遠方の割当乗場呼びの前方にある終
端階を１号機の最終呼び階と予測し，その階でのがごの
到着方向（最上階では下り方向．最下階では上り方向）
も考慮して最終呼び予測乗場ｈ１として設定する．また
，１号機が割り当てられた乗場呼びを持たずにかご呼び
だけを持っている時は，ステップ（５１）→ステップ（
５２〉→ステップ（５４）へと進み，ここで最遠方のか
ご呼び階を１号機の最終呼び階と予測し，そのときのか
ごの到着方向も考慮して最終呼び予測乗場ｈ，として設
定する．さらに，また．１号機が割当乗場呼びもがご呼
びも持っていない時は，ステップ（５１）→ステップ（
５２）→ステップ（５５）へと進み，ここで１号機のか
ご位置階を最終呼び階と予測し，そのときのかご方向も
考慮して最終呼び予測乗場ｈ，として設定する． このようにして最終呼び予測乗場ｈ１を求めると，次に
，ステップ（５６）で１号機が空かごかどうかを判定す
る．１号機が空かごてない（ＡＶ１＝「Ｏ」）時は．ス
テップ（５７）で１号機が空かごになるまでに要する時
間の予測値（以下．空かご予測時間という）１＋を求め
る．空かご予測時間ｔ，は，最終呼び予測乗場ｈ，への
到着予想時間ＡＩ（ｈ１）にその乗場での停止時間の予
測値Ｔ　ｇ　（　”　１０秒）を加算して求める．なお
．かご位置階を最終呼び予測乗場ｈ１として設定した場
合は，がご状態（走行中，減速中．戸開動作中．戸開中
，戸閉動作中など）に応じて停止時間の残り時間を予測
し．これを空かご予測時間ｔ，として設定する．次に，
ステップ（５８》で所定時間Ｔを経過するまでに１号機
が空かごになるかどうかを判定する．１号機の空かご予
測時間ｔ１が所定時間Ｔ以下の時は，所定時間Ｔを経過
するまでに１号機が空かごになるということを意味して
いるので，ステップ（５８〉→ステップ（５９）へと進
み，ここで最終呼び予測乗場ｈ１に基づいてその乗場ｈ
１の階床を所定時ｒｔｊＪＴ経過後の予測かご位置Ｆ＋
（Ｔ）として設定する．また，予測かご方向Ｄ＋（Ｔ）
をｒＱＪに設定する．なお，予測かご方向ＤＩ（Ｔ）は
，ｒＱＪの時は無方向，『１」の時は上り方向，「２」
の時は下り方向を表す．そして，ステップ（６０）で１
号機の予測空かごフラグＰＡＶＩを「１』にセットする
． 一方．１号機のかご予測時間ｔ，が所定時間Ｔよりも大
きい時は，所定時間を経過してもまだ空かごになってい
ないということを意味しているので．ステップ（５８）
→ステップ（６１）へと進み，ここで乗場１−１の到着
予想時間Ａ＋（ｉ−１）と乗場ｌの到着予想時ｒＷＪ　
Ａ　ｔ　（　ｉ　）ｆｆｉ　（　Ａ　ｒ　（　ｉ　−１
　）　＋　Ｔ　ａ　≦Ｔ＜　Ａ　Ｉ（　ｉ　）　＋　Ｔ
　ｓ　｝となるような乗場ｌの階床を所定時間Ｔ経過後
の予測かご位置Ｆ．（Ｔ）として設定し，この乗場ｉと
同じ方向を予測かご方向Ｄ，（Ｔ）として設定する．そ
して，ステップ（６２）で１号機の予測空かごフラグＰ
ＡＶ，をｒＱＪにリセットする． また，ステップ（５６）で１号機が空かごくＡＶ＋−　
’　Ｉ　Ｊ　）の時は．ステップ（６３）で空かご予測
時間ｔ，をＯ秒に設定し，ステップ（６４）で最終呼び
予測乗場ｈ１に基づいてその乗堝ｈ１の階床を所定時間
Ｔ経過後の予測かご位置Ｆｌ（Ｔ）として設定する．ま
た，予測かご方向Ｄ，（Ｔ）を『０」に設定する．そし
て．ステップ（６２〉で１号機の予測空かごフラグＰ　
Ａ　Ｖ　＋を「Ｏ」にリセットする． このようにして．ステップ（６５）で１号機に対する予
測かご位置Ｆｌ（Ｔ）と予測かご方向Ｄ＋（Ｔ）の演算
と予測空かごの検出をするが，２号機〜４号機に対する
予測かご位置Ｆｔ（Ｔ）〜Ｆ４（Ｔ），予測かご方向ｏ
ｘ（’ｒ）〜Ｄ４（Ｔ）．および予測空かごフラグＰ　
Ａ　Ｖ　ｚ〜ＰＡＶ４もステップ（６５）と同じ手順か
らなるステップ｛６６｝〜（６８）でそれぞれ設定され
る． 再び第３図において．ステップ（３６）のかご白数予測
プログラムでは，所定時ｒＨＩＴ経過後に所定階床もし
くは所定階床域にいるかご白数，例えば第８図に示すよ
うに，１階床または連続した複数階床からなる階床域（
ゾーン）２１〜２．に対して予測かご白数Ｎ＋（Ｔ）〜
Ｎ．（Ｔ）をそれぞれ演算する．これを第６ｒｆ！１に
よって詳細に説明する．第６図のかご台数予測プログラ
ム《３６〉において．ステップ（７１）で予測かご台数
Ｎ，（Ｔ）〜Ｎ＠　（Ｔ）をそれぞれ「０」台に，号機
番号ｊおよびゾーン番号ｍをそれぞれ「１」に初期設定
する．ステップ（７２）では，ｊ号機の予測かご位置Ｆ
Ｊ（Ｔ）と予測かご方向ＤＪ（Ｔ）に基づいて，所定時
間Ｔ経過後にｊ号機がゾーンＺｍにいるかどうかを判定
する．ｊ号機がゾーンＺｍにいると予測されると．ステ
ップ（７３）でゾーンＺｍの予測かご台数Ｎｍ（Ｔ）を
１台増加させる．ステップ（７４）では号機番号ｊを一
つ増加させ．ステップ（７５）で全号機について判定し
終わったかどうがをチェックする．終了していなければ
ステップ（７２）に戻り．上述の処理を繰り返す． ゾーン番号ｍのゾーンＺｍについてステップ（７２）お
よびステップ（７３）の処理を全号機終了すると，次に
．ステップ〈７６〉で，ゾーン番号ｍを一つ増加させる
と共に号機番号ｊを「１」に初期設定する．そして．同
じようにステップ（７２〉〜（７５〉の処理を号機番号
ｊ＞４となるまで繰り返す．すべてのゾーン２１〜Ｚ．
について上述の処理を終わるとステップ（７７〉でゾー
ン番号ｍ＞６となり，ゾーンｚ１〜２．についてのかご
台数予測の処理を終了する． ステップ（７８）〜（８４）では．空かごの白数ＮＡＶ
と予測空かごの台数ＮＰＡＶをカウントする．ステップ
（７８〉で台数ＮＡＶ．白数ＮＰＡＶをそれぞれ「０」
台に．号機番号ｊを「１」に初期設定する．ｊ号機が空
かごてあれば，ステップ（７９）→ステップ（８０）と
進み．ここで空かごの台数ＮＡＶを１台増加する．また
．ｊ号機が予測空かごてあれば，ステップ（７９）→ス
テップ（８１）→ステップ（８２）と進み，ここで予測
空かごの台数ＮＰＡＶを増加する．ステップ（８３）で
は号機番号ｊを一つ増加させ．ステップ（８４）で全号
機について判定し終わったかどうかをチェックする．終
了していなければステップ（７９〉に戻り．上述の処理
を繰り返す． このようにして．ゾーン毎の予測かご台数Ｎ，（Ｔ）　
〜Ｎｓ（’ｒ），空かごの白数ＮＡＶ，　および予測空
かごの台数ＮＰＡＶをカウントし．かご台数予測プログ
ラム（３６）の処理を終了する．第３図の群管理１ログ
ラム＜１０）におけるステップ（３７）の割当制限プロ
グラムでは，新たに乗場呼びＣが登録されると，その乗
場呼びＣの階床位置，その時の上記予測かご白数Ｎｌ（
Ｔ）〜Ｎ６（Ｔ），空かどの白数ＮＡＶ，および予測空
かごの台数ＮＰＡＶに基づいて，上記乗場呼びＣに対し
て１号機〜４号機の割当を制限するかどうかを判定し，
上記新規乗場呼び５Ｃに割り当てにくくするための割当
制限評価値２１〜Ｐ，をそれぞれ設定する．なお．ｗ４
当制限評価値Ｐ，〜Ｐ４は大きな値になるほど割当制限
の程度が上がることを意味し．この値が無限大になれば
最初から割当の対象から除くことと等価になる．この判
定手順を第７図によって詳細に説明する． 第７図の割当制限プログラムにおいて．上記新規乗場呼
びＣが上方階ゾーン（ＺｓまたはＺ．）に属し，空かご
が１台もいな＜（ＮＡＶ＜１）て．所定時間Ｔ経過後に
上方階ゾーン（Ｚ３またはＺ４）にいると予想されるか
ごの台数が多＜　（Ｎ　３（Ｔ　）”Ｎ　４（Ｔ）≧Ｎ
ａ）て，予測空かごはいるが多くはな＜（１≦ＮＡＶ≦
Ｎｂ）で，かつ下方階ゾーン（Ｚｔまたはｚ６）で乗場
呼びが発生しやすい交通状態であれば，ステップ（８１
）→ステップ（８２）→ステップ（８３）→ステップ（
８４）→ステップ〈８５）→ステップ（８６〉と進み，
ここで予測空かごｋ（ｋε（１，２，３．４））の割当
制限評価値Ｐｋを「９９９９９』に．予測空かごでない
かごｎ（ｎε（１，２，３．４）かつｎ≠ｋ〉の割当制
限評価値ＰｎをｒＱＪに設定する．上記以外は．ステッ
プ（８７）で全てのかごの割当制限評価値Ｐ，〜Ｐ，を
ｒＯ』に設定する．このようにして割当制限評価値Ｐ１
〜Ｐ，が設定される．第３図の群管理プログラム（１０
）におけるステップ（３８〉の待時間評価プログラムで
は．新規乗場呼びＣを１号機〜４号機にそれぞれ仮割当
した時の各乗場呼びの待時間に関する評価値Ｗｌ〜Ｗ４
を演寡する．この待時間評価値Ｗ１〜Ｗ，の演算につい
ては周知であるので詳細な説明は省略するが，例えば１
号機を仮割当した場合は．１号機を仮割当した時の各乗
