JPS60209475A - Method of controlling group of elevator - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a group management control method for a network operator, and particularly to an improvement of a distributed standby control method using a learning function.

[Technical background of the invention and its problems]

In recent years, group management control devices that control a plurality of elevators have become common, using small computers such as microcomputers. Therefore, we maintain past data such as elevator trends and hall call occurrences (
Memorization) has become easier.

However, most of this learning data has conventionally been used as an element for determining the allocation of newly generated hall calls, and there are almost no examples of it being used in a distributed standby control method for machines that are free and on standby. In the conventional distributed standby system, if there are no machines already on standby in multiple floor zones or floors specified in advance, the machines are put on standby in a certain priority order. Method (Standard floor as the waiting floor.

(generally in ascending order).

With this control method, even if there is no waiting machine on a floor with higher demand than the standard floor of the first priority floor, such as a banquet hall or restaurant, if there is only one waiting machine, the waiting machine will first go to the standard floor. When the machine is on standby at the standard floor, there is a high possibility that a hall call will occur on a high-demand floor and that the hall will be assigned. If the demand forecast for each floor is not used in this way, standby control is often useless.

Furthermore, even if the standard floor and the top floor, for example, the sky restaurant, are in high demand, there is a probability of about 5θ% that simply having customers wait on the standard floor will be effective.

Although the purpose of distributed waiting is to improve service for the next hall call, if there are multiple floors with high demand for hall calls, distributed waiting does not contribute to improving service. do not have.

[Purpose of the invention]

SUMMARY OF THE INVENTION Therefore, the present invention provides an elevator group management control method that can determine an effective waiting floor using learning data in determining the waiting floor of a waiting car, and perform distributed waiting control that matches predicted demand. It is an object. ( [Summary of the Invention] In order to achieve the above-mentioned object, the present invention provides information for each floor according to major factors that determine traffic patterns, such as day of the week, time of day, Rokuyo (Rokuki), and whether or not it is a holiday. Group management that stores learning data such as "hall call occurrence rate," predicts high-demand floors based on this learning data, and performs distributed waiting to match the predicted results, that is, the distribution of high-demand floors. This is a control method.This control method is used, for example, when demand is high on the standard floor and the top floor such as a sky restaurant, and the two sides are equal, and one of the machines is on standby.
Distribute waiting to suit the traffic flow at the time, such as distributing waiting to intermediate floors, and conversely, when demand is low and it is quiet, do not force waiting on a specific floor, but meet demand. The system controls the degree to which it is spread across different distributions, allowing it to respond to changes in demand from time to time.

[Embodiments of the invention]

The following describes a case in which an elevator group management control method according to an embodiment of the present invention is applied to a group of four elevators in an eight-story building. A control device for group management control of such four elevators is constructed as shown in FIG.

That is, a hall call command issued from an elevator hall on an arbitrary floor is registered in the large hall call registration circuit 1 as a hall call command 1. The information is stored separately for the floor where it occurred and the desired direction, and this memory is reset when the elevator car arrives at the floor where the elevator car occurred. 4 elevators (A,
B, C.

D) each has an elevator operation control device 2A.

2B, 2C, and JD are provided (however, 2B and 2C are omitted in 1).

Inside each elevator operation control device 2A to 2D, there are car status buffers 3A to 3D that temporarily store the car status such as the position, driving direction, and load of each elevator car. The call registration circuits 4A to 4D memorize the set call registration floor and reset the registration when the above call arrives at the registered floor.
is accommodated.

5 is a small machine such as a microcomputer with 16 pits, and the above hall call registration circuit l
The information on the hole status stored in the above compact 5g5s
The hall condition table (hereinafter 1 (C
(abbreviated as T) 9. In addition, each car-shaped book buffer 3A to 3D of the multi-elevator operation control device 2A to 2D
The car condition information stored in the RAM is stored in a car condition (hereinafter abbreviated as ccT) 10 provided in the RAM via input registers 7A to 7D, respectively. Similarly, the car call status information stored in each car call metal edge circuit 4A to 4D is stored in a car condition table (hereinafter abbreviated as KCT) 11 provided in RAM via each input register 8A to 8D. be done.

