CA2122869A1 - Elevator dispatching with multiple term objective function and instantaneous elevator assignment - Google Patents

Abstract

Abstract of the Disclosure Assignment of elevators to hall calls is performed directly as a function of one or more system performance parameters, all related to passenger waiting time.
Assignment of elevators to hall calls is performed directly as a function of one or more system performance parameters by calculating said function as an objective (meaning object or goal) function dependent upon those system performance parameters.

Description

~IPR 26 '94 13:55 FROI~I OTIS INT PROP DEPT TO GOWLING C~N~bh Pr~GE.003~033 '- 21X~ 9 Elevator Di~pa~-ching with ~lt~pl~
Ter~ Objecti~ ~uncti~n and ~nst~ntaneous ~levator a~aignment Technical Field ~he present inventlon r~lates tu assi.gnment of elevator6 to hall calls.

Backqroun~ of the Inven~ion An el~vator di~pat~her causes a particul~r ele~ator in ~ ba~ of elevator cars to ~e ~ent to a floor in ~esponse to a ~se~ pres~ing a hall b~tton at that floor. ~raditionally, a hall lantern will ill~minate 3ust p~ior to ~he opening of the ca~ doors in order to infurm the us~r a~ to ~hich car will ~ervice his hall call.
The disp~tcher as~gn6 a car to a hall ~all according to a varic~y ~f el~ator syst~m p~ra~eter~.
It i~ poe~lble ~or valuee o~ t~Q6a sy~tem uaram~t~r~ to change between the ~ime the hall call i~ r~gi~ter~d and ~he tlme t~e hall call ~s 6er~i~ed. There~ore, the dispatcher ~ay reassign the ha~l c~ o oth~r cars ma~y timos be~ore the hall call ~s serviced. The u~er do~s not notice the Xeassi$~ment because t~o hall lantern iS
lit only after thefi~ ~ultipla reassig~nt~ have ~curre~ an~ just before t~e ~a~ axrives at the ~loor.
2S Ac~ordi~g tv a di6patching schem~t called instantaneous car assignment (ICA), ~nce a car has been assigne~ ~o a hall cctll, the a~ignment may not be chang~d. ~like traditional el~tvator as~ign~ent techniguet~ IC~ in~r~5 the u~¢r at t~e instant o~ ~irst a~sign~t~nt (or sh~r~ly ~h~er~af~er) as to Which car Will servi~e hisJher hall call. ~h~t benef~t is that the user cctn be walkin~ to~ard that partic~lar car, o~ the bank cyDl~t.~air~t~# _E~a?~7~44 ~ C~it _ ~y~5-l~s3~ ~ _ 0~--138~ c~ , t;~ t~ ';s ~ .r;~ n6itlYt ___S~lt ',',.' '. ' ~ ' ' ,',' , ~ , ~,' ~ . ' .
i;. ' . ' . ,~,, :

' ~ 'i ' ' ' : ' hPR 26 '94 1(3:55 FliOI~I OT15 INT PROP DEPT TO GO~JLlN~i ChNhl:l~ PhGE.0~34/033 2122~6~

of cars, which ls going to serve hi~ and be positl~ned and rea~y ~o enter that car when it ar~ives. A know-and-qo time i5 tha time f~o~ ~hen a passeng~r knows which car i5 respon~iny to his hall call to ~he tl~e i~
S takes him t~ qo over to the c~r. Therefore, giving ~he user the opport~nity to be in fr~nt of the car when it arrives re~ir~ that numerous rea~si~nments o~ a hall call to di~ersnt cars can~ot take pl~ce. ~o t~e extent a dispatcher ls spending ~me reassigning, the know-and-go time is ~sed up.
~he reason ~r allowin~ m~ltiple reassi~nments in the past was to obtain the ~est assignm~n~; concerD o~er an initial optim~m assignm~nt was mini~iz~d in the past b~oause there ~ould always be reassignments possible and the~for~ the opportunity to correct for an initial assign~nt th~t had beqome less tban optimum in llght of subsequent even~s ~u¢h as new h~ll call~ and ~ar c~lls~
Und~r ICA, howe~er, P~ause there i6 llttl~ or no ~ime for r~assi~nment~, ~he importanc~ Qf a go~d initial ~0 as~ignment i~ increased.
The ~ir6t use~ of ~CA were not as sen6iti~e to this issue a~ they might have ~een. RelatiVe Sy~te~
~sponse (RSR), taught in n.s. Patent ~o. 4,363,38}
nR~lative System Respo~e Eleva~or Call Assignments", i~
one scheme typically used with the exp~ctation ~h~t multip~ reassign~nts would be allowed. ICA wa~ us~d in Gon~ mction ~dith JISR~ 'rhi6 RSR~ICA sc:~e~e, th~efore, fixed the ~i~s~ cRr to hall call assignment u~ing RSR - a s~heme ~or whi~h the initial assi~n~nt did not account ~or f~ture events ~new hall ca~l6 and car calls) which would serve ~o degrade the quali~y of an initial assi~nme~t. ThQ need for ~ better initi~l assig~ent re~ain#d a~ter RSR/ICA.