場呼びｉの予測待時間Ｕ（ｉ）（ｉ＝１．２．　　・・
・．２２：乗場呼びが登録されていなければｒ０１秒と
する）を求め，これらの２乗値の総和，すなわち待時間
評価値Ｗ，・Ｕ（１）”＋　Ｕ　（２　＞”＋・・・＋
Ｕ（２２）宜でもって演算する．次に．ステップ（３９
）の割当かご選択プログラムでは，上記割当制限評価値
Ｐ＋〜Ｐ，と待時間評価値Ｗ，−Ｗ．に基づいて割当か
ごを１台選択する．この実施例では，ｊ号機に新規乗場
呼びＣを仮割当した時の総合評価値ＥＪを，Ｅｊ＝Ｗｊ
＋ｋ−ＰＪ（ｋ：定数）で求め．この総合評価値ＥＪが
最小となるかごを正規の割当かごとして選択するもので
ある．割当かごには乗場呼びＣに対応した割当指令と予
報指令を設定する． さらに．ステップ《４０〉の待機動作プログラムでは，
すべて乗場呼びに答え終わった空かごが生じると，かご
が１か所に固まらないようにするため．上記空かごを最
終呼びの階でそのまま待機させるか，または特定階で待
機させるかを判定し，特定階で待機すると判定した時は
その特定階へ走行させるための待機指令を上記空かごに
設定する．最後に．ステップ（４１〉の出力プログラム
では，上記のようにして設定された乗場釦灯信号（２０
）を乗場に送出すると共に．割当信号，予報信号，およ
び待機指令などを１号機〜４号機用かご制御装置（１１
）〜〈１４）に送出する．このような手順で上記群管理
プログラム（３１〉〜（４］》を繰り返し実行する． 次に，この実施例における群管理プログラム（１０）の
動作を第９図．および第１０５？Ｉによって，さらに具
体的に説明する．なお，簡単のために第１２図〜第１５
図で用いた例を使用して説明する．第１２図に示す状態
において，第９図に示すように９１Ｉ１の下り呼び（９
ｄ〉が登録されたものとする．なお．５階上り呼び（５
ｕ〉の継続時間は１０秒とする．この時．かごＡに９階
の下り呼び（９ｄ）を仮割当した時の９階の下り呼び（
９ｄ）および５階上り呼び（５ｕ）の予測待時間はそれ
ぞれ２４秒と１６秒となり．この時の待時間評価値ＷＡ
は，ＷＡ＝２４”＋１６”＝８３２となる．一方，かご
Ｂに仮割当した時の９ｌｌ１の下り呼び（９ｄ）および
５階上り呼び（５ｕ）の予測待時間はそれぞれ２８秒と
１６秒となり，この時の待時間評価値Ｗ．は，Ｗ．＝２
８”＋１６’＝１０４０となる．したがって．従来の割
当方式であれば，ＷＡ＜Ｗ．であるので９階下り呼び（
９ｄ）はかごＡに割り当てられる．さて．かごＡおよび
かごＢの所定時間Ｔ（Ｔ＝２０秒）経過後のかご位置は
第１０図に示すようにかごＡ′およびかごＢ′のように
なる．シタがって，この時の予測かご台数は，Ｎ．（Ｔ
）＝２台，Ｎｌ（Ｔ）＝Ｎ１（Ｔ）＝Ｎ４（Ｔ）＝Ｎｓ
（Ｔ）＝Ｎ＠（Ｔ）・０台となり．空かご台数ＮＡＶ＝
Ｏ台，予測空かご台数ＮＰＡＶ＝１台となる．なお．こ
の例では無方向のかごは上り方向とみなしたが，かご位
置に応じて適宜方向を決めればよい．また．この例にお
いて．一定値Ｎａ＝２台，Ｎｂ＝１台とすれば．Ｎ３（
Ｔ）＝２台は一つのゾーンに全てのかごがいる場合に相
当するので．第７図の割当制限プログラム（３７）のス
テップ（８６〉においてかごＡの割当制限評価値ＰＡ＝
９９９９９．かごＢの割当制限評価値Ｐ．＝０と設定さ
れる．したがって．総合評価値は，ＥＡ＝ＷＡ＋ＰＡ＝
９９９９９＝１０８３１，Ｅ．＝Ｗ．＋Ｐｓ＝１０４０
＋Ｏ＝１０４０．でＥＡ＞Ｅ−となるので．最終的に９
階下り呼び（９ｄ）はかごＢに割り当てられる． 従来の割当方式だとかごＡに割り当てられて近い将来に
かごはだんご運転となり長待ち呼びが発生しやすくなる
．しかし．所定時間Ｔ経過後のかご配置を考慮してかご
Ｂに割当することにより，このようなだんご運転を防止
することができる．以上説明したように．上記実施例で
は．かごが現時点から呼びに順次応答して所定時間経過
後のかご位置とかご方向とを予測演算し．さらにこれら
に基づいて予測空かごの台数と各ゾーンにおける所定時
間経過後のかご台数を予測演算し，これらの予測かご台
数に応じて予測空かごに対する劃当制限動作を行わせる
ようにしたので，かごが１か所に集中することがなくな
り，現時点から近い将来にわたって乗場呼びの待時間を
短縮することができる． なお．上記実施例では，所定時間Ｔ経過後のかご位置と
かご方向を予測する時，まずかごが最終呼びに答え終わ
って空かごになるであろう階床とそれまでに要する時間
を予測し，その上で所定時間Ｔ経過後のかご位置とかご
方向を予測するようにした．これは，かごが空かごにな
るとその階でそのまま待機するものと仮定したからであ
る．空かごを特定階で必ず待機させることが決まってい
る場合であれば．特定階に走行させるものとしてかご位
置とかご方向を予測すればよい．また，所定時間Ｔを経
過するまでに新たに発生するであろう呼びも考慮してか
ご位置とかご方向を予測することもできる．さらにまた
，最終呼び予測乗場の演算方向もこの実施例のように簡
略化したものではなく，統計的に求めたかご呼びゃ乗場
呼びの発生確率に基づいてきめ細かく予測するものであ
ってもよい． また，上記実施例では，第８図に示すようなゾーンに建
物を分割したが階床数や設置かご台数の他．時間帯や各
階床の用途（主階床，食堂階，集会室階．乗継階なと〉
に応じて逐次ゾーンの設定の仕方を変更することも容易
である．また，必ずしも乗場の方向を考慮してゾーンを
決める必要はない． さらにまた，上記実施例では， （イ）新規乗場呼びＣが上方階ゾーンに属している時． ■　空かごが１台もなく， ■　所定時間Ｔ経過後に上方階ゾーンにいると予想され
るかごの台数が多く， ■　予測空かごが１台はいるが．多くはなく，■　下方
階ゾーンに乗場呼びが発生しやすい交通状態である． という条件■〜■がいずれも成立すると，所定の予測空
かごに対して新規乗場呼びＣへの割当を制限するための
割当制限評価値（〉Ｏ》をそれぞれ設定する． というようにした．しかし．割当制限評価値の設定条件
はこれに限るものではない． （口）新規乗場呼びＣが下方階ゾーンに属している時． ■　空かごが１台もなく． ■　所定時ｌＷＩＴ経過後に下方階ゾーンにいると予想
されるかごの台数が多く． ■　予測空かごが１台はいるが．多くはなく，■　上方
階ゾーンに乗場呼びが発生しやすい交通状態である． という条件（ロ》を適用することもできるし．（ハ）新
規乗場呼びＣが中間階ゾーンに属している時． ■　空かごが１台もなく， ■　所定時ｔｌＩＴ経過後に中間階ゾーンにいると予想
されるかごの台数が多く． ■　予測空かごが１台はいるが，多くはなく．■　上方
階ゾーンまたは下方階ゾーンに乗場呼びが発生しやすい
交通状態である． という条件（ハ）を適用することもできる．さらに，１
つのゾーンにかごが集中するということの裏返しを使用
する． 〈二）新規乗場呼びＣが登録された時，■　空かごが１
台もなく， ■　上方階ゾーン，下・方階ゾーン．または中間階ゾー
ンのいずれかのゾーンで，新規乗場呼びＣが属さないゾ
ーンで．かつ所定時間Ｔ経過後にいると予想されるかご
が１台もないというゾーンが存在して， ■　予測空かごが１台はいるが，多くはなく，■　上記
■を満たすゾーンに乗場呼びが発生しやすい交通状態で
ある． という条件（二）を適用することもできる．なお．この
条件（二）が戒立した時，予測空かごのうち上記■を満
たすゾーンに最も近い予測空かごを割当制限することが
望ましい． ここで，上記条件（イ〉〜（二〉における条件項■は，
予測空かごを温存することによる効果をできるだけ高く
するために設けているものである．この条件項■の交通
状態の条件がなくてもこの発明の効果を大きく損なうも
のではない．また，上記条件《イ）〜（二）における条
件項■の空かごの台数に関する条件を空かご台数＝０台
としたが，空かご台数≧１台の場合にもこの発明を適用
することができる．例えば， 《ホ）新規乗場呼びＣが上方階ゾーンに属している時， ■　空かごが上方階ゾーンに１台いて，■　所定時間Ｔ
軽過後に上方階ゾーンにいると予想されるかごの台数が
多く， ■　予測空かごが１台はいるが，多くはない．という条
件（ホ｝が成立すると，所定の予測空かごに対して新規
乗場呼びＣへの割当を制限−するということも可能であ
る．この条件（ホ）が有効なのは．下方階ゾーンにすぐ
乗場呼びが発生するというような混雑した交通状態にな
い場合である．すなわち．新規乗場呼びＣには空かごを
有効に使用して短時間でサービスさせ，下方階ゾーンの