The HCT9, CCT10, and KCT11 have bit configurations as shown in FIGS. 2, 3, and 4, respectively. That is, in the HCT representing the hall state shown in Fig. 2, %8 bits of information are stored from the descent of the 8th floor (8D) to the ascent of the 7th floor (7U) for the hall sub-indexes (H8) from 0 to 13. Stored. The hall conditions for each floor will be specifically explained. For example, when the up switch is pressed in the elevator hall on the 5th floor, H81
When the 7th bit of 1 (5U) becomes 1 and the service elevator corresponding to this hall call is determined to be car A using the method described later, the 0th bit and 8th bit of H8II become 1. If the above-mentioned Unit A is on the same floor as the 5th floor, H8
Bits 0, 6, and 7 of No. 11 are all reset to 0. That is, the θ~3run bits indicate the number set of each elevator, the 6th bit indicates whether the elevator is assigned to a hall call, and the 7th bit indicates the presence or absence of a hall call.

In the OCT representing the car status in FIG. 3, 16 bits of information are stored for each index from 0 to 3 for elevators A to D. That is, 0~
The third bit indicates the load state of the car in binary notation. The meanings of these bits 0 to 3 are “0001”, 0
010'.

0011”, “0100”, “0101”,
"0110".

"0111", "1000", "1001",
"1010".

For “1011” and “1100”, θ
~10%, 11-20%, 21-30%, 31-40%
.

41-50%, 51-60%, 61-70%, 71-8
0%, 81-90%, 91-100%, 101-110
%, 111% or more. The 5th check pit indicates the running status of the car, 1# indicates running, 0" indicates decelerating. The 7th bit indicates the open/closed state of the door, 1" indicates opening, @0" indicates closing button. The 8th to 13th bits indicate the car position in binary notation.

14.15% pit indicates the direction of movement of the car, '10''
is ascending, ``01'' is descending, and ``oo'' indicates no direction, that is, stopped. In the KCT representing the fourth prisoner's car call state, like the HCT in Fig. 2, 0 to
Three pits indicate whether there is a car call for elevators A to D.

Next, a hall call occurs on any floor, and the most suitable service elevator for that hall call is selected by HOT 9.
, CCT 10. The process up to closing the furnace based on the information of KCT 11 will be explained using FIG.

These processes are performed by software, as shown in FIG. After the start, the task management program 2o determines which task (software module separated by function) is to be started. Here, explanations of each task and the RAM and ROM tables to be used will be briefly explained in the process up to the allocation of the optimum machine.

21 is an external input task that sets external inputs such as the above-mentioned OCT, KCT, )ICT, etc., on the RAM. If this external input task 2 is started with a high priority, it will be restarted every 100 ms or so.

Reference numeral 22 denotes an allocation task that performs allocation, which checks newly generated hall calls every 100 ma or so, and if a new hall call occurs, performs an evaluation based on expected unresponse time, damage such as fullness, etc., and determines the hall number with the best evaluation.

26 is the allocation light adjustment task, which is a low-level task that is activated approximately once every second, and is used to change the allocation in response to a long wait, a full hall, or a predicted hall call. be.

27 is a data processing task, which uses external input or data from a single elevator to set the data table for the current state, and replaces that data when changing to the next state. Yes, and is activated when data or state changes. It is also a low-level task and is activated so as not to harm higher-level group management tasks.

Reference numeral 28 denotes a task for communicating with each individual elevator, and in addition to cyclic data transmission, this also outputs assignments, assignment cancellations, etc. as necessary, and requests data such as the number of people getting off, new car calls, etc. .

29 is an annual timer and various timers, 10m5, 10
It is a routine of various interval timers such as 0m5, 1 second, etc. and a yearly timer combined with them, and
These data are corrected by an external timer. Month for yearly timer.

Contains information such as dates, days of the week, holidays, Rokuyo, and other events.
I10 Task (2) 3 floppy disk drive, Il
o Task (1130(7) CRT etc. grinning tjv
information will be updated.

