JPS6247787B2 - - Google Patents

Description

【発明の詳細な説明】[Detailed description of the invention]

この発明は複数台のエレベータのかごを一群と
して管理する装置に関するものである。 従来、エレベータ・サービスの評価を行う場
合、多くは各乗場待客の待時間の長さが問題にさ
れてきた。すなわち、各乗場待客の待時間が平均
的に短く、しかも長時間待ち呼びが少ないことが
良いサービスであるとされてきた。しかし、最近
のように乗場呼びにサービスするように決められ
たかごを、サービスかごとして乗場待客に前もつ
て案内表示（以後これを「予報」という）するよ
うな方式になると、上記予報されたかごが途中で
満員となつて上記呼びを満員通過したり、また他
のかごがかご呼びなどに応答し、予報かごよりも
先に上記呼びにサービス（以後この状況を「予報
外れ」という）したりするなど、乗場待客の期待
を裏切り乗場待客を混乱させるというエレベー
タ・サービスの上での不具合が新たに生じること
となつた。 従来、上記したような不具合を減少させるため
に、種々の手段が考えられて来たが、待時間、満
員通過、予報外れの各問題点をそれぞれ別個に取
り上げ処理していたために、乗場全体として、乗
場待客への総合的なサービスの質の良さという点
では不十分であつた。一例として、待時間と満員
通過の危険性だけで割当かごを選択する場合を考
えてみると、待時間が最も短く、しかも満員通過
の危険性が最も小さいかごを割当かごとして選択
すれば最良の割当であると言えるが、一般に上記
条件を満たすかごが常に存在しているわけではな
い。上述のような最良のかごがないとき、従来の
方式では満員通過の危険性の確率が所定値（例え
ば0.3）以上のかごには割り当てないようにし
て、残つたかごの中で最も近いかごを割当かごと
して選択するという手段がとられている。しかし
ながら、満員通過の危険性の確率が0.31で待時間
は10秒のかごと、満員通過の危険性の確率が0.29
で待時間は100秒のかごとがある場合には、満員
通過の危険性にはそれ程の差はないので、少しだ
け満員通過の危険性は増すが前者のかごに割り当
てた方がよい。このように、従来の方式は満員通
過の危険性と待時間の長さとを別々に判断するた
めに乗場待客の立場から見るとサービスにきめ細
かさが欠けていた。したがつて、乗場待客へのサ
ービスの質を向上するには、まず乗場待客へのサ
ービス状態を、待時間、満員、予報外れ等を互い
に関連させた上で、正確に把握する必要がある。
その一つとして、乗場待客への心理的影響によつ
てサービス状態を表わすことが考えられる。すな
わち、待時間、満員、予報外れという異なつた性
質のものを待時間の長さによる待客への心理的影
響（焦燥感、不安感、不審感、不満など）、満員
通過、満員積残しが生じる危険性の大きさと、そ
れが生じたときの待客への心理的影響の大きさ、
及び予報外れが生じる危険性の大きさとそれが生
じたときの待客への心理的影響の大きさという同
一次元のものに数値的に置き換えることにより、
乗場待客へのサービス状態を表わすのである。 この発明は上記欠点を改良するもので、各乗場
待客のサービス状態を総合的かつ適確に把握し
て、最良と思われるかごを割り当てることによ
り、乗場待客に対するサービスの質を向上させる
ことのできるエレベータの群管理装置を提供する
ことを目的とする。 以下、第１図〜第９図によつてこの発明の一実
施例を説明する。なお、説明の便宜上、３台のか
ごが６階建の建物に設置されている場合について
示すが、設置台数及び階床数には関係なく適用で
きることは言うまでもない。 実施例の説明に先立つて、満員確率及び予報外
れ確率の演算について説明する。 満員通過、満員積残し、予報外れ等の危険性を
表わす手段として、それぞれ満員通過が生じる確
率（以後満員通過確率という）、満員積残しが生
じる確率（以後満員積残し確率という）、及び予
報外れが生じる確率（以後予報外れ確率という）
を使用するやり方がある。（なお、以後満員通過
確率と満員積残し確率を総称して単に満員確率と
いうことにする。）一般に、各乗場での予測待客
数や予測かご負荷にはある範囲のばらつきがある
ために、例えば、予測かご負荷がかご定員の60％
程度でも実際には満員（100％）になることがあ
り、逆に予測かご負荷がかご定員の100％であつ
ても実際には満員にならないことがある。また、
かごが各乗場に到着するまでの予測時間、すらわ
ち到着予想時間においてもばらつきがあるため
に、例えばかごａの到着予想時間がかごｂの到着
予想時間より小さくてもかごｂの方が先に到着す
ることがあり得る。このように予測値にはばらつ
きがあるので、そのばらつきを考慮して満員や予
報外れを適確に予測する必要がある。この予測値
のばらつきを考慮したものが満員確率であり予報
外れ確率である。以下、満員確率及び予報外れ確
率の演算について説明する。 ある乗場での待客数の予測値（平均値）をｋ
人、この乗場でｉ人待つている確率をｆ（ｉ，
ｋ）、かご到着時にかごが満員になるまで乗車可
能な人数（以後「乗込許容人数」という）の予測
値（平均値）をｍ人、この乗場での乗込許容人数
がｊ人である確率をｇ（ｊ，ｍ）とすれば、この
乗場で待客を積み残して出発する確率、すなわち
満員積残し確率ｑは、待客ｉ人が乗込可能人数ｊ
人を越える確率と等しいので であり、この乗場を満員で出発する確率ｐは で与えられる。ただし、No.はかご定員を表わす。
したがつて、かごが満員で出発したとき次の階に
かご呼びがなければ通過してしまうので、次の階
にかご呼びを持つ確率をｕとすると、次の階を満
員通過する確率ｒは で求めることができる。 今、かご位置階をSo、かごと同方向前方の乗
場をかごに近い方からS₁，S₂，…Ｓ_o階とし、Ｓ_o
階までかご呼びがない場合を考える。乗場S₁，
S₂，…，Ｓ_oの待客数l₁，l₂，…ｌ_oが平均値k₁，
k₂，…，ｋ_oのポアソン分布であるとき、待客数
の和 は加法定理により、平均値 のポアソン分布に従うので、S₁，S₂，…，Ｓ_o階
の待客数の和Ｌ（ｎ）は exp（−Ｋ（ｎ））・Ｋ（ｎ）^ｊ／ｊ〓 （ｊ＝０，１，２，…） …… の確率分布を持つことになる。また、S₁階での乗
込許容人数Ｎは Ｎ＝（かご定員）−（現在のかご内人数） …… であり、Ｓ_o階までの途中の階で一人も降車しな
いと仮定すると、Ｓ_o階でかごが満員となり待客
の積残しが生じる確率ｑ_oは となる。もし、Ｓ_o階までにかご呼びがあつた
り、S₁，S₂，…Ｓ_o-1階で乗車した待客が途中で
降車することを考えて乗込許容人数Ｎを予測し、
式を利用すればより正確に確率ｑ_oを計算でき
ることは言うまでもない。また、上述の説明はか
ごと同方向前方の乗場についてであつたが、かご
と逆方向の乗場、かごと同方向背後の乗場につい
ては Ｎ＝（かご定員） …… として計算する他は全く同様に考えることができ
る。 また、式では計算式が複雑なので近似式とし
て ｑ_o＝Ａ｛Ｌ（ｎ）＋Ｂ｝／Ｎ＋Ｃ−Ｄ …… ＝ｑ_o-1＋Ａ／Ｎ＋Ｃ・ｌ_o …… を与えることができる。ただし、Ａ，Ｂ，Ｃ，Ｄ
は定数値（例えばＡ＝２，Ｂ＝10，Ｃ＝９，Ｄ＝
２）で、ｑ_oが１，０を越えるときはｑ_o＝１，０
に、ｑ_oが０より小さいときにはｑ_o＝０という補
正が必要である。式はまた途中階で降車する乗
客を考慮して ｑ_o＝ｑ_o-1＋Ａ／Ｎ＋Ｃ｛（Ｓ_o階での予測乗車人数）−（Ｓ_o階での予測降車人数）｝ …… とも近似計算することができる。 また、式を近似し、式で求めたｑ_oを用い
て、Ｓ_o+1階を満員通過する確率ｒ_o+1を ｒ_o+1＝ｑ_o・（１−ｕ） …… と計算することができる。 したがつて、かご位置階又は終端階での乗込許
容人数Ｎが分かれば、式により順々に各階床の
満員積残し確率と満員通過確率を求めることがで
きる。 次に、現在のかご位置階から各階床にかごが到
着するまでの時間の予測値を到着予想時間という
が、その到着予想時間は途中の乗場での乗客の乗
降時間、戸閉に要する時間、加減速に要する時
間、定格速度で走行する時間等により決まる。し
たがつて、ある時点である乗場に対する到着予想
時間は、途中の停止予定階での乗車人数、降車人
数、新たに生じる呼びなどの様々な要因によつて
ある範囲でばらつきをもつことになる。 今、ある乗場に乗場呼びが登録されていて、そ
の呼びに対するサービスかごとしてかごａが予報
されているとする。かごａが上記乗場に到着する
までの時間ｔの分布を確率密度関数ｈ（ｔ）（ｔ
ｏ）で表わすとき、この確率密度関数ｈ（ｔ）
は ∫^∞ _ｐｈ（ｔ）dt＝１ …… という条件を満たしていて、実際にかごが上記乗
場に時間T₁とT₂（T₁＜T₂）の間で到着する確率
Ｒ（T₁，T₂）は Ｒ（T₁，T₂）＝∫^Ｔ２ _Ｔ１ｈ（ｔ）dt …… で与えられる。また、他のかご（その一つをかご
ｂとする）の上記乗場への到着予想時間ｔの分布
も同様にして確率密度関数ｖ（ｔ）で表わすと、
第１図からも分かるようにかごｂがかごａよりも
先に到着する確率Ｗは Ｗ＝∫^∞ _ｐ｛∫^ｘ _ｐｖ（ｔ）dt｝ｈ（ｘ）dx …… で求めることができる。したがつて、かごｂが上
記乗場呼びと同方向でかご呼びにより停止する確
率をｅとすると、かごｂが予報されたかごａより
も先に上記乗場呼びに応答する確率、すなわち予
報外れ確率zbは zb＝Ｗ・ｅ …… と求められる。予報されていないかごＣ、につい
ても同様に予報外れ確率Zc，が得られると上記
乗場呼びの予報が外れる確率Ｚは Ｚ＝１−（１−Zb）（１−Zc） …… であり、続けて２度、３度と予報が外されること
を考慮して、予報外れ回数の期待値Z′で予報外れ
の可能性を表現すると Z′＝１×Zb＋１×Zc …… となる。 ところで、式で表わされる先着確率Ｗを簡単
に求めるものとして、到着時間の分布を一様分
布、指数分布、正規分布などに仮定して計算する