近い将来登録されるであろう乗場呼びには近いうちに空
かごになるであろう予測空かごをサービスさせるという
ことを意図している． さらにまた，上記条件（イ）〜（ホ）における条件項■
で．「予測空かごの台数が１台以上で．かつ一定値Ｎｂ
以下』という条件を付けたが，これは予測空かごの台数
が多いときには割当制限による予測空かごの温存は必ず
しも必要ないためである．「一定値Ｎｂ以下」という条
件を無くシ．割当制限すべき予測空かごを選択する処理
の中で，特定のｌ台もしくは２台（例えば，空かご予測
時間１，〜ｔ，を用いて最も早く空かごになりそうなか
ごを選択したり，または予測かご位置Ｆｌ（Ｔ）〜Ｆ．
（Ｔ）と予測かご方向ＤＩ（Ｔ）〜Ｄ．（Ｔ）を用いて
下方階ゾーンに最も近いかごを選択したりする〉を割当
制限するようにしても同様の効果が得られる． この他はも予測かご台数に基づく予測空かごへの割当制
限の条件は種々考えられるが，第７図に示した条件（イ
）と同様に容易に実現できることは明白である． さらにまた，上記（イ）〜（ホ）のほかそれぞれの状況
に応じた複数の条件が設けられている時，同時に２つ以
上の条件が戒立する場合も考えられる．このような場合
，どの条件に従って割当制限するかごを決めるかという
問題が生じる．このような問題を解決する方法には種々
考えられるが，その一つによく知られたファジィ理論を
利用するものがある．例えば，前述の「エレベーターの
群管理装置」　（昭和６３年電気・情報関連学会．予稿
集第２分冊Ｐ２−１　１７〜１２０）で詳細に詳述され
ているように，条件を構成している条件項のそれぞれが
成立する度合いをメンバーシップ関数でＯから１の間の
数値で表し．それらを用いてＡＮＤ結合であれば最小値
を，ＯＲ結合であれば最大値を選択して，条件そのもの
の確信度を計算し，最終的に最も高い確信度を持つ条件
を一つ選択する方式である．また，各条件に対応した割
当制限の仕方に確信度の大きさに応じた重み付けをそれ
ぞれ行い，それに従ってかごを割当制限するという方式
もある．このような方式における条件に対してもこの発
明を適用することができることは明白である．このよう
に．所定階もしくは所定階床域に集中するであろうかご
の予測台数と近い将来空かごになるであろう予測空かど
の台数とを使用した条件であればどのようなものであっ
てもよい． さらにまた，上記実施例では．乗場呼びへ割当を制限す
る手段として．特定のかごに対しては他のかごより大き
な値を持つ割当制限評価値を設定し．これを待時間評価
値に重み付け加算して総合評価値を求め，この総合評価
値が最小のかごを正規の割当かごとして選択する方式を
使用した．このように割当制限評価値を他の評価値と組
み合わせて総合評価し割り当てるということは．割当制
限評価値の小さいかごを優先的に割り当てるということ
に他ならない．すなわち，上記割当制限評価値が大きい
かごは他のかごより割り当てにくくなる． また．乗場呼びへの割当を制限する手段は上記実施例に
限るものではなく．割当制限条件を満たすかごを予め割
当かごの候補から除外する方式であってもよい． さらにまた，上記実施例では．待時間評価値を乗場呼び
の予測待時間の２乗値の総和としたが，待時間評価値の
演算方法はこれに限るものではない．例えば，登録され
ている複数の乗場呼びの予測待時間の総和を待時間評価
値としたり，同じく予測待時間の最大値を待時間評価値
とする方式を使用するもであってもこの発明を適用でき
ることは明白である．もちろん．割当制限評価値と組み
合わせる評価項目は待時間に限るものではなく，予測外
れや満員などを評価項目とする評価指標と組み合わせて
も良いものである． また，上記実施例では，一種類の所定時間Ｔについて所
定時間経過後のかご位置とかご方向を各かごについてそ
れぞれ予測し，これに基づいて割当制限評価値を演算す
るようにしたが，複数種類の所定時間Ｔｌ，　Ｔ２，　
　・・・，　Ｔｒ　（ＴＩ＜Ｔ２〈・・・＜Ｔｒ）につ
いて所定時間経過後のかご位置とかご方向を各かごにつ
いてそれぞれ予測し，さらに複数種類の所定時間ＴＩ，
　Ｔ２，・・・，Ｔｒについて所定時間経過後の予測か
ご白数Ｎ　ｍ　（Ｔ　１）〜Ｎｍ（Ｔｒ）を各ぞーンＺ
ｍ（ｍ＝１．２，　・・１についてそれぞれを演算する
．また．予測空かご台数ＮＰＡＶ（ＴＩ）〜ＮＰＡＶ（
Ｔｒ）をそれぞれ演算する．そして．各組み合わせ｛Ｎ
，（ＴＩ），Ｎ２（Ｔ２）ｓ・・，Ｎ　Ｐ　Ａ　Ｖ　（
ＴＩ））．　（　Ｎ　＋（Ｔ２）．Ｎ　＊（Ｔ２）．−
　−　・，ＮＰＡＶ（Ｔ２Ｎｓ・・．（Ｎ＋（Ｔｒ）．
Ｎ＊（Ｔｒ＞，−　・・ＮＰＡＶ（Ｔｒ））．によって
それぞれ設定された割当制限評価値Ｐ　（ＴＩ）．Ｐ　
（Ｔ２）．・・・Ｐ　（Ｔ　ｒ　）．を重み付け加算す
る．すなわち，Ｐ＝ｋｌ・Ｐ（Ｔｌ）＋ｋＰ（Ｔ２）＋
ｋｙ”Ｐ（Ｔｒ）．（但し．ｋｌ，ｋ２，・・・，ｋｒ
は重み係数〉なる算式に従って演算することにより．最
終的な割当制限評価値Ｐを設定することも容易である．
この場合．ある一時点Ｔだけのかご配置に注目するので
はな＜　．ＴＩ．Ｔ２，・・・，Ｔｒという複数の時点
におけるかご配置を大局的に評価することになるので．
現時点から近い将来にわたって乗場呼びの待時間を一層
短縮することが可能となる．なお．上記重み係数ｋｌ，
ｋ２，・・・，ｋｒは，例えば第１１図に示すように．
どの時点のかご配置を重視するかによって何通りかの設
定方法が考えられるが．交通状態や建物の特性などに応
じて適宜選択すればよい．［発明の効果］ この発明は以上説明したとおり．乗場釦が操作されると
乗場呼びを登録する乗場呼び登録手段と．前記乗場呼び
に対して複数のかごの中からサービスすべきかごを選択
して割り当てる割当手段と，かごの運行方向決定，出発
，停止．および戸開閉等の運転制御を行い．かごをかご
呼びと前記割当乗場呼びに応答させるかご制御手段と．
かごがすべての呼びに答え終わると答え終わった階床で
待機させるか．もしくは所定の階床へ走行させて待機さ
せる待機手段とを有するエレベータの群管理装置におい
て．現時点で応答すべき呼びを持つているかごが呼びに
順次応答して所定時間経過した後に応答すべき呼びを持
たす戸閉状態になっていると予測されるかごを予測空か
ごとして検出する予測空かご検出手段と，かごが現時点
から前記かご呼びと前記割当乗場呼びに厘次応答して所
定時間経過後のかご位置とかご方向とをそれぞれ予測演
算するかご位置予測手段と．前記予測かご位置と前記予
測かご方向に基づいて，前記所定時間経過後に所定階も
しくは所定階床域にいるであろうかごの有無または台数
を予測演算するかご白数予測手段と，前記所定時間経過
後に前記所定階もしくは前記所定階床域内にいると予測
されるかごの予測台数を第１の所定値と比較判定する第
１の判定手段と予測かごの台数を第２の所定値と比較判
定する第２の判定手段並びに前記第１の判定手段と前記
第２の判定手段の判定結果に基づいて前記予測空かごが
前記乗場呼びに割り当てるのを制限するかもしくは割当
対象から除外する指令を出力する第３の判定手段とを有
する割当制限手段とを備えているので．時間経過に伴っ
たかご配置の変化と空かごになる可能性とを考慮し，現
時点から近い将来にわたって乗場呼びの待時間を短縮す
ることができる効果がある． ４．区画の酊単な説明 第１図はこの発明による一実施例によるエレベータの群
管理装置の全休楕戒ブロック図．第２図は第１図の群管
理装置（１０）の楕戒ブロック図，第３図は群管理プロ
グラム（１０）のフローチャート図，第４１５！Ｉは第
３図の空かご検出プログラム（３３）のフローチャート
図，第５図は第３図のかご位置予測プログラム（３６）
のフローチャート図．第６図はかご台数予測プログラム
（３６）のフローチャート図，第７図は第３図の割当制
限プログラム（３７》のフローチャート図．第８図は建
物のゾーン分割を示す図，第９図および第１０図は呼び
とかご位置の関係を示す図，第１１図はこの発明の他の
実施例を説明する図である．第１２図〜第１６図は従来
のエレベータの群管理装置を示し，それぞれ呼びとかご
位置の関係を示す図である．図において，（ＩＯＡ）・
・・乗場呼び登録手段，（ＩＯＣ）・・・割当手段．（
ＩＯＤ＞・・・かご位置予測手段．（ＩＯＥ）・・・か
ご台数予測手段，（ＩＯＦ）・・・待機手段．（ＩＯＧ
）・・・空かご検出手段．（ＩＯＨ）・・・予測空かご
検出手段，（ＩＯＪ）・・・割当制限手段．（１１）〜