Reference numeral 30 denotes I10 task (1), that is, a (character display terminal) transmission task, which is used for transmitting information with external terminals, computers in the pond, and the like.

This task i10 task (1) 30 is time-sliced and activated at a low level so as not to overwhelm other group management tasks.

31 is the I10 task (2), ie, for floppy disk control, which is activated when learning data, etc. are stored on an external floppy disk, and is activated at a lower level like the old task.

32 initializes RAM and CPU registers,
This is an initialization task that initializes the LSI, and is activated when the initial state or operation mode changes.

Generally speaking, these tasks are activated during lower-level tasks. However, if a special flag or a change in loved one 1- is made, the value changes. The task management program 20 manages these tasks, and the RAM for the buffer 7 is used to exchange information between tasks.

23 in Figure 5 is a task for distributed standby control, which is a task that performs distributed standby when it becomes free and there is a huge problem in one room time, and it is a task at a lower level than allocation and external input/output. be.

Next, routines related to the gate beta group management control method of the present invention will be explained step by step.

In FIG. 5, when the start is started, the initialization task 32 is started, and the annual timer and various timers 29 are started.
RAM is cleared, preset, etc. Also,
OCT, KCT,
Inputs such as HCT are input.

Tasema etc. are also counting up.

The learning function part will now be explained with reference to Fig. 11. It is assumed that each floor of the building is configured as shown in Fig. 6, and that it is a group elevator in a hotel.
IF~8F), and the group is assumed to be 4 vehicles.

Based on the above assumptions, the traffic volume on each floor changes in various ways, but the following are the reasons for this change in traffic:
Possible factors include Monday, June 6th, whether it is a holiday, and the time of day. Therefore, the traffic mode set is Monday, Rokuyo,
Determined by holiday and time zone factors. Here, the traffic mode is TRMOD, the month is MON, the day of the week is WEK, the sixth day is ROY, the holiday is HDY, and the time zone is TMB.

Their contents and meanings are shown in Figure 7.

In Figure 7, it is determined by the five factors mentioned above, but other factors such as changes in traffic volume may be added, and the time period is determined by 15.
Minutes are used as one interval, but this set can be changed depending on the building. Also T RMOD
is determined by a matrix of these elements,
This shows the number on the RAM. This number may remain the same even if the above-mentioned elements change, and the set of numbers may be set in advance, or may be automatically corrected by the function of the microcomputer.

TRMOD = TRMOD$DATA (MON,
WEK, ROY.

HDY, TMB) In other words, TRMOD is shown by the above equation.

TRMOD$DATA contains traffic mode information for each element.

The TRMOD$DATA that does not go into the RAM is stored on a floppy disk, etc., and the parts that are used are stored in the RAM. In the eighth example, the TRMOD by 'l'MB (time zone) is loaded onto the RAM.

Every time TMB advances, today's previous part is Figure 11 40
1. It will be updated tomorrow as shown here.

TRMOD=TRMOD$RAM(TMB) Therefore, the above formula is used instead.

To illustrate this example, assume that MON=4 (May) WFjK=0 (Sunday v4) ROM=O (Daian) HAY=1 (holiday) TMB=72 (18:00). From this data, it is assumed that TRMOD = 3 (Daian, holiday, spring, evening, check-in).

As shown in FIG. 11, the following types of learning data for this TRMOD are prepared.

HCT$RAT: Average number of hall calls generated during 15 minutes.

IN$RAT : Average number of passengers oUT$RAT : Average number of people getting off the train KCT$RAT : Average number of car calls ordered KCT$SE
T: Call occurrence rate for each floor These HCT$R
AT, IN$RAT, OUT$RAT.

The KCT$RAT is indicated by the hall sub-index (H8), which takes into account the direction of the floor, as shown in Figure 9.

Further, KCT$SE'l' is the occurrence rate for the position (floor) as shown in FIG. Lobby (IF)
From the guest room (6th floor) call occurrence rate (jKCT$5ET
(1, 6).

Furthermore, as a reservation function, if a high-value floor is specified from a CRT terminal, etc. (when the date and time and floor are specified from the CRT and the floor is unclear), the HCT$ RAT is corrected based on that data. .