s .. .. :- . . , , . , :. : : . :

RFR 26 '94 ~0:5~ F1~011 OrlS INT PROP ~EPT TO GOIIJLING ChNh~ P~lGE.0135/033 .
21~2~.'3 The average registr~ion is the time from when th~
hall call b~tton is p~es~ed to the ti~e that the hall call is canaelled. Thi~ latter point i~ time varies with diff~rent elevator syste~ ~ for some, th~ hall call i8 c~ncelled ~hen th~ car arriv~ at the ~loor and is leveling wh~le fo~ othe~ the hall call is cancelled at a s~op control point typically located where deceleration of the elevator begins a~ it ne~rs th~
floo~ ~egins and a hall lantern is lit. ~o~e that 0 r~gistration time i~ no~ equal to waiting time because not all pass~ngers wait ~e ~a~e ti~e and therefore we cannot easily measure the waiting time of all p~ssengers A
The av~rage r~gis~ration ti~e of an elevator syst4~ i~ a ~o~mo~ me~ric for the p~rfo~mance of that system. ~wever, a good avera~e regi~tration time can be deceptive, hiding an occa~ional, extremely long regi~tration time amonq numsrou~, ve~y ~hort r~gistration ~i~e~. Engineer~ have di~co~r~d that there will u~ually b~ on~ hall call fluring a heavy ~wO-~ay tra~ic ~c~na~o which ~aits a very long tim~ ~or exampl~, 135 ~coond~ hase long waits oc~ur rather ~ quently (for ~xa~ple, once o~ twice in one thousand hall ~alls~. It has heen observed th~t the Associated hall calls ha~e ~f~en ~een bypa6s~d by at least one l~sually sa~eral) car. ~hese b~passe~ h~ppen beca~s~
the Pypassing car wa~ not th~ ~ne assig~ed to the h~ll call a~ the tim~ of the b~pn~ f the bypassing car had stopped ~or the hall call, then the Yery long r~gis~r~tion time could h~ been re~uced~
~ustom~rs ~a~e pointed ~ut ~h~ need t~ red~ce the~e ~ary long reg~tration times. ~y reduaing the nu~ber o~ hall call ~ypasses, a dlspatcher may r~duce the longest reqistration ~ime. At the s~me time, '''. '- .`' , ' ' :
;'. '' ' ~ ' 1' ~ ' - . . . .

` ", ' .. . . ,.' ' ~ ;

`',"~ ' , ,' ~'I" "`' `
.:, ~, .

hPR 2b '~4 1E~:56 FR~rl OrlS INT PROP DEF; TO GOWLI~G C~NflDfl 5RGE.006~033 2~22~6~

however, the average o~ all regi~t~ation tim~s ~ay increa~e because special treatment to a long-w~itin~
call is g~ven at the ~xpense o~ several other hall ~alls. In somQ ~arket~, it is understood t~t ~he m~rket place will accept a higher ~v~ra~e r.egistration ~ime in favor of a lo~er maxim~m regis~r~t~Lon time.
Yigur~ 1 ;llustrat~s thi~ maXi~um regi~tra~ion t~me dilemma a~d the ~ailure of the prior ar~ to addre~
i~. Aceordlng to the prior artI car B is a6signed a hall call at ~loor 7 while car B is ~ading ln the down direction when a new h~ll call ~t ~loor 9 i~ registered, which hall oall i~ as y~ not a~signed. Car A is ~lso ~eading in ~he down direction but i~ ~arther from the hall cal} regi3tered ~t ~loor 9 than car B. According to the prior art RSR schemQ~ aar B Will more than likely be a~s~gne~ to th~ hall call at ~loor 9 beca~se car B is closer ~han aar A ~o ~loor 9. Thi~ is op~imUm for the p~rson who ~istexed the hall call at ~loor 9, b~t tbe p~rson who rQgisterea the hall call at ~laor 7, to Which 2~ car ~ is already c~mmitted, haa ~een waiting ~or ~i~t~
~econd~ ~or a car ~lready wheD t~e hall call at floor 9 was regi~tered~ T~e pe~son at floor 9 has a ve~y short wait, but th~ person at ~loor 7 who ba6 already waited a lon~ timer noW waits even ~ong~r.