（１４）・・・かご制御手段である． なお．各図中同一符号は同一又は相当部分を示す．[Detailed Description of the Invention] [Field of Industrial Application] This invention relates to an elevator group management device that selects and allocates a service car for a hall call from among a plurality of elevator cars. [Conventional technology] When multiple elevators are installed together. Normal group management operation is performed. One of these group management operations is the assignment method. As soon as a hall call is registered, an allocation evaluation value is calculated for each car, and the car with the best evaluation value is selected and assigned as the car to be serviced, and only the assigned car responds to the hall call. In this way. Improving operational efficiency. It also aims to reduce waiting time at the boarding point. In addition, in group-controlled elevators using this type of assignment system, arrival warning lights are generally installed for each car and each direction at the landings on each floor. This allows us to display the forecast for the assigned car to passengers waiting at the landing. Waiting customers can safely wait for their cars in front of the forecast. Now, the allocation evaluation value in the above-mentioned hall call allocation method is calculated based on the viewpoint of which car would be the best car to allocate a hall call to if the current situation continued as it was. In other words, the current car position and car direction. and the predicted value of the time required for the car to respond to the above calls in sequence and arrive at the landing on each floor based on the currently registered landing calls and car calls (hereinafter referred to as expected arrival time). Calculate the estimated waiting time for all currently registered hall calls by calculating the time that has passed since the hall call was registered (hereinafter referred to as duration time) and adding the above expected arrival time and the above duration time. do. Then, the sum of these predicted waiting times or the sum of the squared values of the predicted waiting times is set as the allocation evaluation value, and the hall call is allocated to the car with the minimum allocation evaluation value. In this conventional method, when allocating hall calls, it is determined whether or not it is the best extension of the current situation, so newly registered hall calls after the assignment will last longer. A problem had occurred. The first example of this problem is
This will be explained using Figures 2 to 15. In Figure 12,
A and B are. The irons of No. 1 and No. 2, respectively.
Unit 1 is ascending to respond to the 6th floor call (6C), while Unit 2 is assigned to the up call of 5111 (5u) and is ascending to respond to the up call (5uB). In this situation. Assume that a down call (9d) is registered on the 9th floor as shown in Figure 13. According to the allocation evaluation value of the above conventional allocation method. To assign the 9th floor down call (9d) to Kako A so that the overall waiting time is minimized. Unit 1 has a downlink assignment (9dA) on the 9th floor,
As a result, both cars will travel upwards. If an up call (1u) is registered on a lower floor, for example on the 1st floor, some time after the down call (9d) on the 9th floor is assigned, then this up call (1u) on the 1st floor will be assigned to car A and car This will be a back call for B, and even if it is assigned to any car, it will take a long time to be answered, resulting in a long wait. on the other hand. 9th floor down call (9d)
If you assign B to B. Approximately 15 seconds later, as shown in Figure 14, car A becomes an empty car on the 6th floor and is in a standby state.
