JPS5922870A - Detector for characteristic of transport demand 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]

[Object of the Invention] The present invention d1, an elevator that is controlled according to traffic demand.
In particular, the present invention relates to -1-, an apparatus for detecting a percentage sign of traffic demand; (Prior art) Recently h? Microcomputer (hereinafter referred to as microcomputer)
r) has been applied to various industries, and in the field of Nirebeke, it has been applied to group management control devices that efficiently manage multiple elevators and to machine number control devices that control individual elevator fees. There is. These efforts are based on the small size, high functionality, high reliability, and low cost of microcontrollers.
Because of this, it has made a great contribution to elevator control equipment. For example, in the case of group management control, pole calls (f
Monitor the whole hall online on ll7it!
The most suitable elevator, taking into account the service situation of each individual.
This makes it possible to select and allocate the information, which greatly reduces waiting time. In addition, it is now possible to provide priority services, such as installing multiple elevators in halls with many passengers, and providing e-services with short waiting times for executive floors. It has become possible to perform 1 kana control 1. On the other hand, the elevator monitoring equipment utilizes computers in a fairly advanced manner, and connects the system processor that controls group management with the central monitoring station's telephone line to ensure efficient monitoring. A method has also been proposed. That is, in this method, during the night when the elevator system does not need to be operated, the system processor is separated from the elevator system, and the central 1/7
It is connected to a device that simulates the car response in the processor of the JF viewing station, and efficiently monitors the functions and operating status of the system and processor. (Special Publication No. 56-37145) As mentioned above, by using a computer such as a single microcomputer,
We are confident that this will significantly improve performance and functionality compared to random logic configurations. However, in the conventional elevator group management control device, the operation is controlled using predetermined and fixed control functions and parameters, so the system is not necessarily adapted to the ever-changing building environment. For example, the destination intersection ilO, I'lδ will be different depending on the traffic demand when a building is completed and the traffic demand when there is a subsequent tenant change, business change, etc. Also, among the hardships of -day's father, dispatch, lunch, leaving work,
The usual destination has changed dramatically. In this way, when traffic demand changes significantly, efficient management 11tll? iii becomes difficult, leading to a decrease in hissing. Conventionally, methods proposed in Japanese Patent Publication No. 48-1.5502 and Japanese Unexamined Patent Application Publication No. 141942-1982, etc., have been proposed to reduce traffic demand. All detected and the predetermined 'FηC1
1 (determine which of the multiple representative points it is closest to)
' By selecting the current pattern, all methods of controlling 10 elevators 1i11 were adopted. However, this method has the problem of not being able to adapt to situations such as when traffic demand in a building is not under control, or when a change in the building's environment occurs. In addition, since there are many buildings where traffic demand generally changes significantly even on a single day, a method has been proposed in which the elevator type is controlled by switching the elevator type depending on the day of the week (Japanese Patent Application Laid-Open No. 1983-1999).
130457). However, there are changes due to holidays, 101st edition, etc.
It was necessary to respond to the traffic 71N required each time. Furthermore, almost the same traffic demand and percentage returns may occur irregularly at schools, halls, etc.
It was impossible for K to understand all these requirements and to drive easily in this traffic. [Object of the Invention] The object of the present invention is to develop an elevator that can automatically detect the salient features of the elevator demand in a building with thousands of elevators installed. −
An object of the present invention is to provide a traffic demand characteristic detection device 1K. [Features of the Invention] A feature of the present invention is that it is equipped with an elevator traffic demand detection means, and generates a preset fc, a special mode of traffic demand, or a characteristic mode extracted by itself in accordance with the detected traffic demand. 1) automatically detects the essential characteristics of the traffic in the building where the elevator is installed; (,-Actual Mii example) Below, I will explain one example of this book. (i) (I'll explain. First, elevator-crossing) 1m demand! In 1, the characteristic mode generation equipment will be explained in detail with reference to a specific embodiment shown in FIGS. 9 to 13. In addition, prior to explaining the embodiments, the limitations of the present invention will be explained.