手段もあるが、より簡単に求める手段の一つとし
て、到着予想時間が長くなればそれだけばらつき
も大きくなるだろうという予想のもとに、 Ｗ＝1.5−ｔｂ／ｔａ …… という式で演算することもできる。ただし、taは
予報されているかごａの到着予想時間、tbは予報
されていないかごｂの到着予想時間であり、Ｗ＜
０，０のときはＷ＝０，Ｗ＞１，０のときは、Ｗ
＝１と補正する必要がある。 第２図中、ａ〜ｃはかご定員８人のかごで、そ
れぞれかご負荷は300Kg，420Kg，300Kgである。
１ｃはかごｃ内で登録されたかご呼び、３ａ〜３
ｃはそれぞれかごａ〜ｃ内で登録された３階のか
ご呼び、４ｕは４階の昇り呼びでまだどのかごに
も割り当てられていず、呼びが登録されて５秒経
過しているものとする。５ｕはかごａに割り当て
られた５階の昇り呼びで呼びが登録されて24秒経
過しているものとする。 第３図中、１０はかご制御装置、１１は乗場呼
びを登録する装置、１２はかごａ〜ｃごとに割り
当てられた乗場呼びを記憶する割当記憶装置、１
３はかごに割り当てるための乗場呼びを一つ選択
する乗場呼び選択装置、１４は上記乗場呼び選択
装置１３により選択された乗場呼びをかごに割り
当てる割当装置、１５ａ〜１５ｃは割当装置１４
に設けられ割り当てるべく選択された乗場呼び
を、各かごａ〜ｃに仮りに割り当てたときの割当
呼び信号を出力する仮割当装置、１６ａ〜１６ｃ
は同じく割当装置１４に設けられ各かごａ〜ｃに
呼びを仮りに割り当てたときの総合評価値を出力
する仮割当評価装置、１７ａは仮割当評価装置１
６ａに設けられたサービス予測装置で、各かごａ
〜ｃに対して各階床・方向別ごとに予測待時間、
満員確率、予報外れ確率をそれぞれ予測し出力す
る。１８ａは同じく仮割当装置１６ａに設けられ
たサービス状態評価装置で、各階床・方向別ごと
のサービス状態を演算し出力する階床別評価装置
とそれらを組合せて全体のサービス状態を求め、
総合評価値を出力する総合評価装置とからなる。
１９は割当装置１４に設けられた本割当装置で、
各かごａ〜ｃの仮割当装置１６ａ〜１６ｃの出力
の中から最小の総合評価値を持つかごを選択し正
規に呼びに割り当てる。 第４―Ａ図は乗場呼び選択装置１３の回路図の
一例であり、４階昇り方向の乗場についての演算
回路で、他の乗場についても同様の演算回路が必
要である。図中、２０ａ〜２０ｃは割当記憶装置
１２の出力信号で４階昇り呼びがそれぞれかごａ
〜ｃに割り当てられているときのみ「１」となる
割当呼び信号（一つの呼びに対して１台のかごを
割り当てるようにしているので、信号２０ａ〜２
０ｃのうち２個以上が同時に「１」となることは
ない）、２１は乗場呼び登録装置１１の出力信号
で、４階昇り呼びが登録されているときのみ
「１」となる乗場呼び登録信号、２２は第４―Ｂ
図に示すように４階昇り方向の乗場に対応した時
刻のみ周期的に「１」となるパルス列信号、２３
はノアゲート、２４はアンドゲート、２５は乗場
呼び選択装置１３の出力信号で、４階昇り呼びが
選択されたときのみ「１」となる選択乗場呼び信
号である。 第５図はかごａのための仮割当装置１５ａの４
階昇り方向の乗場についての回路図の一例で、他
の乗場についても同様の回路が必要である。かご
ｂ及びｃに対応する仮割当装置１５ｂ及び１５ｃ
も同様の回路で構成される。図中、２６はオアゲ
ート、２７ａは仮割当装置１５ａの出力信号で、
４階昇り呼びがかごａの割当呼びであるか、又は
かごａに仮割当された呼びであるときのみ「１」
となる割当呼び信号である。 第６―Ａ図及び第６―Ｂ図はサービス予測装置
１７ａの回路図の一例である。第６―Ａ図は予測
待時間を演算する回路の一部であり、図中２８は
かごａによる４階昇り呼びの予測待時間を演算す
る回路である。他の呼びについても同様の回路が
必要である。２９はかごが各乗場に昇り方向及び
降り方向で到着するまでの時間をそれぞれ予測演
算する到着予想時間演算装置で、３０ａはかごａ
が４階に昇り方向で到着するまでの到着予想時間
信号、３１はノツトゲート、３２はＩ点の入力信
号が「０」から「１」になると時間を計数し始
め、Ｒ点に「１」の信号が入ると時間を零にリセ
ツトするようなタイマで、乗場呼びが登録されて
から経過した時間を出力する。３３は加算器、３
４ａはかごａによる４階昇り呼びの予測待時間を
表わす信号（予測待時間信号）である。 第６―Ｂ図はかごａの４階昇り方向の乗場につ
いての回路図の一例で、式を実現した回路であ
る。他の乗場についても同様の回路が必要であ
る。図中、３５ａはかごａが３階を昇り方向で出
発するときの満員積残し確率を表わす信号（演算
回路は図示しない）、３６ａはかごａが４階にか
ご呼びに応答して昇り方向で停止する確率を表わ
す信号（演算回路は図示しない）、３７は１，０
を表わす定数値信号、３８はＹ点の入力からＸ点
の入力を差し引いた値を出力する減算器、３９は
乗算器、４０ａはかごａの４階昇り呼びに対する
満員通過確率信号である。 第７図はサービス状態評価装置１８ａにおける
階床別評価装置の回路図の一例で、４階昇り呼び
に対応する回路である。他の乗場呼びについても
同様の回路が必要である。図中、３４ｂ，３４ｃ
はそれぞれかごｂ及びかごｃにおける４階昇り呼
び４ｕの予測待時間信号、４０ｂ，４０ｃはそれ
ぞれかごｂ及びかごｃにおける４階昇り呼び４ｕ
の満員通過確率、４１ａ〜４１ｃはかごａ〜ｃに
おける４階昇り呼び４ｕの満員積残し確率信号
（演算回路は図示しない）、４２ａ〜４２ｃはかご
ａ〜ｃにおける４階昇り呼び４ｕの予報外れ確率
信号（演算回路は図示しない）、４３〜４６はG₁
〜G₃点の中で、「１」の信号が入力されている点
に対応するI₁〜I₃点の入力信号を選んで信号４３
ａ〜４６ａとして出力する選択回路で、例えば
G₁点に「１」の信号が入力されている場合にはI₁
点の入力信号を出力する。なおG₁〜G₃点にいず
れも「０」の信号が入力されているときは常に零
を出力する。４７〜５０は乗算器、５１は加算
器、５２〜５５は定数値信号で例えばそれぞれ
１，３０，５０，２０と設定される。５６ａは４
階昇り呼び４ｕに対するサービス状態評価値信号
である。 第８図はサービス状態評価装置１８ａにおける
総合評価装置の回路図の一例で、図中５７ａ，
…，５８ａ，…，５９ａは階床別評価装置の出力
信号で、それぞれ１階昇り呼び，…，５階昇り呼
び，…，２階降り呼びに対するサービス状態評価
値信号、６０〜６３は階床別に重みをつけるため
の定数値信号であるが、ここではすべて１，０と
設定されているものとする。６４〜６７は乗算
器、６８は加算器、６９ａは総合評価値信号であ
る。 第９図は本割当装置１９の回路図の一例で４階
昇り呼びに対応する回路である。他の乗場呼びに
ついても同様の回路が必要である。図中、６９
ｂ，６９ｃはそれぞれかごｂ及びかごｃに４階昇
り呼び４ｕを割り当てたときの仮割当評価装置１
６ｂ及び１６ｃの出力で総合評価値信号、７０は
Ia〜Ib点の入力信号の中から最小値を持つものを
選択し、それに対応した信号７０ａ〜７０ｃを出
力する最小値選択回路である。例えばIb点の入力
信号６９ｂが最小の場合、出力信号７０ｂのみが
「１」となり、他の信号７０ａ，７０ｃは「０」
と出力される。７１〜７３はアンドゲート、７４
ａ〜７４ｃはそれぞれかごａ〜ｃに対する割当記
憶指令信号で、例えば信号７４ａが「１」のとき
割当記憶装置１２には４階昇り呼びがかごａの割
当呼びとして記憶される。 今、第２図に示されるように４階昇り呼び４ｕ
をかごに割り当てる場合について具体的に各回路
の動作を説明する。 まず、第４―Ａ図において、４階昇り呼び４ｕ
はどのかごにもまだ割り当てられていないので、
割当呼び信号２０ａ〜２０ｃはいずれも「０」で
あり、ノアゲート２３の出力信号は「１」とな
る。また乗場呼び登録信号２１も「１」となつて
いる。一方、パルス列信号２２はある周期で
「１」のパルスをもつ信号であるからパルス列信
号２２が「１」になつたときのみアンドゲート２
４の出力信号、すなわち選択乗場呼び信号２５は
「１」となつて４階昇り呼び４ｕが割り当てるべ
き呼びとして選択されることになる。このとき他
の乗場呼びに対応する選択乗場呼び信号はすべて
「０」となつている。 前に説明したように、割当装置１４は乗場呼び
選択装置１３によつて選択された乗場呼びをかご
ごとに仮りに割り当ててみて、そのときの各乗場
呼びのサービス状態及び全体のサービス状態を求
め、最良のかごに上記呼びを割り当てる装置であ
るが、説明の便宜上以後の説明においては、かご
ａに４階昇り呼び４ｕが仮割当された場合を中心
に進めて行くことにする。 さて、第５図において、選択乗場呼び信号２５
は「１」であるので、オアゲート２６の出力信
号、すなわち割当呼び信号２７ａは「１」とな
る。このとき仮割当装置１５ａの他の乗場に関す
るそれぞれの出力信号は、割当記憶装置１２のか
ごａの各乗場の割当呼び信号とそれぞれ一致して
いる。 次にサービス予測装置１７ａにおいて予測待時
間、満員確率、予報外れ確率が各かごについて演
算される。第６―Ａ図において、４階昇り呼び４
ｕは登録されて５秒経過しているので、その登録
信号２１は「１」であり、タイマ３２の出力は５
秒である。もし、かごが４階昇り呼びに応答する
と登録信号２１は「０」となるので、タイマ３２
のＲ点には「１」が入力されることになり時間は
零秒にリセツトされる。到着予想時間演算装置２
９は、かごが一周運転（かご位置階→かご前方の
終端階、そこで反転してかご前方の終端階→かご
背後の終端階、再度反転してかご背後の終端階→
かご位置階というような運転）するものとして、
かごが１階床進むのに要する時間（走行時間、例
えば２秒と設定される）と呼びがあるときその呼
びに応答して停止しサービスするのに要する時間
（停止時間、例えば10秒と設定される）とをかご
位置階からかごの進む方向へ順次加算していつて
各乗場にかごが到着するまでの時間を予測演算す
るもので、この装置はすでに公知となつている。