A data table of past learning is prepared in the above manner. These are T in the data processing task in Figure 5.
Every time the time zone TMB of RMOD (traffic volume mode) changes, the previous learning data is transferred to the processing buffer as shown in No. 11 @ 41, and then the previous learning data table is transferred as shown in Fig. 11 @ 42. The next data table is loaded, and the next learning data table is updated as shown in FIG. 11 (43). Note that if the TMB of TRMOD does not change, the processing flow moves to B in FIG. 11. In this case, higher-level tasks such as allocation are prioritized. This completes the explanation of the routine for preparing past learning data, then finishes the explanation of the routine for preparing current learning data, and then describes the current learning data collection routine. As shown in FIG. 12, when collected data is collected, it is set in the learning data collection table as shown in FIG. 12 44. It is set in the learning data collection table in this case. It is set in the learning data collection table in this case. In this case, the learning data collection table shown in Figures 13 and 14 is n)-h4.
1; Z-to fp 4) f) I$DATA: Number of generated hall calls on multiple floors (with direction) C$DATA: Number of car calls on each floor (with direction) CN$
DATA: Power when responding to each hall call) Number of calls IN$DATA: Number of passengers on each hall call floor OUT$
DATA: Number of people getting off at car call floor CZ$DATA:
Occurrence of car calls on the destination floor for hall calls on each floor I fixed numbers H$DATA, C$DATA, CN$DATA
, IN$DATA.

OUT $ DATA has respective data by H8 (hall sub index) which is a directional floor,
CZtDATA G'i is held by the octopus 11 and is set as a set of two q indexes of the destination floor and the start floor.

The data to be set in these tables is sent by the external input task 21 shown in FIG. 5, etc., and the boarding/alighting information is sent to the buffer RAM by the data processing task.

This buffer RAM f is as shown in FIG.
The time, status, and data are attached to form a queue, and the format of this data is as shown in FIG. These are processed and set in the tables shown in FIGS. 13 and 14, and if this is the data processing interval time, the learning data is updated with the contents of the processed data buffer as shown in FIG. 12 45.

Next, a routine for reflecting the learning data in the processing buffer to the previous learning data table will be explained. The aforementioned H$D
ATA, C$DATA, CN$DATA.

IN$DATA, OUT$DATA', CZ$DAT
Since each of AZaraζTi1E (recorded in the RAM at the interval time) is possessed by TRMOD (transaction amount mode), processing is performed using it.

The data processing interval IITMIi: is performed once a day, usually during off-peak hours at night.

The used and collected TRMODs are processed accordingly. The actual data and the previous learning data table correspond to the relationship below.

(Learning data) (Collected data) (+ Nito calculation) I NORAT・・・・・・IN$DATA/)i$D
ATAOUT$RAT・・・・・・OUT$DATA
/C$D ATAKCT $RAT・・・・・・=C
N$ DATA / H$ D ATAKCT$5BIT
(PO8I, PO32) -...Hatanishimaka screams ■X
100ΣCZ$DATA When converting to the same direction and flattening the newly collected data to learning data, smooth the new learning data 2 weights new collected data + (1-singing) ≠ old learning data (0≦weight≦1) (Normal exponential smoothing).

Next, the automatic correction +1 & function of TRMOD will be explained.

For each TRMOD, if the square of the difference between the respective learning data is within a certain upper limit, the TRMOD is absorbed into the lower numbered mode. In this case, the learning data will be the average of both.

The group management control method of the present invention is characterized in that learning data obtained by the learning function described above is used. The focus is on simulating elevator trends in the near future using these learning data and then making assignments.

Therefore, as long as accurate learning data can be collected without being limited to the learning function part, an uninvented method will be described with reference to FIG. 17. At 21 in FIG. 17, the distributed standby control task 23 shown in FIG. 5 first searches for number a which is free and on standby.

Then, when a certain period of time (for example, 10 seconds) has passed after the free operation, and there are M new standby machines that are not controlled by the distributed standby command, it is determined whether the maximum number of distributed standby machines has been exceeded. Note that the distributed standby command is cleared by assignment or call. When the function for determining whether there is a new standby machine determines that there is no new standby machine, the process goes to 23A and returns.