Disclo6ure of the In~e~tion Objectives of the present i~ention incl~de reducing the ~aX~m~m re~istration ti~e, while ma~imizing the know-~nd-g~ time and 6till ac~ie~ing a good i~itial .
el~vator assignment.
A~cording to the pre~ent inyentton~ a~signment of cars to Pall ca~ls i~ performed dir~tly as a function of syste~ perform~n~e paramet~rs, rel~ted to pas~nger w~i~ing ~ime including 1) remaini~g response ~i~e ~R~
^ 4 -,: . " ' '' . ''. ' ' RPR C6 '94 113:56 FRC111 OTIS I~T PROP l)EPT TO GOWLING CPNflD~ GE.E~07~33 21~2~
and one or more o~: 2) predicted r~gistr~tion ~im~
(PRT), 3) maximum pr~dic~ed re~istration time (maxPR~) and ~ a relati~e system response ~RSR) qu~n~ty. In still f~r~her accordance with the p~e~nt inven~ion, thQ
~ssignmen~ of elev~tor~ to hall call~ is perfo~ed according to the ~yst~m param~ter~ and reassignQent i3 discouraged ~a in~orporate an instantaneous elevator a~si~n~nt f~ature.
~n advantage is ~hat the ~aiting ti~e of long-~aiting calls is ~educ~d~

Brl~f Description of the Drawin~s Figure 1 is a prior art ~hart of floors m~pped against the location of car~ in a ba~lk o~ ele~ators and registered hall calls.
Fig~re 2 maps ~loors agains~ the location o~ a car B ~nd car call-q and hall calls ~or assignment to car ~.
Figure 3 Ls a mappi~g of floors ~gain~t the location of car~ ~, C and elevator calls associated with t~o~e ~lQ~ators and a hall call a~60ciat~d with car B.
~0 Fig~re 4 i~ a map o~ ~loo~s again~t re~i~terQd hall calls, ~nd the locat~on of car B, C.
Figure 5 i~ a ~ster ~low chart fo~ ~llu~trating th~ ~othod of the present in~ention.
Figure 6 is a flo~ chart ~ a ~a}1 call a~sign~ent algori~hm.
Figure 7 i~ a ~low chart for determinin~ an o~iective function.
Figure 8 ie a graphical repre6entation of an ob3~ctive func~ion ~ith a single i~d~pendent ~ariib}e, showing the e~istence of a ~inim~m ~alue for ~4e o~jecti~e function.

, . -- -- . . .

~ ~ : : :: :. ::-- .
.. , . : . . .
, .
,: . : .,:

:,: . :.~
-.: : ,, ,. . .
-: . .~ .

QPR 26 '94 11~:56 FRO~`1 OTIS l~r PROP DEPT TO GOWLING Cf~ DR P~J5E.00~3~0'3 21~2~6~

Be~t ~ode ~or carrying out the Inv~t.ion The dispa~ching method of the present inv~ntion consiStS o~ t~o parts~ First~ for a n~w ~all call, a c~r i6 as~igned to the call by choo~ing th~ car which provi~es the ~ni~um ~aluc o~ the objQctiYe (~ean~ng goal) function:
OBJ ~iaar) = A-RRT ~ B-¦PRT - 20l ~ t C~maxPR~ - ~o~ ~ D~RSR.
E~ch t~rm is discussed in detail b~low.
~jectlve functions used in el~v~tor dispa~ching are not new, see U.S. Patent 4,947,965 ~uzun~ki et ~1, ~Group Control Method and ApparatUs for an ~levator System with Plural ~ages". ~he R5~ ~lgorithm uses an objectivo ~unction. ~he RSR ~lgorithm and vario~
modifications of it can ~e 6aid to include ~ariou~
terms, dependin~ on the RSR algorithm employed. ~
basic co~ponent of the RSR ~l~ntity is an esti~ate of th~ number of second~ a eleva~or would re~uire to reach a hall call.
~ow~ve~, th~ ~ce of ~he ~ar~i~ular objectlve function, the sel~¢tion of the elQ~Rn~s of the ob~ct funation, thQ u~e of an objective ~unction in combination with ICA ~nd the assignment o~ car~ ta hall calls dlrec~ly Ag a ~unotion of elevator sy~tem performance ~etrics are, a~ong o~har ~hing~ pre~nt~d here, ne~.
~he second part of the inven~ion i~ th~
lnst~ntaneou~ car assign~ent tIC~) feature in combina~ion with the ob~ective ~un~io~ For a hall c~ll tha~ t~as been ~aiting for so~e ti~e wi*;h a c~r alrea~y assignea, ~wi~ching th~ aB~ign:ment to anot~r car iR unlikely according to ~he p~e~ent i~ntion.
Under no ciraums~nces will more than one reas~igD~ent ~e allowed. ~ switc~, that is a rea signment, i6 : ~ . , , .:: ,.......... ~ : -: : -~PR ~b '~4 lE3:C7 FRQM OTIS INT PROP DEPr TO GOWLI~G CQN~ PfGE.13~19/0~3 2 ~
pQrmi~sible under t~o Qxceptional circumstancc~
ther~ is a car other than the a6aigned one that can reach the call ~ignifica~tly faster (for ~a~ple, ~y at least ~0 ~e~and6~ a~d ~) the ~sslgned c~ :Ls traveling away ~rom the call ~for exampl~, ~he car a~5~gned to an up ~11 call i5 traveling upwardly a~ove ~he call~. In the case wher~ a ~witch is p~rmissi~l~, the assignment iS made based on the ob~ective ~'~nction. The values o~
the coefficients ~, ~, C, and ~ can be v~rie~ to r~lect the preference of the buildin~ owner. It is al~o clear tha~ ~y etting all but one coef~icient to z~ro, dispa~ching assignments c~n ~e made based on a single metric.