かごａの４階昇り方向の乗場の到着予想時間信号
３０ａが14秒と演算されているとすれば、加算器
３３により４階昇り呼びのかごａによる予測待時
間信号３４ａは５＋14＝19秒と演算される。一
方、かごｂの割当呼びである５階昇り呼び５ｕの
予測待時間は、到着予想時間が26秒と演算されて
いるとすれば、24＋26＝50秒と演算されている。 この実施例では満員確率を，および式で
近似計算する。（式における定数値をＡ＝２，
Ｂ＝10，Ｃ＝９，Ｄ＝２とする。）かご負荷を人
数に変換するのに60Kg／人とすればかごａのかご
位置階での乗込許容人数Ｎは、８−300／60＝３
人と設定される。３階で昇り方向にはかごａの割
当呼びがなくかごａに乗算する予定はないので、
かごａの３階昇り方向の乗場での満員積残し確率
は式から｛２・（０＋10）／（３＋９）｝−２＝
−0.33、したがつて、補正を行なつて０，０と求
められる。各階床・方向別の乗客発生率を0.15
人／秒と設定したとき、４階昇り方向の乗場での
待客数は予測待時間が19秒なので0.15×19＝2.85
人と予測される。（かごａの割当呼びのない乗場
の予測待客数は無視する。）したがつて、４階昇
り方向の乗場での満員積残し確率は式から−
0.33＋｛２／（３＋９）｝×2.85＝0.145と求められ
る。以下同様にして順次５階昇り、６階降り、
…，２階降り、１階昇り、方向の乗場の満員積残
し確率が求められる。 第６―Ｂ図において、信号３６ａはかごａが４
階にかご呼びを持つとき１，０，かご呼びを持た
ないときは一定値（例えば０，３）となるように
与えられている。したがつて、かごａの４階昇り
方向の乗場での満員通過確率信号４０ａは、0.0
×（１−0.3）＝0.0と出力される。他の乗場につい
ても同様で、５階昇り方向の乗場では0.145×（１
−0.3）＝0.101と出力される。 予報外れ確率はこの実施例では，，式で
求められる。例えば４階昇り方向の乗場の到着予
想時間は、かごａが14秒、かごｂが16秒、かごｃ
が32秒であり式における確率ｅをかご呼びをす
でに持つているときは１，０、持つていないとき
は一定値（例えば0.7）となるように与えるよう
にすれば、かごｂによる確率Ｗは式から1.5−
１６／１４＝0.357かごｃによる確率Ｗは1.5−３２／１
４＝− 0.786、したがつてＷ＝0.0かごｂ，ｃは４階にか
ご呼びを持つていないので結局４階昇り呼びの予
報外れ確率は式から0.357×0.7＋0.0×0.7＝
0.25と出力される。 以上のようにして、各乗場呼びの予想待時間、
満員確率及び予報外れ確率が演算される。その予
測結果をまとめると、４階昇り呼び４ｕをかごａ
に仮割当したときは表１、かごｂに仮割当したと
きは表２、かごｃに仮割当したときは表３のよう
になる。 The present invention relates to a device for managing a plurality of elevator cars as a group. Conventionally, when evaluating elevator services, the issue has often been the length of waiting time for passengers waiting at each landing. In other words, it has been considered that a good service is one in which the waiting time of each passenger at each landing is short on average, and the number of long-term waiting calls is small. However, as has been the case recently, when a system has been adopted in which a car designated for service at a landing call is displayed in advance as a service car to passengers waiting at the hall (hereinafter referred to as a "forecast"), the above forecast is not possible. If the car becomes full midway through and passes the above call, or another car responds to the car call, the above call is serviced before the forecast car (hereinafter, this situation is referred to as ``forecast failure''). A new problem has arisen in the elevator service, which is causing confusion and disappointing the expectations of passengers waiting on the platform. In the past, various measures have been devised to reduce the above-mentioned problems, but because each problem such as waiting time, crowded passage, and incorrect forecasts were dealt with separately, it was difficult for the boarding area as a whole to However, the overall quality of service provided to passengers waiting at the boarding station was insufficient. As an example, if we consider the case where the allocated car is selected based only on the waiting time and the risk of passing full people, it is best to select the car with the shortest waiting time and the lowest risk of passing through the crowded area as the allocated car. Although this can be said to be an allocation, generally there are not always cars that satisfy the above conditions. When there is no best car as described above, in the conventional method, the probability of the danger of passing through the car being full is not assigned to a car that is higher than a predetermined value (for example, 0.3), and the closest car among the remaining cars is selected. A method is used to select the item as an allocated basket. However, if the probability of the danger of passing through a crowded car is 0.31 and the waiting time is 10 seconds, the probability of the danger of passing through a crowded car is 0.29.
If there is a car with a waiting time of 100 seconds, there is not that much of a difference in the risk of a full car passing, so it is better to allocate it to the former car, although the risk of a full car passing increases slightly. As described above, in the conventional system, the risk of a crowded train passage and the length of waiting time are judged separately, so from the perspective of passengers waiting at the boarding point, the service lacks detail. Therefore, in order to improve the quality of service to passengers waiting at the platform, it is first necessary to accurately understand the status of service to passengers waiting at the platform, in relation to waiting time, fullness, missed forecasts, etc. be.