Here, the number of waiting machines (TAIK1$N) is the total number of the new standby machine and the currently waiting machine, and the new standby machine is the 19th machine.
As shown in the figure, the corresponding machine number bit of NEW$TAIK1$C is on, and for Funaiso, which is currently on standby, the corresponding machine number bit of 'rAIKI$C is also on, as shown in FIG. The maximum number of distributed waiting machines is a preset number (η), and if it exceeds this number, the waiting floor is not determined.
It goes to 23A and is returned.

Here, it is assumed that the current maximum number of distributed standby machines = 3 machines, and that the new standby machine is machine A on the 4th floor, and that there is no machine on standby. In other words, N EW$'l'A] Kl $C= 01 (f()
Assume that 'I'AlK1$C=00 (H) indicating hexadecimal representation.

In this case, the process proceeds to the 17th-22nd waiting floor determination routine.

Figure 18 is for explaining the details of this waiting floor determination routine.From TAIK1$N, demand data for each floor (the above-mentioned learning data), HCT$RAT (average number of hall calls generated in 15 minutes)1, As shown in FIG. 18 and 24, the distributed zone division is divided into TAIK1$N.

In this example, there is one zone because TAIK1$N=1. Also, H CT $ RAT is shown in Figure 21 (a
Suppose there is a shift like l.

In this embodiment, the example shown above is used. Hotel 5
9th of the month, 1 tomb, Daian, holiday, 18:00 data is 1'R
Assume that this is an example where MOD=3.

Note that if there is a change in TRMOD, the distributed waiting floors will naturally be reviewed. In the routine that changes TRMOD, T
AIKISC is cleared. In the chiful processing of this example, since the index structure of the table is made up of H8 (hole sub-index), the 21st index (b
Rise to each floor (POS) as shown in l,
HCT$RAT1 (POS
).

Here, the number of zone divisions is 1 because TAIK1$N = 1
Zone (no division).

When dividing into multiple zones, HCT$R as shown in Figure 18 24
Divide the total ATL by the number of zones and calculate HCT$R from the lower floor.
Add AT1 and divide the zones by the value of [total/number of zones]. The details of this will be described later.

Next, the routine proceeds to a waiting floor determination routine in each of the 25 zones shown in FIG. As shown in FIG. 22, the point X where the demand is average is set as the waiting floor. Figure 22 shows the demand and unsatisfactoriness of each floor in stone, and it is calculated so that it is balanced at the point X on the left side.

However, it is set below HCT$RAT$MIN (assumed to be 5 here) because the possibility of a hall call occurring is not small. If X is not an integer, round it to the nearest whole number.

Next, an evaluation formula for determining the dispersion rank to find the ideal dispersion rank X will be shown.

Average number of hall calls made in 15 minutes = (8-x) x 9.5 + (3-x) x 13.5 + (1-
x) x18.5=OHCT$RAT1(POS)<HC
If T$RAT$MIN, HCT$RAT1 (PO
Assume that S ) = 0, and that PO8$MAX=8.

From the above formula, ΣPO8XHCT$RAT1(PO8')P08tl +3.2 Therefore, PO8X=POINT(x)=3However, POI
The rounded distribution floor posx of NT(x)=x is the third floor.

Therefore, as shown in FIG. 17, a distributed standby command to the third floor is set for the A machine, and at the same time, the command is reset. Then, the process goes to 23A and returns to the distributed standby routine. After that, assume that Unit B becomes free on the 6th floor. After 10 seconds, as shown in zl in Figure 17, by searching for the standby machine, NEW$TAIK1 $C= 0
2H TAIK1$C= 01)1.

Here, from TAK1$N=02H (assuming there are a maximum of three vehicles), the process proceeds to step 171N22, that is, the waiting floor determination routine of FIG. 18.

Here, from 'l'AlK1$N=2, the dispersion zone is divided into two zones as shown in FIG. 18, 24.

The division is performed by adding HCT$RAT1 (POS) according to the lower floor, and dividing it at the part shown in Fig. 24.