The ter~ remaining respons~ ti~e is fully desori~ed in U.S. Paten~ 5,146,053 entitled ~Ele~ator Di~pat~h~ng Based on Remaining Respon~ Time", i~sued to t~e ~a~e i~entor~ a~ t~e pr~s~nt inve~ti~n. It 15 an e~timate of the n~mber of seco~ds an clevator would reguire to rea~h the h~ll call unde-r con~deration given its cu~r~t ~et ~f a~signed car ~alls and hall ~al1~.
t is sometimes ref~rred to in ~he eleva~o~ indus~ry as esti~ated time o~ arri~l t~A)~
F~gur~ 2 illustrates ~ car a ~oving in the down 2~ direc~ion ~nd positioned a~ floor 12 on its ~ay ~o service a car call at floor 9. At this poin~, a n,ew ~all call is regis~er~d at floar S. The remain~ng re~on~e tlm2 for t~e new h~ll call fo~ ~ar ~ i~ an exempl~ry lS seconds. A ~BW ~econds lat~r, a~other hall 3~ call i~ assigned when the car ~, still moving downwardly i~ the dirac~ion o~ its car c~ll at ~loor 9 ~n~ a~signed hall ~all ~dt floor 6, when another hall ca}l i6 asgig~ed to it ~t fioor 10. The additional hall call a~ floor 10 ~:,: , , :~

.. . : ~
~:. ~ :. : -:

. :: - . . - , .
.-f~PR 26 '94 10:5~ FROM OTIS INT PROP DEPT TO GOIIJLING Cl~hDh P~GE.010/033 2~22~9 increas~s t~Q remainlng response time of t~e call at ~loor 6 to 25 seconds ~ro~ 15 s~conds.
Flgure 3 maps ~lo~rs in a ~uildlng against car calls ~or cars B and C and a hall call assigned to car B. Figure 3 ~llustrates the rem~ining response time comcept af~er ~ h~ll call has alrea~y been waiting an exemplary time o~ 20 ~econds. In Figu~e 3 a car B i~
tra~eling in the do~nward direction ~o 6er~ice two c~r call~ be~re servicing a hall call as6igned to car B
lo wh~re ~he pasS~nger h~s a~ready ~Qen wai~ing or ~o seco~ds. Meanwbile, a car C is mo~lng in the up~ard direc~ion to serv~Ge a car cal~ at a floor abo~e th0 loca~ioP of the ~all call. The question arises as to whether the hall call should r~ain assigned to car s or b~ re~ssig~ad to car C.
~here thQ ~signme~t of ~ars to hall ca~ls is based purely on re~Ainin~ response ~i~e, t~ re~aining r~ponse ti~e ~or a~ nment to aar B i~ co~par~d to the re~a~n;ng res~on~e ti~ Por car C to e~aluat~ th~ merit o~ ~h~ curren~ asslgnment aT-d d~termln~ wh~her ~
~witch, ~at i6 a r~a~;s~gn~entr ~om car B to c~r C
would be a good id~a.
Al~o~ if the trip to raach a hall call in the opposite direction includes an assi~ned hall call in the ~irecti~h o$ tra~el, thQn fo~ the purposes of remaining re~pons~ ~i~B computation the car 1~ as~umed to go to ~he ~erminal ~loor. (For example, consid~r ~ car t~aveli~g u~ a~ floor five wit~ a c~r call at 7 and an assigned ~all call at floor 9. Now, a ~own c~l} is 3~ registered at floor ln. To e~ti~ate ~he remaini~g respo~se ~i~e o~ the car, the car is a~aum~d to bQ s~nt to the ~op terminal to f~lfill th4 car aall resultin~
~rom the hall call at floor 9 be~ore it can r~a~h ~loor lQ in the down direction). Upon reflection, it can ~e :.: : : ~ ~ . . : : , . : .