One possible way to do this is to express the service status based on the psychological impact on passengers waiting at the hall. In other words, we consider the different characteristics of waiting time, full capacity, and missed forecasts, as well as the psychological impact on waiting customers (feeling of impatience, anxiety, suspiciousness, dissatisfaction, etc.) due to the length of waiting time, and the effects of waiting time, full capacity, and not being fully booked. The magnitude of the danger that arises and the magnitude of the psychological impact on waiting customers when it occurs,
By numerically replacing it with the same dimensions: the magnitude of the risk of forecast failure and the magnitude of the psychological impact on waiting customers when it occurs,
It represents the state of service to passengers waiting at the platform. This invention improves the above-mentioned drawbacks, and improves the quality of service to passengers waiting at the landing by comprehensively and accurately understanding the service status of each passenger waiting at the landing and allocating the car that is considered to be the best. The purpose of the present invention is to provide an elevator group management device that can perform the following tasks. An embodiment of the present invention will be described below with reference to FIGS. 1 to 9. For convenience of explanation, a case will be described in which three cars are installed in a six-story building, but it goes without saying that the invention is applicable regardless of the number of cars installed and the number of floors. Prior to explaining the embodiments, calculations of the full probability and the forecast failure probability will be explained. As a means of expressing the dangers of a packed passage, a full passage, and an incorrect forecast, the probability of a full passage occurring (hereinafter referred to as the probability of a packed passage), the probability that a full passage occurs (hereinafter referred to as the probability of a packed passage), and the failure of the forecast, respectively. probability of occurrence (hereinafter referred to as forecast failure probability)
There is a way to use . (Hereinafter, the probability of a full passenger passing through and the probability of a full passenger remaining on board will be collectively referred to as the "full probability.") In general, there is a certain range of variation in the predicted number of waiting passengers and predicted car load at each landing. For example, the predicted car load is 60% of the car capacity.
Even if the car is only 100% full, it may actually be full (100%), and conversely, even if the predicted car load is 100% of the car capacity, the car may not actually be full. Also,
Because there are variations in the predicted time for a car to arrive at each landing, and even in the expected arrival time, for example, even if car a's expected arrival time is smaller than car b's expected arrival time, car b may arrive earlier. It is possible that you will arrive at As described above, there are variations in the predicted values, so it is necessary to take these variations into consideration in accurately predicting full occupancy and incorrect forecasts. The fullness probability and the forecast failure probability take into account the dispersion of the predicted values. Hereinafter, calculations of the full capacity probability and the forecast failure probability will be explained. The predicted value (average value) of the number of customers waiting at a certain boarding point is k
The probability that i people are waiting at this landing is f(i,
k), the predicted value (average value) of the number of people who can board until the car is full when the car arrives (hereinafter referred to as "permitted number of passengers") is m people, and the number of people allowed to board at this landing is j people. If the probability is g(j, m), then the probability that this boarding area will depart with no passengers left on board, i.e., the probability q that the board will be full is equal to
Since it is equal to the probability of exceeding a person , and the probability p of leaving this boarding station full is is given by However, No. indicates the car capacity.
Therefore, when a car departs with a full car, it will pass if there is no car call on the next floor.If the probability of having a car call on the next floor is u, then the probability r of a full car passing through the next floor is It can be found by Now, the car location floor is So, and the landings in front of the car in the same direction as the car are S <sub>1</sub> , S <sub>2</sub> , ...S <sub>o</sub> floors from the one closest to the car, and S <sub>o</sub>
Consider the case where there is no car call up to the floor. Boarding area S <sub>1</sub> ,
The number of waiting customers <sub>l 1 ,</sub> l <sub>2</sub> , ... l <sub>o</sub> of S 2 , ..., S <sub>o</sub> is the average value k <sub>1</sub> ,
When it is a Poisson distribution of k <sub>2</sub> ,..., <sub>ko</sub> , the sum of the number of waiting customers is the average value by the addition theorem Therefore, the sum L(n) of the number of waiting customers on floors S <sub>1</sub> , S <sub>2</sub> , ..., S <sub>o</sub> is exp(-K(n))・K(n) <sup>j</sup> /j〓 (j=0, 1, 2,...)... It has a probability distribution of... Also, the number of people allowed to board the S <sub>1st</sub> floor N is N = (car capacity) - (current number of people in the car)... Assuming that no one gets off on the way to the S <sub>o</sub> floor, then The probability that the car will be full on <sub>the o</sub> floor and there will be a backlog of waiting customers q <sub>o</sub> is becomes. If a car call is received by <sub>the So</sub> floor, and the number of passengers N allowed to board the train is predicted, taking into consideration that waiting passengers who boarded on the S <sub>o-1 floor</sub> will get off the train on the way.
It goes without saying that the probability q <sub>o</sub> can be calculated more accurately by using the formula. Also, the above explanation was about the landing in the front in the same direction as the car, but the landing area in the opposite direction to the car and the landing in the back in the same direction as the car are exactly the same except that they are calculated as N = (car capacity)... You can think about it. Furthermore, since the calculation formula is complicated, the following can be given as an approximate formula: q <sub>o</sub> =A{L(n)+B}/N+C-D . . . =q <sub>o-1</sub> +A/N+C·l <sub>o</sub> . . . However, A, B, C, D
are constant values (e.g. A=2, B=10, C=9, D=
2), when q <sub>o</sub> exceeds 1,0, q <sub>o</sub> =1,0
In addition, when q <sub>o</sub> is smaller than 0, it is necessary to correct q <sub>o</sub> =0. The formula also takes into account passengers who get off at intermediate floors, and is approximated by q <sub>o</sub> = q <sub>o-1</sub> + A/N + C {(Predicted number of passengers on S <sub>o</sub> floor) - (Predicted number of passengers alight on S <sub>o</sub> floor)}... can be calculated. Also, by approximating the formula and using q <sub>o</sub> obtained by the formula, calculate the probability r o <sub>+ 1</sub> that the first floor will be filled with people as r <sub>o+1</sub> = q <sub>o</sub>・(1−u)... be able to. Therefore, if the permissible number of passengers N at the car position floor or the terminal floor is known, the probability that each floor will remain full and the probability that the car will pass full can be determined in turn using the formula. Next, the predicted value of the time it takes for the car to arrive at each floor from the current car position floor is called the expected arrival time, and the expected arrival time includes the time required for passengers to get on and off at the landings along the way, the time required to close the door, It is determined by the time required for acceleration/deceleration, the time to travel at the rated speed, etc. Therefore, the expected arrival time at a certain boarding point at a certain point in time will vary within a certain range depending on various factors such as the number of passengers getting on, the number of people getting off at the scheduled stopping floor, and new calls. Suppose now that a hall call is registered at a certain hall, and car a is predicted to be the service car for that call. The distribution of time t until car a arrives at the landing is expressed as probability density function h(t)(t
o), this probability density function h(t)
satisfies the condition ∫ <sup>∞</sup> <sub>p</sub> h(t)dt=1..., and the probability that the car actually arrives at the above landing area between times T <sub>1</sub> and T <sub>2</sub> (T <sub>1</sub> < T <sub>2</sub> ) is R(T <sub>1</sub> , T <sub>2</sub> ) is given by R(T <sub>1</sub> , T <sub>2</sub> )=∫ <sup>T2</sup> <sub>T1</sub> h(t)dt... Similarly, the distribution of the expected arrival time t of other cars (one of which is called car b) at the landing area is expressed by a probability density function v(t) as follows:
As can be seen from FIG. 1, the probability W that car b arrives before car a can be found as W=∫ <sup>∞</sup> <sub>p</sub> {∫ <sup>x</sup> <sub>p</sub> v(t)dt}h(x)dx . Therefore, if the probability that car b stops due to a car call in the same direction as the hall call is e, then the probability that car b responds to the hall call before the predicted car a, that is, the probability that the forecast is incorrect is zb. is calculated as zb=W・e... Similarly, for car C, which is not forecasted, if we obtain the probability of incorrect forecast Zc, then the probability Z that the forecast for the above hall call is incorrect is Z=1-(1-Zb)(1-Zc)...and continue. Taking into account that the forecast is missed two or three times, the probability of a forecast miss is expressed by the expected value Z' of the number of forecast misses, as follows: Z' = 1 x Zb + 1 x Zc... By the way, there is a way to easily calculate the first-arrival probability W expressed by the formula by assuming the distribution of arrival times to be uniform distribution, exponential distribution, normal distribution, etc., but one of the easier ways to calculate it is to Based on the prediction that the longer the expected arrival time, the greater the variation, the following formula can be used: W=1.5-tb/ta... However, ta is the forecasted expected arrival time of car a, tb is the predicted arrival time of car b, which is not forecasted, and W<
When 0,0, W=0; when W>1,0, W
It is necessary to correct it to =1. In FIG. 2, a to c are cars with a capacity of 8 people, and the car loads are 300 kg, 420 kg, and 300 kg, respectively.