Car B is expected to be in service at 5JIl (assuming that the car call for the 10th floor (10c) is registered on this floor). therefore. Even if Illif's up call (1u) is registered after that, there will be no long waiting time because car A waiting on the 6th floor will provide direct service as shown in Figure 15. To prevent this, consider how the basket arrangement will be in the near future and whether there is a possibility that there will be empty baskets, and make sure that the baskets do not gather in one place, even if you make allocations that will temporarily lengthen the waiting time. It is necessary to allocate hall calls. [Problem to be solved by the invention] Conventionally, various methods of allocating hall calls to prevent cars from gathering in one place have been proposed as described below, but none of them have been sufficiently effective. The elevator group control device shown in Japanese Patent Publication No. 55-32625 is similar to the above-mentioned zone allocation method.It is designed to prevent cars from gathering in one place and improve operational efficiency. In this allocation method, when a hall call is registered, a car that is scheduled to stop at a floor near that call is assigned.In this allocation method, too, attention is paid to the presence or absence of cars scheduled to stop at a nearby floor. How long does it take for a car scheduled to stop to arrive at that floor? How are other landing calls distributed and registered, and when are they likely to be answered? The system does not make judgments that accurately capture changes in car arrangement over time, such as whether there are any cars that are likely to become cars, which floors other cars are on, and in which direction they are trying to move. There was also the problem that a long-lasting call was generated.Furthermore, the elevator group management control method disclosed in Japanese Patent Publication No. 62-56076 is for waiting cars at the drop-off position, but when a new hall call is received. When a landing call occurs, the car is temporarily assigned to each car in order, the drop-off position of the temporarily assigned car is predicted, and the dispersion degree of the car is calculated from the predicted drop-off position of the temporarily assigned car and the positions of other cars.At least the above dispersion degree is calculated. This is an allocation method in which the evaluation value of each allocated car is such that the higher the degree of dispersion, the easier it is to be allocated, and the allocated car is determined from the above evaluation value of each car. This has the effect of being distributed in a distributed manner, preventing wasteful operation of empty cars due to distributed standby, and having a great effect on energy conservation, as well as eliminating the sense of suspiciousness among building occupants.However, this allocation As is clear from its purpose, the method is intended for off-peak hours such as nighttime, and assumes that one hall call is registered while all the cars are empty and waiting. Therefore, this allocation method cannot be applied to platform call assignment in traffic conditions where platform calls are registered one after another and each car is operating while responding to the calls. There was a problem with long waiting times. In other words, this problem arises because the purpose is to balance the arrangement of empty cars, and the structure is not designed to take into account changes in car position over time for cars other than provisionally allocated cars (
Based on that premise, there is no need to consider changes in the cage positions of other cages. This is due to the fact that the boarding hall call assignment is determined by focusing only on the position of the temporarily assigned car at the time of abandonment (at that point, all the cars are empty and on standby). .Furthermore, in 1163, the Electrical and Information Technology Association (Showa 6)
Held from October 3rd to 5th. Venue: Niigata University Faculty of Engineering) "Group Management of Elevators" (Proceedings Vol. 2 P2-117-120). A new group management method applying fuzzy theory has been proposed. For this. The following is an example of a fuzzy rule. (Rule Rm) IF (A hall call occurs on the upper floor) and (If assigned to a certain car (A), the cars will be concentrated on the upper floor) THEN << Cars with the above property (A) are excluded from being assigned candidate cars. ) (Assign the car with the minimum evaluation value from among the allocation candidate cars) Furthermore. As a simulation example, when an IOF downlink call is registered in the situation shown in Figure 16. We will preserve Cars 2 and 4, which are empty, and choose the car with the best evaluation value (3) from among Cars 1 and 3, which have car calls on the upper floors.
It is recommended to select the number structure) in the allocation basket. If this is followed, allocation can be done taking into account call occurrences in the near future, and long waiting times can be prevented. however,"
The determination that "cars are concentrated on the upper floor" is simply "the car has a car call on the upper floor" or "the car has an assigned call on the upper floor"
This is done only under the following conditions. The structure is not designed to take into account changes in the mutual positional relationship of the cars over time. Therefore, although the cars may eventually reach the upper floor, since the positional relationship with other cars is not clearly defined, it is quite possible that the cars will not actually be ``concentrated in the upper part'', and in that case, empty cars may However, there was a problem in that the wait time was lengthened because of the waiting time.Furthermore, in the near future, after answering the call, the car would be waiting with the door closed. This invention did not take into consideration any prediction of the occurrence of long waiting times.There was a problem that the effect of preventing long waiting times from occurring in the near future was not sufficient.This invention was made to solve this problem. The purpose of this study is to obtain an elevator group management system that can reduce the waiting time for hall calls from now to the near future, taking into account changes in the car arrangement and the possibility of empty cars. Means for Solving] An elevator group management device according to the present invention includes: a hall call registration means for registering a hall call when a hall button is operated;
Allocating means for selecting and allocating a car to be serviced from a plurality of cars in response to the hall call. Determining the running direction of the car. Starting, stopping. and a car control means that controls operations such as door opening and closing, makes the car respond to car calls and the assigned landing calls, and makes the car wait at the floor where it has answered all calls. Alternatively, in an elevator group control device having a waiting means for causing the elevator to travel to a predetermined floor and wait. A predictive empty car that detects, as a predicted empty car, a car that has a door to be answered and is predicted to be in a closed state after a predetermined period of time after the cars that currently have a call to answer respond to the calls one after another. A car detection means.
a car position prediction means for predicting and calculating the car position and car direction after a predetermined period of time when the car sequentially responds to the car call and the assigned hall call from the current time, and based on the predicted car position and the predicted car direction; a means for predicting the number of cars that will be on the predetermined floor or in the predetermined floor area after the elapse of the predetermined time; a first determining means that compares and determines the predicted number of cars predicted to be with a first predetermined value, a second determining means that compares and determines the predicted number of cars with a second predetermined value, and the first determining means and a third determining means for outputting a command for restricting the allocation of the predicted empty car to the hall call or excluding the predicted empty car from being allocated based on the determination result of the second determining means. It is equipped with the following. [Operation] In the present invention, in the allocation operation for a hall call, a predetermined operation is performed using a predicted value of the number of cars that will be on a predetermined floor or a predetermined floor area after a predetermined period of time and a predicted number of empty cars. I do. [Example] Figures 1 to 10 are diagrams showing an example of the present invention. In this example, it is assumed that four cars are installed in a 12-story building. Figure 1 shows the overall configuration.
The group control device (10) and the No. 1 unit that is controlled by it ~
It consists of the car control device for car No. 4 (11) to <<14>. (10A) is a hall call registration means that registers and cancels the hall calls (up calls and down calls) for each floor, and calculates the elapsed time since the hall call was registered, that is, the duration. (IOB) is an expected arrival time calculation means, and is a predicted value of the time required for each car to arrive at the landing (by direction) on each floor. In other words, calculate the expected arrival time. (I OC) is an allocation means that selects and allocates the best car to service a hall call, and performs an allocation calculation based on the predicted waiting time of the hall call and the determination result by the allocation restriction means described later. (IOD) is a car position prediction means that predicts and calculates the car position and car direction after a predetermined time T has elapsed from the current time. (IOE) is a car number prediction means that predicts and calculates the number of cars that will be in a predetermined floor area after a predetermined time T has elapsed based on the predicted car position and predicted car direction. (1 0 F) is a waiting means, and when the car has answered all the calls, it is made to wait at the floor where the call has been answered or a specific floor. (IOG) is an empty car detection means. A car with a closed door and no calls to answer is detected as an empty car. 《10H) is a predictive empty car detection means,
A car that currently has a call to be answered (including a car that is currently responding to a call) that is expected to become empty after a predetermined time T has passed is detected as a predicted empty car. (
IOJ) is an allocation restriction means that determines whether to limit the allocation of hall calls to the predicted empty car or to exclude it from allocation targets. (11A) is a well-known hall call cancellation method. It is installed in the car control device (11) for car No. 1 and outputs a hall call cancellation signal for each floor's hall call. (11B) is a well-known car call registration method. Similarly, register the car call for each floor. (
11C) is a well-known arrival warning light control means. Similarly, the lighting of the arrival forecast lights (not shown) on each floor is controlled (1
1 D) is a well-known running direction control means that determines the running direction of the car. (11E) is a well-known operation control means, which controls the running and stopping of the car in order to respond to the assigned hall call. (11F) is a well-known door control means, which controls the opening and closing of the door. Control.Incidentally, Unit 2~
The car control devices (12) to (14) for the No. 4 car are also configured in the same way as the car control device (11) for the No. 1 car. FIG. 2 is a block diagram of the group management device (10), which is composed of a microcomputer (hereinafter referred to as microcomputer) and an MPU (microprocessing unit) (101). ROM(102).RAM(
1 03), input circuit (104), and output circuit (104)
05). In the input circuit (104). Hall button signals (19) from the hall buttons on each floor. and cars 1 to 4 from car control devices (11) to (14) for cars 1 to 4.
The status signal of the machine is input. The output circuit (105) outputs a signal (20>. to the hall button light built into each hall button) and a command signal to the car control devices (11) to (14) for cars 1 to 4.Next Now, the operation of this embodiment will be explained with reference to Figs. 3 to 8. Fig. 3 shows the group management program stored in the ROM (102) of the microcomputer constituting the group management device (1o). FIG. 4 is a flowchart showing the empty car detection program (33), FIG. 5 is a flowchart showing the car position prediction program (35), and FIG. 6 is a flowchart showing the car number prediction program (36). .Figure 7 is a flowchart showing the allocation restriction Hamatan program <37>, and Figure 8 is a diagram showing a state in which a building is divided into multiple floor areas (zones).First, group management in Figure 3 An overview of the operation will be explained.The input program in step (31) is the hall button signal (19),
Status signals from car control devices (11) to (14) for cars No. 1 to No. 4 (car position, direction. Stop. Traveling, door opening/closing status, car load. This is a well-known device for inputting car calls, hall call cancellation signals, etc. In step (33), the empty car detection program detects as an empty car a car that has no calls to answer and whose door is closed, as shown in Figure 4. In FIG. 4, step (46). That is, steps (42) to (45) are 1
This represents the procedure for detecting whether the car number is currently empty. Car No. 1 has a car call or an autopsy landing call,
If there is no direction and the door is closed, the process goes to step (42) → step (44> → step (45)), where the empty car flag AV1 is set to "1".Other than the above.Step (43) Reset the empty car flag AVI to "0" with Step 47 for Cars 2 to 4.