flFR 26 '94 lE~:57 FROI~I OTIS INT PROP DEPT TO GOl~lLlNG Ci:NhrlR PhGE.011/033 2122~69 seen th~t ~is assumptio~ that the cars go to the terminal floor is ~o~ necessarily the ~ors~ case.
We assume that only one c~r ~all results fro~ the up hall cail at ~loor g, and that is to the terminal floor (the top). A ~uch wor~e situation w~uld be if several people wero waitin~ behin~ t~e hall call at floor 9, and each pressed a di~erent car call button.
For this worse case, the RRT would obvio~sly be muoh longer, due to addi~ional s~ops.
~
~hi~ metric i~ the sum of ~he a~oun~ of th~ time that the call ~as already been w~iting (the wait ti~e-~o-far) a~ the RR~. For a new hall call, PXT ~
~i~ure 4 illustrates why assignment of hall calls based solel~ on re~aining re~pon~ time is not su~icient for .good hall call a~sigPment~ and why predicted regis~ration ti~e ~s importa~t. Car B i~ presently at floor 11, car B is moving downwardly to service a ha~1 c~ll as~igned to it ~t ~loor 6 where the passenger'~
~ait tim~-~o-far is ~a very lon~ 50 s~cond~ whe~ ~ new hall call i~ regist~red At Cloor g. Anoth~r car C a~
~loor 14 i~ al80 ~o~ing downwa~d~y. Th~ re~ainlng re~ponse ti~e o~ car B ~or th~ new hall c~ll a~ floor 9 is six seeonds. ~he remaining re~pon~e ti~e o~ the car Z5 C ~ respect to the new hall call a~ 100r ~ is 15 seconds, becauce the car C i~ ~art~er a~ay from the ne~
hall than car B. It would soem at this point that the logical ~el~ct~on for ~he as6igpment ~or th~ hall call is car B. Under certain ~ircumstances, this assign~nt 3a ~ould not be approp~te, however, bec~use of t~e e~e~t of that assign~en-t on other cal~. The pre~iated re~i~tration time ~or tha call at ~loor six if car B i~
a~igned to the h~ll call ~t floor 9 is increase~ to 65 se~o~ds. ~e predi~ted rogi.stration ti~e ~or t~e ~all _ g _ ,. ... ,, , .,;:: . . .. ::. , .,.: ,., .. , . .. : . ~ :
,. . : . ; .: , .: .

: , . :: :::, .. : .: :
::: : . . - :
: : . .

,~,:: : .. ~

., : ~., : :

hPR 2~ 'q4 10:58 FRO~ OTIS INT PROP IIEPT TO GC~ LIN5 C~ D~ PflGE.al2/033 ~12~6~
dt floor ~ i~ car B i~ as~igned to the hall call a~
floor ~ is 55 second~. ~hus, assignin~ the car B to the new hall call at floor ~ based on ~he ~horte5t r~maining response time co~pari~on ~or the two cars res~lts in a very long predict~d regis~r~tion tim~ ~or ~he passenger a~ ~loor 6. Ihe p~ed~cted r~gistration tine resu~ts wher~ an assignment is ~ade purely as a f~otion ~P ~be remaining response ti~e ~tric is poign~nt where ~s ~n ext.ra 10 secon~s of waitin~ for the passe~ger at floor 6 is the di~fer~nce between ~n anxious passen~er and a fu~io~s passenger, as a conse~uence of ~he nonlinearlty of passenger fru~tration as a fun~tion o~ waiting ti~e.
~ce, the wisdom of including the predicted regi~t~ation ti~e in the o~jective function.
l T~e predicted re~istration time metric is incl~ded ln the objeotive func~ion as the ah~olute value o~ the dlfferenc~ be~ween the pr~dicted r~gistration tim~ An~
the term, T1, 0~ 20 ~cond3. I~ the predicted registration tim~ lther v~ry short or ~er~ long~
then t~e ten~, T1, punaliz~s a car. ~his rQ~lects the philosophy in fiome mark~s t~at a passen~er i5 willing to wai~ approximately 20 secon~s without a~y level of discom~ort. 0~ cour~o, thi~ penalty ter~ is variahle and need ~ot be ~0 ~econds. Therefore~ a car t~at could 2S r~ach the hall call in a very s~ort time ( or example, five seconds) might be~ t~r proc~ed to answer other.~ore urgent elev~to~ ~ystem demands.