1c is the car name registered in car c, 3a to 3
Assume that c is a 3rd floor car call registered in cars a to c, and 4u is a 4th floor ascending call that has not yet been assigned to any car, and that 5 seconds have passed since the call was registered. . It is assumed that 5u is an up call for the 5th floor assigned to car a, and 24 seconds have passed since the call was registered. In FIG. 3, 10 is a car control device, 11 is a device for registering hall calls, 12 is an allocation storage device for storing hall calls assigned to each car a to c, and 1
3 is a hall call selection device that selects one hall call to be assigned to a car; 14 is an assignment device that assigns the hall call selected by the hall call selection device 13 to a car; 15a to 15c are assignment devices 14;
Temporary allocation devices 16a to 16c that output an allocation call signal when a hall call selected to be allocated is provisionally allocated to each of the cars a to c;
17a is a provisional allocation evaluation device which is also provided in the allocation device 14 and outputs a comprehensive evaluation value when a call is provisionally allocated to each car a to c; 17a is a provisional allocation evaluation device 1;
With the service prediction device installed in 6a, each car
Predicted waiting time for each floor/direction for ~c,
Predict and output the probability of fullness and probability of forecast failure. Reference numeral 18a denotes a service status evaluation device also provided in the provisional allocation device 16a, which calculates and outputs the service status for each floor and direction, and a floor-based evaluation device that calculates and outputs the service status for each floor and direction.
It consists of a comprehensive evaluation device that outputs a comprehensive evaluation value.
19 is a main allocation device provided in the allocation device 14;
The car having the minimum overall evaluation value is selected from the outputs of the provisional allocation devices 16a to 16c for each of the cars a to c, and is normally allocated to the call. FIG. 4-A is an example of a circuit diagram of the hall call selection device 13, and is an arithmetic circuit for the hall in the ascending direction on the fourth floor, and a similar arithmetic circuit is required for the other halls. In the figure, 20a to 20c are the output signals of the allocation storage device 12, and the calls for going up to the 4th floor are respectively assigned to car a.
Assigned call signal that becomes "1" only when assigned to ~c (one car is assigned to one call, so signals 20a~2)
21 is the output signal of the hall call registration device 11, which is a hall call registration signal that becomes "1" only when a 4th floor ascending call is registered. , 22 is the 4th-B
As shown in the figure, the pulse train signal 23 periodically becomes "1" only at the time corresponding to the landing in the ascending direction of the 4th floor.
is a Noah gate, 24 is an AND gate, and 25 is an output signal of the hall call selection device 13, which is a selective hall call signal that becomes "1" only when a 4th floor ascending call is selected. FIG. 5 shows provisional allocation device 15a-4 for car a.
This is an example of a circuit diagram for a landing in the ascending direction; similar circuits are required for other landings. Temporary allocation devices 15b and 15c corresponding to cars b and c
is also composed of a similar circuit. In the figure, 26 is an OR gate, 27a is an output signal of the temporary allocation device 15a,
``1'' only when the 4th floor ascending call is a call assigned to car a or a call provisionally allocated to car a.
This is the assigned call signal. FIG. 6-A and FIG. 6-B are examples of circuit diagrams of the service prediction device 17a. FIG. 6-A shows a part of a circuit that calculates a predicted waiting time, and 28 in the figure is a circuit that calculates a predicted waiting time for a 4th floor ascending call by car a. Similar circuitry is required for other calls. 29 is an expected arrival time calculating device that predicts and calculates the time until the car arrives at each landing in the up direction and the down direction, and 30 a is the expected arrival time calculation device for car a.
31 is the not gate, 32 is the expected arrival time signal until the arrival at the 4th floor in the ascending direction, and 32 starts counting the time when the input signal at the I point changes from "0" to "1", and when the input signal at the R point changes from "1" to "1". The timer resets the time to zero when a signal is received, and outputs the time elapsed since the hall call was registered. 33 is an adder, 3
4a is a signal (predicted waiting time signal) representing the predicted waiting time for a fourth-floor ascending call by car a. Figure 6-B is an example of a circuit diagram of a landing for car a in the ascending direction of the fourth floor, and is a circuit that realizes the formula. Similar circuits are required for other landings. In the figure, 35a is a signal representing the probability that car a will be full and left unloaded when car a leaves the third floor in the ascending direction (the calculation circuit is not shown), and 36a is a signal indicating the probability that car a will leave the 3rd floor in the ascending direction in response to a car call. A signal representing the probability of stopping (the calculation circuit is not shown), 37 is 1,0
38 is a subtractor that outputs a value obtained by subtracting the input at point X from the input at point Y, 39 is a multiplier, and 40a is a full passage probability signal for a call for the fourth floor of car a. FIG. 7 is an example of a circuit diagram of a floor-specific evaluation device in the service status evaluation device 18a, and is a circuit corresponding to a call for going up to the fourth floor. Similar circuits are required for other hall calls. In the figure, 34b, 34c
are predicted waiting time signals of the 4th floor up call 4u in car b and car c, respectively, and 40b and 40c are the 4th floor up call 4u in car b and car c, respectively.
, 41a to 41c are probability signals of the 4th floor up call 4u in the cars a to c being full and remaining unloaded (the calculation circuit is not shown), and 42a to 42c are the prediction failure of the 4th floor up call 4u in the cars a to c. Probability signal (calculation circuit not shown), 43 to 46 are G ₁
~G Among _{the 3} points, select the input signal of the 3 points I ₁ ~ _I that corresponds to the point where the signal "1" is input, and input the signal 43
A selection circuit that outputs as a to 46a, for example
G If a signal of "1" is input to _one point, I ₁
Outputs the point input signal. Note that when a signal of "0" is input to all _three points _G1 to G, zero is always output. 47 to 50 are multipliers, 51 is an adder, and 52 to 55 are constant value signals set to, for example, 1, 30, 50, and 20, respectively. 56a is 4
This is a service status evaluation value signal for the floor-climbing call 4u. FIG. 8 is an example of a circuit diagram of a comprehensive evaluation device in the service status evaluation device 18a, in which 57a,
..., 58a, ..., 59a are output signals of the evaluation device for each floor, respectively, service status evaluation value signals for the 1st floor up call, ..., 5th floor up call, ..., 2nd floor down call, and 60 to 63 are floor These are constant value signals for weighting, but here it is assumed that they are all set to 1 and 0. 64 to 67 are multipliers, 68 is an adder, and 69a is a comprehensive evaluation value signal. FIG. 9 is an example of a circuit diagram of the present allocation device 19, and is a circuit corresponding to a call going up to the fourth floor. Similar circuits are required for other hall calls. In the figure, 69
b and 69c are provisional allocation evaluation device 1 when 4th floor ascending call 4u is allocated to car b and car c, respectively.
The output of 6b and 16c is the comprehensive evaluation value signal, 70 is
This is a minimum value selection circuit that selects the one having the minimum value from among the input signals at points Ia to Ib and outputs the corresponding signals 70a to 70c. For example, when the input signal 69b at point Ib is the minimum, only the output signal 70b becomes "1" and the other signals 70a and 70c become "0".
is output. 71-73 are and gates, 74
Reference numerals a to 74c are assignment storage command signals for cars a to c, respectively. For example, when signal 74a is "1", a 4th floor ascending call is stored in the assignment storage device 12 as an assigned call for car a. Now, as shown in Figure 2, the 4th floor climbing call 4u
The operation of each circuit will be specifically explained in the case of assigning a car to a car. First, in Figure 4-A, the 4th floor climbing call 4u
has not been assigned to any basket yet, so
The assigned call signals 20a to 20c are all "0", and the output signal of the NOR gate 23 is "1". Further, the hall call registration signal 21 is also set to "1". On the other hand, since the pulse train signal 22 is a signal that has a pulse of "1" in a certain period, the AND gate 2 is used only when the pulse train signal 22 becomes "1".
The output signal 4, that is, the selected hall call signal 25 becomes "1", and the fourth floor ascending call 4u is selected as the call to be assigned. At this time, all selected hall call signals corresponding to other hall calls are "0". As previously explained, the allocation device 14 tentatively allocates the hall calls selected by the hall call selection device 13 to each car, and determines the service status of each hall call and the overall service status at that time. , is a device that allocates the above-mentioned call to the best car. However, for convenience of explanation, the following explanation will focus on the case where the 4th floor ascending call 4u is provisionally allocated to car a. Now, in FIG. 5, the selected hall call signal 25
is "1", so the output signal of the OR gate 26, that is, the assigned call signal 27a, becomes "1". At this time, each output signal of the temporary allocation device 15a regarding the other landings matches the allocation call signal of each landing of the car a of the allocation storage device 12, respectively. Next, the service prediction device 17a calculates the predicted waiting time, full probability, and forecast failure probability for each car. In Figure 6-A, 4th floor climbing call 4
Since 5 seconds have passed since u was registered, its registration signal 21 is "1" and the output of the timer 32 is 5 seconds.
Seconds. If the car responds to the call for going up to the 4th floor, the registration signal 21 becomes "0", so the timer 32
``1'' is input to point R, and the time is reset to zero seconds. Estimated arrival time calculation device 2
9, the car runs once (car position floor→terminal floor in front of the car, then reverses, terminal floor in front of the car→terminal floor behind the car, reverses again, terminal floor behind the car→
As a device that operates (such as car position floor),
The time required for the car to advance one floor (travel time, set to 2 seconds, for example) and the time required for the car to stop and serve in response to a call (stop time, set to 10 seconds, for example) This device is already known, and calculates a prediction of the time it will take for the car to arrive at each landing by sequentially adding up the car's car position from the car's floor in the direction in which the car moves.