The division is made as evenly as possible, as shown in Figure 24 and Figure 27, and the last floor of that zone is designated as ZNE. This means that the zone has been divided.

This point is also one of the characteristics of the method of the present invention, and in the example, 1i HCT$RAT1(POS) = 22.8P
08 Self 1 The calculated value 28.5 is, from (pos), the last floor ZNE (1) of zone l
= 2 (floor). Then, as shown in FIG. 18 (25), that is, FIG. 23, the process proceeds to the next routine, the waiting floor determination routine for each zone.

The same procedure as before is carried out for each zone, and as shown in Fig. 25, for zone 1, HCT$RAT1 of 5 or more is only on the first floor, and the planned distribution floor PO8X (1) =
1 (floor).

For zone 2, HCT$RAT1 has the 8th floor and the 3rd floor is 5 or more, so 8x9.5 + 3x13.5.

X=□〒5.1 9.5 + 13.5 Therefore, the planned distribution floor PO8X(2) = 5 (floors).

The process shown in FIG. 18, 25, is thus completed, and the process proceeds to step 26.

At z6, the waiting floor of the distributed car is determined.

(C) during the zone as in the two examples shown here.
HCT of T$RAT$MIN (temporarily set with 5) or more
If $ RA T, 1 exists, its height ′a
Since there is a high possibility that hall calls will occur on important floors, it is effective to perform distributed waiting according to the demand, but
In a quiet zone where this is low and all HCT$RAT$MIN is less, there is less need to force people to wait at the scheduled distribution floor determined by the above method. For this reason, in FIG. 18, 26, this condition is checked for each zone, and when there is low demand, it is sufficient if there is a free car in the zone, and the free car is not directed to the waiting floor for fingers and toes.

Only in the case where a number of free cars are waiting in one zone, wait for the reserve distribution floor (diffusion method) In this example, since demand is high in all zones, 25 A is made to wait on the 1st floor and B is made to wait on the 5th floor in order to make the nearby cars wait.

Here, an example of a quiet state will be given, and the portion shown in FIG. 26 will be explained with reference to FIG. 26.

Unit A, 7th floor free standby Unit B, 5th floor F11-standby ZNE = 2 (zone tit
1st floor), and as shown in factor 27, zone 1 is PO8x (1)
= 1 (IF), MAX (HCT$PA in zone
T1 ) = 6.3 Here, the median value of the quiet zone is used during off-peak hours. On the other hand, zone 2 is PO8X (2) = 6
(6F), MAX (HCT$RAT1 in zone)
= 2.1.

TAIKI$PO8 (1) = 1 (IF) Zone 2 in Zone 1
Therefore, TAIKI$PO8(2) = OFFH.

However, 0FFH means that nothing is particularly necessary, and is used to mean nothing. At 28 in Figure 26, first go to the required zone of TAIKI$PO8 (zone).
Select the closest one with two or more standby machines within one zone, add your fingers, and set it in TAIKI$CAR (zone).

Therefore, TAIKI$CAR(1) =01H (unit B) then T
AIKI$PO8 (zone) is set to t in order not to wastefully move the free standby machine in the zone where it is no longer needed.

TAIKI$CAR (2) = OOH (machine A) Similarly, set the pos of that machine in TAIKISPO8 (zone).

TAIKI$PO8(2) = 07H and above, at the 180th 26 (Saw 71) rhrKr end cAno) = oxH(s
No.; Sho) TAIKI$PO8(1) =01H(I F
) (Zone 2) TAIKI$CAR(2) =0
0H (Unit A) TAIKI$PO8 (2) =07K
(7F) is specified and returns.

In FIG. 17, a command is set in the transmission buffer with each elevator. Also NEWtTAIKI $C= 1)OH(fx shi)T
AIKI $C=03H (machines A and B) is also set.

This concludes the explanation of this routine.

If the +@e state in this example changes, Takie's T RMo D changes and the learning data changes, so this routine is used again and the distributed standby is not performed. However, as I explained above, unnecessary movements are rare.