〉 ~ (49) Detects the car and sets the empty car flag AV2
~AV. Set. In the expected arrival time calculation program in step (34).
Each landing i (1=1.2,3,・
..., 11 are B2, Bl., respectively. 1,... Uphill landing on the 9th floor. i=12.13, ...21.22 is. 10.9 respectively. ...The expected arrival time Aj<1) to the car J (J=
1.2.3.4) Calculate each. Estimated arrival time. For example, it is assumed that it takes 2 seconds for the car to advance one floor and 10 seconds for each stop, and the calculation is performed assuming that the car sequentially runs around all the landings. Note that the calculation of estimated arrival time is well known. In the car position prediction program in step (35>), the predicted car position Fl(T
)~F4(T) and predicted car direction D+(T)~D4(T)
Perform prediction calculations for each car. This will be explained in detail using Figure 5. 1st! In the car position prediction program (35) of I, step (65). That is, steps (51) to (64) are. Predetermined time Tf of No. 1 machine
The prediction of & is the position p+('r) and the predicted car direction Dt(T
) represents the procedure for calculating. When there is a landing call assigned to Car 1, step (51) → step (53)
Here, the terminal floor in front of the farthest assigned landing call is predicted to be the final call floor of Unit 1, and the arrival direction of the car at that floor (downward for the top floor, upbound for the bottom floor) is predicted. direction)
This is also taken into consideration when setting the predicted final call landing h1. Also, when car No. 1 has only a car call without an assigned hall call, step (51) → step (
52> → Proceed to step (54), where the farthest car call floor is predicted to be the final call floor of car No. 1, and the car's arrival direction at that time is also taken into account and set as the predicted final call landing h. ．． Furthermore, again. When the No. 1 car does not have an assigned landing place call or a call, step (51) → step (
52) → Proceed to step (55), where the car location floor of car No. 1 is predicted to be the final call floor, and the car direction at that time is also taken into consideration and set as the predicted final call landing h. After determining the predicted final call landing h1 in this way, it is then determined in step (56) whether or not car No. 1 is empty. When car No. 1 is not empty (AV1 = "O"). In step (57), the predicted value (hereinafter referred to as the predicted empty car time) 1+ of the time required for car No. 1 to become empty is calculated. The predicted empty car time t, is determined by adding the predicted arrival time AI (h1) at the final call predicted landing point h, to the predicted stop time T g ("10 seconds) at that landing. Note that the car position When a floor is set as the final call predicted landing h1, the remaining stop time is predicted according to the car state (running, decelerating, door opening, door opening, door closing, etc.). Set this as the empty car predicted time t.Next,
In step (58), it is determined whether car No. 1 will become empty before the predetermined time T elapses.If the predicted empty car time t1 of car No. 1 is less than the predetermined time T, it is determined whether the car will be empty until the predetermined time T elapses. This means that car No. 1 will be empty in
Predicted car position F+ after the specified time rtjJT on the 1st floor
(T). Also, predicted car direction D+(T)
Set rQJ. The predicted car direction DI (T) is rQJ when there is no direction, ``1'' when it is in the up direction, and ``2'' when it is in the up direction.
indicates the downward direction. Then, in step (60), 1
The predicted empty car flag PAVI of car No. 1 is set to "1." On the other hand, when the predicted car time t, of car No. 1 is greater than the predetermined time T, it is determined that the car has not become empty even after the predetermined time has elapsed. This means that step (58)
→Proceed to step (61), where the expected arrival time A+(i-1) of landing 1-1 and the expected arrival time rWJ of landing 1
A t (i)ffi (A r (i −1
) + T a ≦ T < A I ( i ) + T
s }, the predicted car position F after a predetermined time T has passed. (T), and the same direction as this landing spot i is set as the predicted car direction D, (T). Then, in step (62), the predicted empty car flag P of the first car is
Reset AV, to rQJ. Also, in step (56), the first machine is empty or AV+-
'I J) When. In step (63), the predicted empty car time t, is set to O seconds, and in step (64), based on the predicted final call landing h1, the predicted car position Fl( T). Also, set the predicted car direction D, (T) to ``0''. and. In step (62>), the predicted empty car flag P of car No. 1 is
Reset A V + to “O”. In this way. In step (65), the predicted car position Fl(T) and predicted car direction D+(T) for car No. 1 are calculated, and the predicted empty car is detected. (T), predicted car direction o
x('r)~D4(T). and predicted empty car flag P
AV z to PAV4 are also set in steps {66} to (68), which are the same procedure as step (65). Again in Figure 3. In step (36), the empty car number prediction program calculates the number of empty cars on a predetermined floor or a predetermined floor area after a predetermined time rHIT has elapsed, for example, as shown in FIG. The floor area (
Zone) 21-2. Predicted basket white number N+(T)~
N. Calculate each of (T). This is the 6th rf! 1 will be explained in detail. In the car number prediction program <<36>> in Figure 6. In step (71), the predicted number of cars N, (T) to N@(T) are each initialized to "0", and the car number j and zone number m are each initialized to "1". In step (72), the predicted car position F of car j
Based on J(T) and predicted car direction DJ(T), it is determined whether car No. J is in zone Zm after a predetermined time T has elapsed. When it is predicted that machine j is in zone Zm. In step (73), the predicted number of cars Nm(T) in zone Zm is increased by one. In step (74), the machine number j is incremented by one. In step (75), it is checked whether the determination has been completed for all machines. If not completed, return to step (72). Repeat the above process. After completing the processing of step (72) and step (73) for all machines for zone Zm with zone number m, next. In step <76>, the zone number m is incremented by one, and the machine number j is initialized to "1". and. In the same way, repeat steps (72> to (75>) until machine number j>4. Repeat steps (72> to (75>) for all zones 21 to Z.
When the above-mentioned process is completed, in step (77>), the zone number m>6, and the process of predicting the number of cars for zones z1 to 2 is completed. In steps (78) to (84), the number of empty cars is calculated. N.A.V.
and count the predicted number of empty cars NPAV. At step (78), set the number NAV and white number NPAV to "0".
On the stand. Initialize machine number j to "1". If car No. J is empty, proceed from step (79) to step (80). Here, increase the number of empty baskets NAV by 1. Also. If car number j is predicted to be empty, the process proceeds from step (79) to step (81) to step (82), where the predicted number of empty cars NPAV is increased. In step (83), the machine number j is incremented by one. In step (84), it is checked whether the determination has been completed for all machines. If it has not been completed, return to step (79>) and repeat the above process. In this way, the predicted number of cars for each zone N, (T)
~Ns('r), the number of empty cars NAV, and the predicted number of empty cars NPAV are counted. The process of the car number prediction program (36) ends. In the allocation restriction program in step (37) in the group management 1 program <10) in Fig. 3, when a new hall call C is registered, the floor position of the hall call C, (
Based on T) to N6(T), the number of empty cars NAV, and the predicted number of empty cars NPAV, it is determined whether or not to limit the allocation of cars No. 1 to No. 4 to the hall call C,
Allocation limit evaluation values 21 to P are set respectively to make it difficult to allocate to the new hall call 5C. In addition. w4
The restriction evaluation values P, ~P4 mean that the larger the value, the higher the degree of the allocation restriction. If this value becomes infinite, it is equivalent to excluding it from the allocation target from the beginning. This determination procedure will be explained in detail with reference to FIG. In the quota restriction program shown in Figure 7. The new hall call C belongs to the upper floor zone (Zs or Z.) and there are no empty cars (NAV<1). The number of cars expected to be in the upper floor zone (Z3 or Z4) after the predetermined time T has passed is large < (N 3 (T)" N 4 (T) ≧ N
a), there are predicted empty baskets, but not many <(1≦NAV≦
Nb) and if the traffic condition is such that a hall call is likely to occur in the lower floor zone (Zt or z6), step (81
) → Step (82) → Step (83) → Step (
84) → Step <85) → Step (86>)
Here, the allocation limit evaluation value Pk of predicted empty car k (kε (1, 2, 3.4)) is set to "99999". Car n (nε (1, 2, 3.4) and n≠ k>'s quota limit evaluation value Pn is set to rQJ.Other than the above.In step (87), the quota limit evaluation value P, ~P, of all cars is set to rO''.In this way, the quota limit evaluation value value P1
~P, is set. Group management program (10) in Figure 3
) In the waiting time evaluation program in step (38>), the evaluation values Wl to W4 regarding the waiting time of each landing call when the new landing call C is provisionally allocated to cars 1 to 4, respectively.
perform. The calculation of the waiting time evaluation values W1 to W is well known, so a detailed explanation will be omitted, but for example,
If a machine number is provisionally assigned. Predicted waiting time U(i) for each hall call i when car No. 1 is provisionally allocated (i=1.2.