If the expected arrival time signal 30a of the landing for the 4th floor ascending direction of car a is calculated to be 14 seconds, then the adder 33 calculates that the predicted waiting time signal 34a for the 4th floor ascending landing is 5+14=19 seconds. Calculated. On the other hand, the predicted waiting time for the 5th floor ascending call 5u, which is the assigned call for car b, is calculated as 24+26=50 seconds, assuming that the expected arrival time is 26 seconds. In this embodiment, the fullness probability is approximately calculated using the following formula. (The constant value in the formula is A=2,
Let B=10, C=9, and D=2. ) To convert the car load into the number of people, if we take 60Kg/person, the number of people allowed to board the floor of car A is 8-300/60=3
It is set as a person. There is no assigned call for car a in the ascending direction on the 3rd floor, so there is no plan to multiply car a.
The probability that car a will remain full at the landing in the direction of going up to the 3rd floor is given by the formula: {2・(0+10)/(3+9)}-2=
−0.33, therefore, it can be corrected to be 0,0. Passenger incidence rate for each floor/direction is 0.15
When set as people/second, the number of customers waiting at the landing on the 4th floor is 0.15 x 19 = 2.85 because the predicted waiting time is 19 seconds.
Predicted to be a person. (Ignore the predicted number of waiting passengers at the landings where there is no assigned call for car a.) Therefore, the probability of full passengers remaining at the landings in the ascending direction of the 4th floor is calculated from the formula -
It is calculated as 0.33+{2/(3+9)}×2.85=0.145. In the same way, ascend the 5th floor, descend the 6th floor,
..., get off the second floor, go up the first floor, the probability of the boarding area being full and remaining in the direction is determined. In Figure 6-B, signal 36a indicates that car a is 4.
It is given a value of 1 or 0 when the floor has a car call, and a constant value (for example, 0 or 3) when there is no car call. Therefore, the full passage probability signal 40a at the landing in the ascending direction of the fourth floor of car a is 0.0.
×(1-0.3)=0.0 is output. The same goes for other landings, and for the landing on the 5th floor in the ascending direction, 0.145×(1
−0.3) = 0.101 is output. In this embodiment, the forecast failure probability is determined by the formula. For example, the expected arrival time at the landing for the 4th floor ascending direction is 14 seconds for car a, 16 seconds for car b, and 16 seconds for car c.
is 32 seconds, and the probability e in the formula is given as 1 or 0 when the car already has a car call, and a constant value (for example, 0.7) when it does not have a car call, then the probability W due to car b is 1.5− from Eq.
16/14=0.357 Probability W due to basket c is 1.5-32/1
4 = - 0.786, therefore W = 0.0 Cars b and c do not have car calls on the 4th floor, so the probability that the forecast for the 4th floor ascending call will be incorrect is 0.357 x 0.7 + 0.0 x 0.7 =
0.25 is output. As described above, the expected waiting time for each hall call,
The fullness probability and the forecast failure probability are calculated. To summarize the prediction results, the 4th floor climbing call 4u is placed in car a.
Table 1 shows provisional allocation to car b, Table 2 shows provisional allocation to car b, and Table 3 shows provisional allocation to car c.

【表】【table】

【表】【table】

【表】 さて、次に階床別のサービス状態の評価が行な
われる。４階昇り呼びについてみると、４階昇り
呼び４ｕがかごａに仮割当されたとき第７図にお
いて割当呼び信号２７ａは「１」、割当呼び信号
２０ｂ，２０ｃはそれぞれ「０」であるので、結
局選択回路４３〜４６の出力信号４３ａ〜４６ａ
は表１からわかるようにそれぞれ19秒、0.145，
0.0，0.25となる。また、定数値信号５２〜５５
はそれぞれ、１，30，50，20と設定されているの
で加算器５１の出力信号、すなわち４階昇り呼び
のサービス状態評価値信号５６ａは19×１＋
0.145×30＋0.0×50＋0.25×20＝28.35となる。４
階昇り呼び４ｕをかごａ〜ｃにそれぞれ仮割当し
たときの各呼びに対するサービス状態の評価値信
号は表４のようになる。[Table] Next, the service status by floor will be evaluated. Regarding the fourth floor ascending call, when the fourth floor ascending call 4u is provisionally assigned to car a, the assigned call signal 27a is "1" in FIG. 7, and the assigned call signals 20b and 20c are each "0", so After all, the output signals 43a to 46a of the selection circuits 43 to 46
As can be seen from Table 1, the times are 19 seconds, 0.145, and 0.145, respectively.
0.0, 0.25. In addition, constant value signals 52 to 55
are set to 1, 30, 50, and 20, respectively, so the output signal of the adder 51, that is, the service status evaluation value signal 56a for the 4th floor ascending call is 19×1+
0.145×30+0.0×50+0.25×20=28.35. 4
Table 4 shows the evaluation value signals of the service status for each call when the ascending call 4u is provisionally assigned to each of the cars a to c.

【表】 次に、すべての乗場呼びについての総合評価が
行われる。４階昇り呼び４ｕがかごａに仮割当さ
れたとき第８図において、４階昇り呼び及び５階
昇り呼びに対するサービス状態の評価値信号５６
ａ，５８ａはそれぞれ28.35，85.38であり、他の
呼びに対する評価値信号はすべて０となつてい
る。したがつて、総合評価値６９ａは乗算器６４
〜６７、加算器６８により表４に示すように
113.73と出力される。同様にして、４階昇り呼び
４ｕをかごｂおよびかごｃに仮割当したときの総
合評価値は表４のように求められる。 したがつて、本割当装置１９は第９図において
最小値選択回路７０は総合評価値を最小とするか
ごｂを選択し、信号７０ｂを「１」、信号７０ａ
を「０」、信号７０ｃを「０」として出力する。
４階昇り呼び４ｕに対する選択乗場呼び信号２５
は「１」であるので割当記憶指令信号７４ａ〜７
０ｃはそれぞれ「０」，「１」，「０」となり結局４
階昇り呼び４ｕはかごｂに割ぃ当てられ、かごｂ
の割当呼びとして割当記憶装置１２に記憶され
る。 上述のようにこの実施例においては、割り当て
るべく選択された乗場呼びを各かごに仮りに割り
当ててみて、各呼びごとに予測待時間、満員確
率、予報外れ確率を予測し、それらを組合せてサ
ービス状態を評価し、更に呼び全体としての総合
評価値を求めてそれが最小となるようにかごを割
り当てるようにした。その結果、乗場待客に対す
るサービスの質を向上させることが可能となる。 上記実施例では、階床別評価装置（第７図）に
おいて、予測待時間と満員確率と、予報外れ確率
とを線形結合したものを４階昇り呼びに対するサ
ービス状態の評価値信号５６ａとして出力した
が、この線形結合したものを更に２乗して出力
し、これをあらためて４階昇り呼びに対するサー
ビス状態の評価値信号５６ａをすることは容易に
実現可能である。線形結合したものを２乗するこ
とによつてサービス状態の悪さを加速度的に強調
するので、極端にサービスの悪い乗場呼びを減少
させる効果が生じ、待客に対するサービスの質を
向上させることが可能となる。もちろん、２乗に
限らず1.5乗、３乗、４乗…でも同様の効果が期
待できることは言うまでもない。 第１０図及び第１１図はこの発明の他の実施例
を示し、サービス状態評価装置１８ａに相当する
回路である。第１０図は予測待時間、満員確率、
予報外れ確率に関して別々に階床全体のサービス
状態を評価し出力する装置のうち、予測待時間に
関する評価装置の回路図である。満員確率、予報
外れ確率に関しての評価装置も予測待時間信号の
代わりに満員確率、予報外れ確率信号を使用する
他は全く同様の回路で構成される。図中、７０，
…，７１，…，７２はそれぞれ１階昇り呼び，
…，４階昇り呼び，…，２階降り呼びについての
演算回路、７３はG₁〜G₃点の中で「１」の信号
が入力されている点に対応するI₁〜I₃点の入力信
号を選んで信号７４として出力する選択回路で、
例えばG₁点に「１」の信号が入力されている場
合にはI₁点の入力信号を出力する。なお、G₁〜G₃
点にいずれも「０」の信号が入力されているとき
は常に零を出力する。７５は乗算器、７６は４階
昇り呼びに対する重み付けのための定数値信号
（ここでは1.0と設定されているものとする。）７
７は加算器、７８は加算器７７の出力信号で、予
測待時間に関する階床全体の評価値を表わす。 第１１図は予測待時間、満員確率、予報外れ確
率に関する階床全体の評価値を組合せて総合的に
評価する装置の回路図の一例である。図中、７９
〜８１はそれぞれ満員積残し確率、満員通過確
率、予報外れ確率に関する階床全体の評価値信
号、８２〜８５は定数値信号でそれぞれ１，50，
100，50と設定される。８６〜８９は乗算器、９
０は加算器である。 さて、４階昇り呼び、５階昇り呼びに関して
〔表１〕のように予測待時間、満員確率、及び予
報外れ確率が求められている場合を考える。この
とき、第１０図においては予測待時間に関する階
床全体の評価値信号７８が演算される。割当呼び
信号２７ａ＝「１」なので選択回路７３により４
階昇り呼びの予測待時間信号３４ａ（＝19秒）が
選択され、出力信号７４となる。４階昇り呼び、
及び５階昇り呼び以外の呼びの予測待時間は０秒
なので結局、乗算器７５、加算器７７により評価
値信号７８は19秒×1.0＋50秒×1.0＝69秒とな
る。同様にして満員確率、予報外れ確率の場合、
また４階昇り呼び４ｕをかご６及びｃに仮割当し
た場合には〔表５〕のように求められ、結局、か
ごｂに４階昇り呼び４ｕが割り当てられることに
なる。[Table] Next, a comprehensive evaluation of all hall calls is performed. When the 4th floor up call 4u is provisionally assigned to the car a, in FIG. 8, the service status evaluation value signal 56 for the 4th floor up call and the 5th floor up call is shown.