The embodiment described above is a distributed waiting method using the demand level based on hall call occurrence prediction data of learning data,
It has the following three characteristics. That is, (1) A shared distributed standby method for high demand floors.

(2) Distributed zone division method according to demand and number of Freener units.

(3) It can be said that this is a distributed standby method that emphasizes diffusion without wasted movement during off-peak hours.

S1 The present invention is not limited to the embodiments described above, but may make more effective use of demand forecasting and make variable the delay time (10 seconds in the embodiment) from free to distributed standby.

For example, if there are a small number of high-demand floors, it is effective to have free cars waiting near those floors as soon as possible.

Furthermore, the maximum distributed number of waiting machines may be made variable depending on whether the overall demand is high or low, and in particularly quiet areas, a small number of waiting machines may be sufficient.

〔Effect of the invention〕

As described above, according to the present invention, efficient distributed waiting can be performed by effectively utilizing the demand forecast for each floor in the learning data. If the number of waiting cars is small and there are many floors with predicted high demand, the waiting cars will be distributed to shared standby positions on those floors. In addition, when dividing the distribution zone according to the number of free machines, the demand forecast value of each floor is used, and the demand is divided so that the demand is approximately equal.
Since the machines are placed on standby in a distributed manner to the waiting floors according to the demand in each zone, the cars placed on standby in a distributed manner can respond effectively to the next hall call that is predicted to occur. When the height is low, the cloud is especially strong, and instead of dispersing and waiting, I have them stand by at that length as much as possible, and I try to spread out depending on the zone, so I can avoid unnecessary movements. There is.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of a control device for carrying out the elevator group management control method according to the present invention, and FIGS.
Figures 4, 8 to 10, 13 to 16, 1
Figures 9 to 21 (a). FIG. 6 is a block diagram of a building to which turning is applied; FIG. 7 is a diagram showing the 5e open mode of the elevator; FIG. 411, FIG. 12, and FIG. 17. Figures 18, 23, 24, and 26 are all jt%.
Figures 22, 825, and 27 are for explaining specific examples in which the same af-sugage method is applied, respectively. This is a diagram. 1°° Hall call registration circuit, 2A to 2D... Elevator operation control device, 38 to 3D... Car status buffer, 4A to 4'D... Car call registration circuit, 5... Small computer , 6, 28-7D, HA-8D...input register, 9-HCT, 1O-CCT, 11 ・KCT. Applicant's agent Patent attorney Takehiko Suzue @7 Figure 8 Figure 1) CT$SET in Figure 1O (X, V) Figure 1f Figure 2 Figure 13 Figure 11 C2S DATA (X, V) Figure 15: ``Yuta ◆ sword on the top. Fig. 16 Fig. 17 Fig. 18 Fig. 21 (a) (b) Fig. 24 Fig. 25 Fig. 26 Fig. 27 Procedural amendments dated May 9, 1973 Kazuo Wakasugi, Commissioner of the Patent Office, 1. Indication of the case Japanese Patent Application No. 5CJ-63660 No. 2 Name of the invention Elevator group management control method 3 Person making the amendment Relationship to the case Patent applicant αm Toshiba Corporation 4, Agent 7, Description of amendments engraving (no changes in content)

Claims (2)

[Claims] (1) In an elevator group management control method in which multiple elevators are put into service for multiple service floors and these elevators are collectively controlled, the occurrence status of hall calls from each hall is calculated on a monthly basis. By 2 hours, by day of the week,
It has a learning function that memorizes major factors that cause changes in traffic volume, such as by Rokuyo (Rokuki), and when issuing a standby command to the cars scheduled for distributed standby, the above learning function is used to calculate the demand on each floor. 1. A group management and control method for elevators, which comprises dividing the waiting area into zones and determining the waiting floor by a predetermined calculation using prediction data. (2) Determination of waiting floors and waiting zones ζ Calculate the predicted demand value for each floor using an evaluation formula consisting of the number of hall calls occurring in each direction (based on learning data) and the position of the distributed standby scheduled machine, and An elevator group management control method according to claim (1), characterized in that positions where the predicted values are equal are set as distributed waiting floors.