Waiting times in eXc~;s of ~0 seconds are considered very long wbil~ their ~requency i~ low (onc~
or twlce in a two hour heaY~ two-way traffic). Their effect is a ~ajor irritant to passengers. It is i~porta~t to reduce th~ ~agnitude and freguency of these long-waiting calls. The present invention p~opo~es to ~ 10 --;'~' , ' ' ' ~
,:. ~ , ,:

RPR 26 '94 10:5S FROt~l OTI5 INT PROP DEPT 'O GOWLING CRNRDR K~GE.01:~/033 2122~

address thesu lan~ aall6 by penalizi~q ~h~ car for an a~ign~nt only when that a~sign~ent Will cau~Q the longe~t waiting call (o~ all hall calls presently waiting) to ~ait long~r than a term, T2~ 6~ second~. It i5 thought that a call ~ha~ has alre~dy wait2d 60 seconds has a poten~ial to cros~ ~he 90 seconds thre~hold and therefo~e should be given sl?ecial conside.ration. ~e p~alty terL i~ variable ~nd neQd no~ be 60 seaonds. ~he ~erm i~ ~q~ared i;n the o~jective ~unction to reflect ~he passengers gro~ing irrita~ion whioh i~ ~lt to be nonlin~ar and inc~easing as the waiting time inoreases beyond 60 secon~s. Obviou~y, the term ~axP~T, like PR~, need not be sguared ~ut could ~e the a~gument for any o~her function to ~odel pass~ng~r irritation~ ~he Dirac ~el~a operator ensures th~t the third term is zero ~here ~axPRT i not longer than 60 s~conds.

T~is ~etric is used cur~tly in the objQctive ~nction in or~er tc allow the ~ilding awn~r to re~rt ~o ~he prior a~t RSR di~p~tchi~ methodology.
The v~lue of the RBR ~erm ~lected depend3 upon whieh ~or~ of RSR is de~ired, as it ha~ many modi~ica~ions. ~he b~ia compon~nt of tho RSR ~uantity i5 the e~ti~ated a~ount o~ ti~e ~or a c~r to reach ~he hal~ call whose assign~ent i5 bQi~g qeter~ined. The valuo select~d, ho~ever, for the ~SR value may be ~ny o~
those shown in U~S. Pate~t 5,146,053 issued to Powell et al entitl~d Elevator ~ispat~hing Based on Remainin~
Response ~im~; ~.S. Patent 4,3~3,381 issued to Bi~tar, entitled Relative System Response Elevator Call Assign~ænt~; U.S. Patent ~,~85,5~3 ~o Bi~tar en~itled Weighted Relative System Rlevato~ Car Assignment System with Varia~le Bonu~es and Penal~ies; U.5. Patent ~F'R 26 '94 10:5~ FRClr~ OTIS I~T PROP DEPT TO GOI~.ILING Chl`l~Df~ PRGE.~114/033 - 21%~69 4,7~2,g21 ~o MacDonald ~t al. entitled ~oin~ident Call Optimization in an Elevator ~i~pa~ching Syste~5 ~.S.
Patent ~,202,540 is~ued to A~er antitled Two-way Ring communicatio~ sy~tem ~or Ele~ator ~roup Control; U.S.
Patent 5,168,136 i6sued to Thangavelu ~t al en~itled Lear~ing ~ethodology fo~ Improving ~ra~fic Prediction Accuracy Or Elevator System ~sin~ Artificial Intelligence; U.S. Paten~ ~,o~5,302 i~sue.d to Thangavelu entitled Art~icia} lntelligence based L~arni~g System Predicting Peak-Period Time4 ~or Elevator Dispatching; U.S. Pa~nt 5,024,295 i~sued to Thangavelu entitled Relative Sy~tem ~espon~e Elevator Di~patc~er S~stem Using Artificial Intelligenc~ to ~ary ~onuses and Penal~ies; U.s. Patent 5,0~2,4g7 issued to Thang~velu entitled Artificial Intclli~ence s~e~ C~owd Sensing System for Elevdt~r Car Assignmenti and ~S.
Patent 4, 83R, 384 issued to ~hangavelu en~itled Q~ue Based Elev~tor Di~patchinq System Using Pe.ak Period Traffic Prediction, inc.orpora~ed by reference. Th~
~0 bonuses und penalt~es m~king up the RSR te~ c~n be ~aried or fixed.
Figu~ 5 i6 a ma~ter -flow cbart for lmplemQntin~
the method of the pr~sent invention. After a start, a hall c~l1 at a floor N in a given direction is 2~ regi~ere~. Then, an el~ator dispatcher determines if the hall call was pre~ou~ly ~ss~gned to a car and records the car o~ the a~signment. ~ext, the remaining respon~e time i5 calculated ~or each c~r in the bank and the lowest remaining response t~me ~nd the car associated with it is deter~ined.
A series of tests i~ now ex~c~ted to determin~ if a hall call assign~ent algorith~ (Pig. ~) for reassigning th~ call should be executed~ ~h~ rou~ines of ~igures 5, 6 and 7 ~nGorpora~e the basic concsp~ of .'.,' ~' ': ' ' " '' ;-' -: , :~ , , ', ~ '':, ~, .~: ~ ' '' . '' '~'':
.,:''` ' ' . :,' ".;. . : ., ~F:R 26 '44 1E) J8 FROI1 OrlS INT PROP DEPT TO GOWLII`IG Cfd`~FlDi~ P~GE.015/033 ~22~S~