a and 58a are 28.35 and 85.38, respectively, and the evaluation value signals for other calls are all 0. Therefore, the overall evaluation value 69a is the multiplier 64
~67, as shown in Table 4 by adder 68
113.73 is output. Similarly, when the 4th floor ascending call 4u is provisionally assigned to cars b and c, the overall evaluation value is obtained as shown in Table 4. Therefore, in the present allocation device 19, the minimum value selection circuit 70 selects the car b that minimizes the overall evaluation value in FIG.
is output as "0", and the signal 70c is output as "0".
Selective hall call signal 25 for 4th floor ascending call 4u
is "1", so the allocation storage command signals 74a to 7
0c becomes “0”, “1”, and “0” respectively, and ends up being 4
Climbing call 4u is assigned to car b;
is stored in the allocation storage device 12 as an allocated call. As described above, in this embodiment, the selected hall calls are tentatively assigned to each car, and the expected waiting time, full probability, and forecast failure probability are predicted for each call, and these are combined to provide service. The condition was evaluated, and the overall evaluation value for the entire call was determined, and the cars were allocated so that the overall evaluation value was minimized. As a result, it is possible to improve the quality of service for passengers waiting at the boarding point. In the above embodiment, the floor-specific evaluation device (FIG. 7) outputs a linear combination of the predicted waiting time, the full probability, and the predicted failure probability as the service status evaluation value signal 56a for the 4th floor ascending call. However, it is easily possible to further square this linear combination and output it, thereby reproducing the service status evaluation value signal 56a for the fourth-floor ascending call. By squaring the linear combination, poor service conditions are emphasized at an accelerated rate, which has the effect of reducing the number of boarding calls with extremely poor service, thereby improving the quality of service to waiting customers. becomes. Of course, it goes without saying that the same effect can be expected not only with the 2nd power but also with the 1.5th power, the 3rd power, the 4th power, etc. 10 and 11 show another embodiment of the present invention, which is a circuit corresponding to the service status evaluation device 18a. Figure 10 shows predicted waiting time, full probability,
FIG. 2 is a circuit diagram of an evaluation device regarding predicted waiting time among devices that separately evaluate and output the service status of the entire floor with respect to the probability of failure in forecast. The evaluation device for the probability of fullness and the probability of failure in the forecast is also configured with a completely similar circuit, except that the probability of fullness and the probability of failure in the forecast are used instead of the predicted waiting time signal. In the figure, 70,
..., 71, ..., 72 are respectively 1st floor climbing calls,
..., 4th floor going up call, ..., 2nd floor going down call, 73 is a calculation circuit for the 3 points I ₁ - I corresponding to the point where the signal "1" is input among _{the 3} points G ₁ - G A selection circuit that selects an input signal and outputs it as a signal 74.
For example, if a signal of "1" is input to _one point G, the input signal of _one point I is output. In addition, G ₁ to G ₃
When a signal of "0" is input to any point, zero is always output. 75 is a multiplier, 76 is a constant value signal for weighting the fourth-floor ascending call (here, it is assumed that it is set to 1.0) 7
7 is an adder, and 78 is an output signal of the adder 77, which represents the evaluation value of the entire floor regarding the predicted waiting time. FIG. 11 is an example of a circuit diagram of a device that performs a comprehensive evaluation by combining evaluation values for the entire floor regarding predicted waiting time, full probability, and probability of forecast failure. In the figure, 79
- 81 are evaluation value signals for the entire floor regarding the probability of being left full, the probability of passing full, and the probability of forecast failure, respectively, and 82 to 85 are constant value signals of 1, 50, respectively.
It is set as 100, 50. 86 to 89 are multipliers, 9
0 is an adder. Now, let us consider a case where the predicted waiting time, full probability, and forecast failure probability are calculated as shown in Table 1 for calls going up to the 4th floor and calls going up to the 5th floor. At this time, in FIG. 10, an evaluation value signal 78 for the entire floor regarding the predicted waiting time is calculated. Since the assigned call signal 27a is "1", the selection circuit 73 selects 4
The predicted waiting time signal 34a (=19 seconds) for the floor-climbing call is selected and becomes the output signal 74. 4th floor climbing call,
Since the predicted waiting time for calls other than the 5th floor ascending call is 0 seconds, the multiplier 75 and adder 77 result in the evaluation value signal 78 being 19 seconds x 1.0 + 50 seconds x 1.0 = 69 seconds. Similarly, in the case of full capacity probability and forecast failure probability,
Furthermore, when the 4th floor up call 4u is provisionally assigned to cars 6 and c, the results are calculated as shown in Table 5, and the 4th floor up call 4u is eventually assigned to car b.

【表】 この実施例は、まず予測待時間、満員確率、予
報外れ確率、に関して別々に階床全体のサービス
状態を評価し、その後で上記別々の評価値を組合
せて総合的に評価値を求めそれが最小となるよう
にかごを割り当てるようにした。その結果、乗場
待客に対するサービスの質を向上させることが可
能となる。 上記各実施例では、予測待時間、満員確率、予
報外れ確率をすべて組合せてサービス状態を評価
したが、必ずしもすべて組合せる必要はない。交
通状態に応じて適当に１個又は２個以上の組合せ
で評価するようにもできるし、また組合せるとき
に乗じる定数値信号５２〜５５や同じく信号８２
〜８５を交通状態に応じて変えるようにもできる
し、また階床ごとに定数値信号５２〜５５を変
え、例えば待時間を重視する階は信号５２を大き
めに設定するなど、階床に応じたサービスが可能
となる。 また、階床全体として評価するとき、上記実施
例では乗場呼びのある乗場だけを評価の対象とし
たが、呼びがなくとも近い将来乗場呼びが発生す
ることを考慮して、その呼びの予測待時間や満員
確率、予報外れ確率を予測して評価することも考
えられる。例えば最も近いかごの到着予想時間を
上記呼びの予測待時間としてサービス状態を評価
すればよい。したがつて、将来発生するであろう
呼びに対してもサービスの質を向上させることが
可能となる。更にまた、階床ごとの重み付けをす
るための定数値信号６０〜６３や同じく信号７６
を上記実施例ではすべて1.0に設定したが、主階
床のように交通量の多い階床や重役室のある階な
ど特に良いサービスが要求される階床などは、上
記定数値信号を例えば1.5という具合に他の階床
より大き目に設定することにより、上記階床を優
先的にサービスさせることができる。 また、上記実施例では将来満員や予報外れが生
じる可能性を予測したが、その他呼びが発生して
から現在に至るまでに生じた満員積残しや満員通
過及び、予報外れなどを検出して、それらが生じ
たことによるサービス状態の悪化も加味して各呼
びのサービス状態を評価し、それにより、上記サ
ービス状態の悪い呼びを優先的にサービスさせる
ことができることは容易に類推できる。また、上
記実施例では予測値のばらつきを考慮して満員確
率や予報外れ確率を採用したが、これに限るもの
ではない。満員、予報外れを表わすものであれば
この発明を適用できることは容易に類推できる。 以上説明したように、この発明では各呼びの予
測待時間、満員確率、予報外れ確率を予測し、各
乗場待客のサービス状態を総合的にかつ適確に把
握して最良のかごを割り当てるようにしたので、
乗場待客に対するサービスの質の向上をはかるこ
とができる。[Table] In this example, the service status of the entire floor is first evaluated separately in terms of predicted waiting time, full probability, and forecast failure probability, and then the above-mentioned separate evaluation values are combined to obtain a comprehensive evaluation value. I tried to allocate the baskets so that the number of baskets is minimized. As a result, it is possible to improve the quality of service for passengers waiting at the boarding point. In each of the embodiments described above, the service status is evaluated by combining the predicted waiting time, the full probability, and the forecast failure probability, but it is not necessary to combine all of them. It is also possible to evaluate one or a combination of two or more depending on the traffic condition, or use the constant value signals 52 to 55 or the signal 82 to be multiplied when combining them.