・．． 22: If no hall call is registered, r01 seconds) is calculated, and the sum of these squared values, that is, the waiting time evaluation value W, ・U(1)"+ U (2 >"+...+
U(22) Calculate as desired. next. Step (39
), the allocation limit evaluation values P+ to P, and the waiting time evaluation values W, -W. Select one car to allocate based on . In this example, the overall evaluation value EJ when new hall call C is provisionally assigned to car j is expressed as Ej=Wj
Obtained by +k-PJ (k: constant). The car with the minimum overall evaluation value EJ is selected as the officially assigned car. The assignment command and forecast command corresponding to hall call C are set in the assigned car. moreover. In the standby operation program in step <<40>>,
This is to prevent the cars from clumping in one place when all the hall calls are answered and there are empty cars. Determines whether the empty car is to be left on standby at the last called floor or on a specific floor, and when it is determined to be on standby on a specific floor, a standby command is set for the empty car to run to that specific floor. do. lastly. In the output program of step (41), the hall button light signal (20
) to the boarding area. The car control device (11
) to <14). The above group management program (31〉~(4)》) is repeatedly executed in this manner.Next, the operation of the group management program (10) in this embodiment is further explained in Fig. 9 and 105?I. A detailed explanation will be given.For the sake of simplicity, Figures 12 to 15
This will be explained using the example shown in the figure. In the state shown in Fig. 12, as shown in Fig. 9, the down call of 91I1 (9
d> is registered. In addition. 5th floor up call (5
The duration of u> is 10 seconds. At this time. When the 9th floor down call (9d) is provisionally assigned to car A, the 9th floor down call (9d) is
The predicted waiting times for 9d) and the 5th floor up call (5u) are 24 seconds and 16 seconds, respectively. Waiting time evaluation value WA at this time
Then, WA=24”+16”=832. On the other hand, when provisionally assigned to car B, the predicted waiting times for the down call (9d) and the up call (5u) on the 5th floor of 9ll1 are 28 seconds and 16 seconds, respectively, and the waiting time evaluation value W. is W. =2
8" + 16' = 1040. Therefore, in the conventional allocation method, 9th floor down call (
9d) is assigned to car A. Now. The car positions of car A and car B after a predetermined time T (T=20 seconds) have passed are car A' and car B', as shown in FIG. Looking back, the predicted number of cars at this time is N. (T
)=2 units, Nl(T)=N1(T)=N4(T)=Ns
(T)=N@(T)・0 units. Number of empty baskets NAV=
O cars, predicted number of empty cars NPAV = 1 car. In addition. In this example, a car with no direction is considered to be in the upward direction, but the direction can be determined as appropriate depending on the car position. Also. In this example. If constant values Na = 2 units and Nb = 1 unit. N3(
T) = 2 cars corresponds to the case where all the cars are in one zone. In step (86> of the quota limit program (37) in FIG. 7, the quota limit evaluation value PA=
99999. Allocation limit evaluation value P for car B. =0 is set. therefore. The overall evaluation value is EA=WA+PA=
99999=10831,E. =W. +Ps=1040
+O=1040. So EA>E-. finally 9
The downstairs call (9d) is assigned to car B. In the conventional allocation method, the car will be allocated to car A, and in the near future the car will be in a dangling operation, causing long-waiting calls to occur. but. By allocating the car to car B in consideration of the car arrangement after the predetermined time T has elapsed, this kind of dangling operation can be prevented. As explained above. In the above example. The car responds to calls sequentially from the current moment, and predicts and calculates the car position and car direction after a predetermined period of time has elapsed. Furthermore, based on these, the predicted number of empty cars and the number of cars after a predetermined period of time in each zone are calculated, and the allocation restriction operation for the predicted empty cars is performed according to the predicted number of cars. Cars will no longer be concentrated in one place, and the waiting time for hall calls can be reduced from now to the foreseeable future. In addition. In the above embodiment, when predicting the car position and car direction after the elapse of a predetermined time T, first predict the floor where the car will become empty after answering the final call and the time required until then. In the above, the car position and car direction after a predetermined time T has elapsed are predicted. This is because we assumed that when a car becomes empty, it will remain on standby on that floor. If it has been decided that empty carts will always be kept on standby on a specific floor. It is sufficient to predict the car position and direction as if the car is to be run to a specific floor. It is also possible to predict the car position and car direction by taking into account new calls that will occur before the predetermined time T elapses. Furthermore, the calculation direction of the final call predicted landing is not simplified as in this embodiment, but may be finely predicted based on the statistically determined probability of occurrence of a car call or a landing call. In addition, in the above embodiment, the building is divided into zones as shown in Fig. 8, but the number of floors and the number of installed cages may vary. Time zone and purpose of each floor (main floor, dining room floor, meeting room floor, transfer floor)
It is also easy to change the way zones are set up depending on the situation. Also, it is not necessary to decide the zone by considering the direction of the landing area. Furthermore, in the above embodiment, (a) When the new hall call C belongs to the upper floor zone. ■ There is no empty car, ■ There are many cars expected to be in the upper floor zone after the predetermined time T has elapsed, and ■ There is one predicted empty car. There are not many, and the traffic conditions are such that it is easy for a landing call to occur in the lower floor zone. When all of the conditions ■ to ■ are satisfied, an allocation restriction evaluation value (〉O》) is set for each of the predetermined predicted empty cars to restrict allocation to new hall call C. However, .The conditions for setting the allocation limit evaluation value are not limited to these. (Expression) When the new landing call C belongs to the lower floor zone. ■ There is no empty car. ■ After the specified time lWIT has elapsed, the lower floor The number of cars expected to be in the zone is large. ■ There is one predicted empty car, but not many, and ■ The traffic conditions are such that hall calls are likely to occur in the upper floor zone. (c) When the new hall call C belongs to the intermediate floor zone.■ There are no empty cars, and ■ There are no cars expected to be in the intermediate floor zone after the predetermined time tlIT has passed. The following conditions (c) can also be applied: ■ There is one predicted empty car, but not many. ■ The traffic condition is such that a hall call is likely to occur in the upper floor zone or lower floor zone. ．Furthermore, 1
We use the flipside of the fact that cars are concentrated in one zone. (2) When new hall call C is registered, ■ 1 empty car
There is no platform. ■ Upper floor zone, lower/downward floor zone. Or in any intermediate floor zone to which the new hall call C does not belong. And there is a zone in which there is no car expected to be there after the predetermined time T has elapsed, and ■ there is one predicted empty car, but not many, and ■ a hall call occurs in a zone that satisfies the above ■. Traffic conditions are easy. Condition (2) can also be applied. In addition. When this condition (2) is satisfied, it is desirable to restrict the allocation to the predicted empty car that is closest to the zone that satisfies the above (■) among the predicted empty cars. Here, the conditional term ■ in the above conditions (a) to (ii) is
This is provided to maximize the effect of preserving predicted empty baskets. Even if the traffic condition condition (■) is not present, the effects of this invention will not be significantly impaired. In addition, although the condition regarding the number of empty cars in condition item (2) in conditions (a) to (2) above is set to the number of empty cars = 0 cars, the present invention can also be applied when the number of empty cars is 1 car. can. For example, <<E) When new landing call C belongs to the upper floor zone, ■ There is one empty car in the upper floor zone, and ■ For a predetermined time T.
The number of cars expected to be in the upper floor zone after a light accident is large; ■ There is one predicted empty car, but not many. If the condition (e) is satisfied, it is also possible to restrict the allocation of a predetermined predicted empty car to the new landing call C. This condition (e) is valid because there is a landing immediately in the lower floor zone. This is the case when the traffic condition is not congested such that a call is generated.In other words, empty cars are effectively used for the new landing call C to be serviced in a short time, and the lower floor zone is registered in the near future. It is intended that a predicted empty car that will become empty in the near future will be serviced when the deaf boarding hall calls.Furthermore, the conditional terms in the above conditions (a) to (e)
in. ``The predicted number of empty cars is 1 or more, and a constant value Nb
This is because when the number of predicted empty cars is large, it is not necessarily necessary to preserve predicted empty cars through allocation restrictions. Eliminate the condition of "below a certain value Nb". In the process of selecting predicted empty cars to be allocated, it is possible to select a specific car or two cars (for example, select a car that is likely to become empty the earliest using predicted empty car times 1, ~t). , or predicted car position Fl(T) to F.