instantaneous car assignment in th~t the call is not 7-eassign~d unle6s there are s~rong incentiVa~ ~or doing so; eve~ then, no more ~han one rea~ign~ent is allow~d, The ~irst t~st ask~ "Is thi~ a new hall c~ll?n. If ~o, co~pletion of the routine o~ Figure 5 ~ait~ for exec~tion O~ the h~ll ca}l ~ssignment algorith~
illustrated in Figure 6. If not~ the ~ext three teqts may ~e executed ~or determining w~ether ~he ~r~viously as~i~ne~ call should be rea6si~ned. In te~t two, if the remaining respon6e time of the assigned e]Le~ator is greater than the lowest ra~aining response ti~e plus seconds, execution of ~he routine-at Fi~ure 5 wait~
un~il execution of the hal~ ~all as6ign~ent algorith~
(Fig~ 6) for possible rea~ignment o~ the hall call to another car. Thi~ test indicates that reassignment ~s strong~y discouraged but i~ ~he re~aining respon~e tlm~
o~ the ~resent car is extremely poor with respec~ to the lowest re~aining re~ponse timR then reassi~n~snt should be Go~sidered~ Extrem~ly poor i~ d~fined by a ~ariable pr~d1Gted re~i~tration time differe~ce, here 40. Th~
third and fourth test~ ~tall execution o~ the rou~in~ o~
Figuro 5 until the h~ll call a~signment algorithm ~
executed i~ th~ assigned ca~ ~s traveling awa~ ~rom the as~igned call. None of these te~ts ~eing met in the af~irmative, th~re is no re~sslgnmen~. .
~ig~re 6 lllustra~es th~ hall call as~ignment algorithm. First, ~hQ rem~ini~g re~ponse ti~e ~lready computed ~or the current ~et of a~;signmer~t:6 of ha~ 1 calls to cars i~ ~ead and used for computing th~
3n predicted registra~io~ ~ime (PRT) for all hall calls, by adding th~ w~it time-so-far ~or e~c~ call to the as~ociated remaining respon~e ti~. ~ext, a c~r index ~car i~ set to zero. The index is incremented by one ~or each c~r i~ the ~anX, and a m~lti-terc ob~eative s~i .:

~PR 26 '94 10:59 FROM OTIS INT PROP ~EPT TO GO~LING Cf~`lRDR P~GE.016~033 function is ~omputed for that car~ u~til ~11 cars ha~
be~n ~onsidered. Next, the car with the lowest objective ~un~tion is d~termined and gi~en a label ~AR~
A series of tests is then exec~ted for dete~mlning Whether there should be a r~a~ignment. ~hese thre~
tests are si~ilar to the ~our ~es~s o~ Figure 5 insofar a~ their ~xecution infre~ently res~lts in reassign~ent of a call out of deference tn insta~taneous c~r a~ignmant. In the first te t, i~ the ha~l call is a new one, then t~e hall call is assi~ned. If ~h~ hall Gall is not a new call (test two~ ~nd ~he call has alrea~y been switched once from t~e car of first assign~n~, then ~he hall call i5 no~ reassi~ned. If t~e call is not a ~aw one, then the predic~ed registration time (PRT) of the a~sign~d car is comp~red with the predicted re~ ration time ~PRT~ o~ the car, ~aR", wi~h th~ lowes~ object~ve function. If the predlated registra~ion ti~e (PRT) o~' ~he ~6signed c~r is ~r greate~ t~an the predi~ed regist~ation ~imo o~ the el~vator with the lowest objective ~unotiQrl, then t~e hall call i~ r~n~lgned ~o the el~ator c~ ~RaR) Wi~h the lowest objective ~unction, but otherwi~e~ no reassignment occurs.
Figur~ 7 illustrates ca}culation of the multi-term ~5 objeati~e function. First, the ~it time-So-~ar ~or e~ch hall call is ~tored a~d ~appe~ a~ainst th~
direction of that hall call. Next, the car ~or which the ob~ecti~ function is being calculated is ass~m~d to b~ igned to ~he call ~ein~ considered ~or reassignment in t~e ~ater flow ch~rt routine, Thi~d, ~h~ remain~ng respo~se time (RRT)~ predi~ted ~gistration time (PR~, maximum predict~d registration tim~ ~maxPRT), and the RSR value are c~lculat~d. The valu~s for the four ter~ of ~he m~ erm obj~ativ¢