~85 can be changed depending on the traffic condition, and constant value signals 52~55 can be changed for each floor, for example, signal 52 can be set larger on a floor where waiting time is important. services will become possible. In addition, when evaluating the floor as a whole, in the above embodiment, only the hall with a hall call was subject to evaluation, but considering that even if there is no hall call, a hall call will occur in the near future, the predicted waiting time for that call is It is also possible to predict and evaluate the time, probability of fullness, and probability of forecast failure. For example, the service status may be evaluated using the predicted arrival time of the nearest car as the predicted waiting time for the call. Therefore, it is possible to improve the quality of service even for calls that may occur in the future. Furthermore, constant value signals 60 to 63 and a signal 76 for weighting each floor are also provided.
are all set to 1.0 in the above example, but for floors where particularly good service is required, such as floors with heavy traffic such as the main floor or floors with executive offices, the above constant value signal may be set to 1.5, for example. By setting the floor to be larger than other floors, the floor can be given preferential service. In addition, in the above embodiment, the possibility of full occupancy and forecast deviations occurring in the future was predicted, but it is also possible to detect full vacancies, full passages, and forecast deviations that have occurred from the time the call occurred to the present. It can be easily inferred that the service status of each call is evaluated taking into account the deterioration of the service status due to these occurrences, and thereby calls with poor service status can be serviced preferentially. Furthermore, in the above embodiments, the probability of being full and the probability of not being forecasted are adopted in consideration of variations in predicted values, but the present invention is not limited to this. It can be easily inferred that the present invention can be applied to any event that indicates full occupancy or unforeseen conditions. As explained above, this invention predicts the expected waiting time, full probability, and forecast failure probability for each call, comprehensively and accurately grasps the service status of each passenger waiting at the platform, and allocates the best car. So,
It is possible to improve the quality of service for passengers waiting on the platform.

【図面の簡単な説明】[Brief explanation of the drawing]

第１図は予報外れ確率の演算を説明するための
図、第２図〜第９図はこの発明によるエレベータ
の群管理装置の一実施例を示し、第２図はかごと
各呼びの関係説明図、第３図は全体の構成を示す
ブロツク図、第４―Ａ図は第３図の乗場呼び選択
装置の論理回路図、第４―Ｂ図はパルス列信号の
波形図、第５図は第３図の仮割当装置の論理回路
図、第６―Ａ図は第３図のサービス予測装置にお
ける予測待時間を演算する回路図、第６―Ｂ図は
第３図のサービス予測装置における満員通過確率
を演算する回路図、第７図は第３図のサービス状
態評価装置における階床別の評価値を演算する回
路図、第８図は第３図のサービス状態評価装置に
おける階床全体の評価値を演算する回路図、第９
図は第３図の本割当装置の論理回路図、第１０図
〜第１１図は第３図のサービス状態評価装置の他
の実施例を示す回路図である。 １０……かご制御装置、１１……乗場呼び登録
装置、１２……割当記憶装置、１３……乗場呼び
選択装置、１４……割当装置、１５ａ〜１５ｃ…
…仮割当装置、１６ａ〜１６ｃ……仮割当評価装
置、１７ａ……サービス予測装置、１８ａ……サ
ービス状態評価装置、１９……本割当装置、３４
ａ〜３４ｃ……予測待時間信号、４０ａ〜４０ｃ
……満員通過確率信号、４１ａ〜４１ｃ……満員
積残し確率信号、４２ａ〜４２ｃ……予報外れ確
率信号、５６ａ……階床別の評価値信号、６９ａ
……総合評価値信号、７４ａ〜７４ｃ……割当記
憶指令信号。なお、図中同一部分は同一符号によ
り示す。 FIG. 1 is a diagram for explaining the calculation of the prediction failure probability, FIGS. 2 to 9 show an embodiment of the elevator group management device according to the present invention, and FIG. 2 explains the relationship between the car and each call. 3 is a block diagram showing the overall configuration, FIG. 4-A is a logic circuit diagram of the hall call selection device of FIG. 3, FIG. 4-B is a waveform diagram of the pulse train signal, and FIG. 3 is a logic circuit diagram of the provisional allocation device, FIG. 6-A is a circuit diagram for calculating predicted waiting time in the service prediction device of FIG. 3, and FIG. 6-B is a circuit diagram for calculating the predicted waiting time in the service prediction device of FIG. 3. A circuit diagram for calculating the probability, FIG. 7 is a circuit diagram for calculating the evaluation value for each floor in the service condition evaluation device of FIG. 3, and FIG. 8 is a circuit diagram for calculating the evaluation value for each floor in the service condition evaluation device of FIG. 3. Circuit diagram for calculating values, No. 9
This figure is a logic circuit diagram of the present allocation device shown in FIG. 3, and FIGS. 10 and 11 are circuit diagrams showing other embodiments of the service status evaluation device shown in FIG. 3. 10... Car control device, 11... Hall call registration device, 12... Allocation storage device, 13... Hall call selection device, 14... Allocation device, 15a to 15c...
...temporary allocation device, 16a-16c...temporary allocation evaluation device, 17a...service prediction device, 18a...service status evaluation device, 19...main allocation device, 34
a~34c... Predicted waiting time signal, 40a~40c
...Full passage probability signal, 41a to 41c...Full to capacity remaining probability signal, 42a to 42c...Forecast failure probability signal, 56a...Evaluation value signal for each floor, 69a
. . . Comprehensive evaluation value signal, 74a to 74c . . . Allocation storage command signal. Note that the same parts in the figures are indicated by the same reference numerals.

Claims (1)

【特許請求の範囲】 １ 各かごのサービス状態を評価して総合評価値
信号を発し、これにより乗場呼びに対してかごを
割り当てるようにした装置において、予報外れ確
率及び満員確率のうちの少くともひとつと予測待
時間とを階床ごとにかつ上記乗場呼びの方向別に
演算しそれに対応する出力をそれぞれ発するサー
ビス予測装置、これらサービス予測装置の出力を
組み合わせて上記総合評価値信号を出力するサー
ビス状態評価装置を備えたことを特徴とするエレ
ベータの群管理装置。 ２ かごのサービス状態を評価して総合評価値信
号を発し、これにより乗場呼びに対してかごを割
り当てるようにした装置において、予報外れ確率
及び満員確率のうちの少くともひとつと予測待時
間とを階床ごとにかつ上記乗場呼びの方向別に演
算しそれに対応する出力をそれぞれ発するサービ
ス予測装置、これらサービス予測装置の出力を予
測外れ確率及び満員確率のうちの少くともひとつ
と上記予測待時間とのそれぞれごとに全階床のサ
ービス状態を評価しそれに対応する出力をそれぞ
れ発する第１の評価装置とこれら第１の評価装置
の出力を組み合わせて上記総合評価値信号を出力
する第２の評価装置とからなるサービス状態評価
装置を備えたことを特徴とするエレベータの群管
理装置。 ３ サービス予測装置の出力のそれぞれに対応す
る信号を加算して出力する第１の評価装置を用い
た特許請求の範囲第２項記載のエレベータの群管
理装置。 ４ サービス予測装置の出力のそれぞれに対応す
る信号を加算しそれを２乗して出力する第１の評
価装置を用いた特許請求の範囲第２項記載のエレ
ベータの群管理装置。 ５ 第１の評価装置の出力のそれぞれを階床ごと
にかつ乗場呼びの方向別に重み付けによる補正を
行つて加算して出力する第２の評価装置を用いた
特許請求の範囲第２項から第４項までのいずれか
に記載のエレベータの群管理装置。 ６ 乗場呼びを各かごに仮りに割り当てたときの
サービス状態評価装置の出力が最小となるかごに
上記乗場呼びを正規に割り当てるようにした特許
請求の範囲第３項から第５項までのいずれかに記
載のエレベータの群管理装置。[Scope of Claims] 1. In a device that evaluates the service status of each car and issues a comprehensive evaluation value signal, and allocates a car to a hall call based on the signal, at least one of the forecast failure probability and the full probability is determined. a service prediction device that calculates the expected waiting time for each floor and for each direction of the hall call and issues corresponding outputs; and a service state that combines the outputs of these service prediction devices to output the overall evaluation value signal. An elevator group management device characterized by comprising an evaluation device. 2. In a device that evaluates the service status of a car, issues a comprehensive evaluation value signal, and uses this to allocate a car to a hall call, the system calculates at least one of the probability of failure in the forecast and the probability of fullness and the predicted waiting time. A service prediction device that calculates for each floor and for each direction of the hall call and outputs a corresponding output, and calculates the output of these service prediction devices by combining at least one of the prediction failure probability and the fullness probability with the predicted waiting time. a first evaluation device that evaluates the service status of all floors and outputs corresponding outputs; and a second evaluation device that combines the outputs of these first evaluation devices and outputs the overall evaluation value signal. An elevator group management device comprising a service status evaluation device comprising: 3. An elevator group management device according to claim 2, which uses a first evaluation device that adds and outputs signals corresponding to each of the outputs of the service prediction device. 4. An elevator group management device according to claim 2, which uses a first evaluation device that adds signals corresponding to each of the outputs of the service prediction device, squares the result, and outputs the result. 5. Claims 2 to 4 using a second evaluation device that adds and outputs the outputs of the first evaluation device by performing weighting correction for each floor and for each direction of hall calls. The elevator group control device according to any of the preceding paragraphs. 6. Any one of claims 3 to 5, wherein the hall call is normally assigned to the car whose output from the service status evaluation device is the minimum when the hall call is provisionally assigned to each car. The elevator group control device described in .