(T) and predicted car direction DI(T) to D. A similar effect can be obtained by restricting the allocation of (T) to select the car closest to the lower floor zone. There are various other possible conditions for restricting allocation to predicted empty cars based on the predicted number of cars, but it is clear that this can be easily achieved like condition (a) shown in Figure 7. Furthermore, in addition to the above (a) to (e), when multiple conditions are set depending on each situation, it is possible that two or more conditions may be in place at the same time. In such a case, the problem arises as to which conditions should be used to decide which baskets are subject to allocation restrictions. There are various ways to solve such problems, one of which is to use the well-known fuzzy theory. For example, the conditions are configured as detailed in the above-mentioned "Elevator Group Control Device" (1988 Electrical and Information Related Society. Proceedings Vol. 2 P2-1 17-120). The degree to which each conditional term is satisfied is expressed by a membership function as a number between O and 1. A method that uses them to select the minimum value for an AND combination or the maximum value for an OR combination, calculates the confidence of the condition itself, and finally selects the one condition with the highest confidence. It is. There is also a method in which the method of restricting allocation corresponding to each condition is weighted according to the degree of certainty, and the allocation of cars is restricted accordingly. It is clear that the present invention can be applied to conditions in such a system. in this way. Any condition may be used as long as it uses the predicted number of cars that will be concentrated on a given floor or a given floor area and the predicted number of empty cars that will be empty in the near future. Furthermore, in the above embodiment. As a means of restricting allocation to hall calls. A quota limit evaluation value that is larger than other baskets is set for a specific basket. This was weighted and added to the waiting time evaluation value to obtain a comprehensive evaluation value, and a method was used in which the car with the smallest overall evaluation value was selected as the officially allocated car. In this way, the quota limit evaluation value is combined with other evaluation values to comprehensively evaluate and allocate. This is nothing more than giving priority to the basket with the lowest quota limit evaluation value. In other words, a car with a large allocation limit evaluation value is more difficult to allocate than other cars. Also. The means for restricting allocation to hall calls is not limited to the above embodiment. A method may also be used in which cars that meet the allocation restriction conditions are excluded in advance from candidates for allocated cars. Furthermore, in the above embodiment. Although the waiting time evaluation value is the sum of the squares of the predicted waiting times for hall calls, the method of calculating the waiting time evaluation value is not limited to this. For example, even if a system is used in which the sum of the predicted waiting times of a plurality of registered hall calls is used as the waiting time evaluation value, or a method in which the maximum predicted waiting time is used as the waiting time evaluation value, the present invention is still applicable. It is obvious that it can be applied. of course. The evaluation items to be combined with the quota limit evaluation value are not limited to waiting time, but may also be combined with evaluation indicators such as unpredictedness or fullness. In addition, in the above embodiment, the car position and car direction after a predetermined time T for one type of predetermined time T are predicted for each car, and the allocation limit evaluation value is calculated based on this. The predetermined time Tl, T2,
..., Tr (TI<T2<...<Tr), predict the car position and car direction after a predetermined time for each car, and further predict multiple types of predetermined time TI,
For T2, ..., Tr, the predicted car empty number N m (T 1) to Nm (Tr) after the elapse of a predetermined time is calculated for each zone Z.
m(m=1.2, . . . calculate each for 1. Also, predict the number of empty cars NPAV(TI) ~ NPAV(
Tr) respectively. and. Each combination {N
, (TI), N2 (T2)s..., N P A V (
TI)). (N + (T2).N * (T2).-
- ・,NPAV(T2Ns...(N+(Tr).
N*(Tr>,-...NPAV(Tr)). The allocation limit evaluation value P (TI). P
(T2). ...P (Tr). Add weights. That is, P=kl・P(Tl)+kP(T2)+
ky”P(Tr).(However, kl, k2,..., kr
is calculated according to the formula 〉weighting coefficient〉. It is also easy to set the final allocation limit evaluation value P.
in this case. Don't just focus on the cage arrangement at a certain point T. T.I. This is because the cage arrangement at multiple points in time, T2, ..., Tr, will be evaluated from a global perspective.
It will be possible to further reduce the waiting time for hall calls from now on into the near future. In addition. The above weighting coefficient kl,
k2,...,kr are, for example, as shown in FIG.
There are several ways to set this up, depending on which point in time you want to place more emphasis on cage placement. It can be selected appropriately depending on traffic conditions, building characteristics, etc. [Effect of the invention] This invention is as explained above. A hall call registration means for registering a hall call when a hall button is operated. Allocating means for selecting and allocating a car to be serviced from a plurality of cars in response to the hall call, and determining the running direction, departure, and stop of the car. and controls operations such as opening and closing doors. A car control means for causing the car to respond to the car call and the assigned landing call.
When the car finishes answering all calls, do you have it wait on the floor where it finished answering? Alternatively, in an elevator group control device having a waiting means for causing the elevator to travel to a predetermined floor and wait. A predictive empty car that detects, as a predicted empty car, a car whose door is predicted to be in a closed state after a predetermined time has elapsed by sequentially responding to calls that have a call to be answered at the present time. a car detecting means; and a car position predicting means for predicting and calculating the car position and car direction after a predetermined period of time when the car responds to the car call and the assigned landing call from the present moment. car empty number prediction means for predicting the presence or absence or number of cars that will be on a predetermined floor or predetermined floor area after the elapse of the predetermined time based on the predicted car position and the predicted car direction; a first determination means for comparing and determining the predicted number of cars that are later predicted to be on the predetermined floor or within the predetermined floor area with a first predetermined value; and comparing and determining the predicted number of cars with a second predetermined value. outputting a command to limit the allocation of the predicted empty car to the hall call or exclude it from being allocated based on the judgment results of the second judgment means and the first judgment means and the second judgment means; and a third determination means. By taking into consideration changes in car arrangement over time and the possibility of empty cars, this system has the effect of shortening the waiting time for hall calls from now to the near future. 4. Brief Explanation of Sections FIG. 1 is a block diagram of an elevator group control system according to an embodiment of the present invention. Figure 2 is an elliptic block diagram of the group management device (10) in Figure 1, Figure 3 is a flowchart of the group management program (10), and Figure 415! I is a flowchart of the empty car detection program (33) in Figure 3, and Figure 5 is the car position prediction program (36) in Figure 3.
Flow chart diagram. Figure 6 is a flowchart of the car number prediction program (36), Figure 7 is a flowchart of the quota restriction program (37) of Figure 3. Figure 8 is a diagram showing the zoning of buildings, Figures 9 and Fig. 10 is a diagram showing the relationship between calls and car positions, and Fig. 11 is a diagram explaining another embodiment of the present invention. Figs. 12 to 16 show conventional elevator group control devices, and each This is a diagram showing the relationship between call and car position. In the diagram, (IOA)
... Hall call registration means, (IOC) ... Assignment means. (
IOD＞...Car position prediction means. (IOE)... Car number prediction means, (IOF)... Standby means. (IOG
)...Empty car detection means. (IOH)...Predictive empty car detection means, (IOJ)...Allocation restriction means. (11)～
(14)... Car control means. In addition. The same symbols in each figure indicate the same or equivalent parts.

Claims (1)

【特許請求の範囲】[Claims] 乗場釦が操作されると乗場呼びを登録する乗場呼び登録
手段と、前記乗場呼びに対して複数のかごの中からサー
ビスすべきかごを選択して割り当てる割当手段と、かご
の運行方向決定、出発、停止、および戸開閉等の運転制
御を行い、かごをかご呼びと前記割当乗場呼びに応答さ
せるかご制御手段と、かごがすべての呼びに答え終わる
と答え終わった階床で待機させるか、もしくは所定の階
床へ走行させて待機させる待機手段とを有するエレベー
タの群管理装置において、現時点で応答すべき呼びを持
っているかごが呼びに順次応答して所定時間経過した後
に応答すべき呼びがない状態になっていると予測される
かごを予測空かごとして検出する予測空かご検出手段と
、前記予測空かごが前記乗場呼びに割り当てるのを制限
するかもしくは割当対象から除外する指令を出力する割
当制限手段とを備えたことを特徴とするエレベータの群
管理装置。A hall call registration means for registering a hall call when a hall button is operated; an allocating means for selecting and allocating a car to be serviced from a plurality of cars in response to the hall call; determining the direction of operation of the car, and determining the departure of the car. a car control means that controls operations such as , stopping, and door opening/closing, and causes the car to respond to car calls and the assigned landing calls; and after the car has answered all the calls, causes the car to wait at the floor where it has answered, or In an elevator group control device having a waiting means for causing cars to run to a predetermined floor and wait, the cars that have a call to be answered at the present time respond to the calls one after another, and after a predetermined period of time has elapsed, the car to be answered is received. Predicted empty car detection means for detecting a car that is predicted to be empty as a predicted empty car, and outputting a command to limit the allocation of the predicted empty car to the hall call or exclude it from being allocated. What is claimed is: 1. An elevator group management device comprising: allocation restriction means.