',:: . :
:.. , ' ' ~ ' . . ~

: ~,: . ., . ' . , hPR 26 '94 1~3:59 FRO~I OTIS INT PROP DEPT TO GOWLING Cf~ DFl Pf~GE.017/033 ~ ' 2~C~2~)9 function ~re now calculated ~nd summ~d for producing the ~ult~-ter~ objecti~e function for use in the assign~nt algorithm hall c~
Fi~ure ~ is a gra~ of the oPjectiYe function af the ~rs i~ a bank; the car wi~h the mlni~l~m value o~
the obj~ctiYe ~unction (c~r B) is assigned ~o a hall c~
Various chan~es may be made ~ithout departing ~ro~
the ~piri~ and soope of the inve~tion.

- 15 ~

, .

Claims (13)

1. A method for assigning a hall call to an elevator, comprising:
a) providing a remaining response time associated with a elevator and said hall call;
b) providing a predicted registration time associated with said elevator and said hall call;
c) providing an objective function in response to said remaining response time and predicted registration time;
d) performing steps (a) through (c) for every elevator available for assignment of said hall call;
e) comparing said objective function for all elevators available for assignment to said hall call for producing a comparison result;
f) assigning an elevator to said hall call in response to said comparison result.

2. The method of claim 1, further including providing said objective function in response to a maximum predicted registration time associated with said car and said hall call.

3. The method of claim 1, wherein a relative system response (RSR) quantity is included in said objective function.

4. The method of claim 1, wherein said remaining response time, and said predicted registration time, are combined linearly to perform said objective function.

5. The method of claim 1, wherein said assignment is disallowed if said hall call has been previously.
assigned more than a limited number of times previously.

6. The method of claim 1, wherein said assignment is disallowed if said hall call has been previously assigned more than a limited number of times previously and wherein the limit on the number of reassignments is excepted where:
a) the predicted registration time of the elevator presently assigned to said hall call is greater than the predicted registration time of an elevator associated with the lowest value objective function by predicted registration time difference, or b) the elevator assigned to said hall call is traveling away from the floor of said hall call.

7. The method of claim 1, wherein said objective function is of the form:
OBJ (icar) = A?RRT + B? ¦PRT - T1¦
+ .delta. ? C ? [(maxPRT) - T2]2 + D?RSR
where OBJ (icar) is the objection function;
where RRT is the remaining response time;
PRT is the predicted registration time;
maxPRT is the maximum predicted registration time;
T1 is a time value;
T2 is another time value;
RSR is a relative system response quantity;
.delta. = 1 if maxPRT > T2 0 otherwise.

8. A method for assigning a hall call to an elevator, comprising:

a) providing a remaining response time associated with an elevator and said hall call;
b) providing a predicted registration time associated with said elevator and said hall call;
c) providing a maximum predicted registration time associated with said elevator and said hall call;
d) providing an objective function in response to said remaining response time, said predicted registration, said maximum predicted registration time;
e) performing steps (a) through (d) for every elevator available for assignment of said hall calls;
f) comparing said objective function for all elevators available for assignment to said hall call for producing a comparison result;
g) assigning an elevator to said hall call in response to said comparison result.

9. The method of claim 8, wherein a relative system response (RSR) quantity is included in said objective function.

10. The method of claim 8, wherein said remaining response time, and said predicted registration time, are combined linearly to perform said objective function.

11. The method of claim 8, wherein said assignment is disallowed if said hall call has been previously assigned more than a limited number of times previously.

12. The method of claim 8, wherein said assignment is disallowed if said hall call has been previously assigned more than a limited number of times previously and wherein the limit on the number of reassignments is excepted where:

a) the predicted registration time of the elevator presently assigned to said hall call is greater than the predicted registration time of an elevator associated with the lowest value objective function by predicted registration time difference, or b) the elevator assigned to said hall call is traveling away from the floor of said hall call.

13. The method of claim 8, wherein said objective function is of the form:
OBJ (icar) = A?RRT + B? ¦PRT - T1¦
+ .delta. ? C ? [(maxPRT) - T2]2 + D?RSR
where OBJ (icar) is the objection function;
where RRT is the remaining response time;
PRT is the predicted registration time;
maxPRT is the maximum predicted registration time;
T1 is a time value;
T2 is another time value;
RSR is a relative system response quantity;
.delta. = 1 if maxPRT > T2 0 otherwise.