11(l, lj' The transportation demand situation from 8:00 a.m. to 2,116 p.m. at El, where the elevator was installed, is shown in the diagram as an element representing the IP!I' translation of traffic demand. The figure is omitted, and the morning dispatch time and h
It shows the traffic demand situation during the 1st and 11th hours. In the figure, the curve CI represents the total pressure of the traffic volume, and the time chart)
9 L, l~l-+ 11) Is Yl by floor? , )A quasi-concentration degree is shown as a digital value consisting of 311α. The upper part of Figure 7 shows the occurrence in the same building during the day! The magnitude of the transport demand characteristic with lrt of 1 is expressed by a curve φ1
1. (Details will be explained later) In general, in order to optimally control elevators, especially when multiple elevators installed in parallel are group-controlled, it is necessary to Ideally, it would be possible to predict in advance which direction they will go from floor K, but in practice Hayabusa, which requires thousands of customers to register their plans for elevator use, is a practical problem. This is difficult for AI6, and it is impossible for buildings with many visitors or buildings with many events. Therefore, the present invention extracts the characteristic mode of the transit demand of the building as the most effective means to solve the above problem by learning past traffic demand and controlling to predict today's traffic demand. By learning the extracted feature mode 1r, the size of multiple elements that can be properly expressed, and elements such as occurrence time and occurrence synchronization across multiple branches, it is possible to perform optimal control by prediction in elevator control from tomorrow. Judgment 1 ~ This principle is adopted. Traffic demand includes various factors, but the ones that put a big burden on the electric heater at 1 o'clock are the traffic volume and the movement status between each floor. Traffic ij f: Indication 1 function C

The number of people remaining in the elevator per 5 minutes in [5 minutes] has been defined using the above equation (1). However, the number of people remaining
J: 11', which can be seen as a standard number, and will be treated as a mere constant. This traffic 1Ii- is used to compare the magnitude of traffic demand between different elevators and shared residences installed in different buildings, and a rough comparison can be made. In other words, normally the traffic rate is around 4 to 6%, and when traffic n( is 12%), it can be roughly expressed as an extremely congested situation.
There is. However, it is insufficient in expressing the magnitude of stress on the elevator. First, are the elevators installed in line with the number of people remaining? Also, from the elevator's perspective, there is no need to compare it with other buildings. Therefore, the authors set the unit of traffic m function C([) as 5
Elevator per minute - expressed in terms of number of users and l~
Ta. Next, consider the amount of traffic during a given period (at least the amount of traffic per day).
H. If possible, use a traffic learning function with a time constant of 7 days or more to measure the average amount of correspondence (converted to a day) by n1.
A frequency distribution map is created, and traffic M: tff value level, α weight, α2. α3 was created and this level value was created to obtain an evaluation value (traffic volume level function) CV(t)e indicating the magnitude of the prime value of the traffic volume C of each traffic demand. Even if the traffic volume is the same, if all floors are evenly crowded, or if people are yelling to get on and off only on the 1st and 2nd floors, the stress caused by elevators will be The difference is more than double. Therefore, there is a mind that recognizes this as an element in some form and creates it. However, since there is no established theory that defines the characteristics of the movement situation between floors (hereinafter abbreviated as traffic flow), the authors used the congestion concentration floor relationship 1 (t) and the congestion concentration function V (1,000 floors for all floors). +
) to express it. In addition to this, it is also possible to calculate and express the distribution coefficient for each single pressure floor, but if you simply use the distribution situation as a coefficient, it will not be possible to correspond to each individual floor. We decided to include the floor itself as an element that expresses the specific traffic demand for elevators. In addition, this congestion concentration floor relationship [11 (an example is shown in equation (2) below) and 11 are also expressed by the crowd moderation function V (LH in equation (3) below, f911 'fC - t) &:j :Due to the nature of the elevator regulations, the number of passengers N (symbol S) and the number of passengers disembarking (symbol 1)
．． 1. ), it is better to calculate the upper row direction and the downward direction (symbol 1) by dividing into "C" to more accurately express the timing of various traffic flows associated with various traffic demands. The following eight R″F filli values can be considered as element functions representing the factors of these 4+;Ii. n in the time period [-Δ1.1+Δt] Evaluation value indicating the factors of the downhill personnel distribution situation. in (su, Vnr, (t). (3) Factors of the upward direction personnel distribution situation 111U(11,,Vn,(t)) (4)Evaluation value indicating the factor of the descending direction of downbound car personnel distribution element.In(t),V(1) 1)D The variable +1 ranges from 1 to the maximum number of floors (generally the maximum floor, several tons, 5 to 40 floors, especially around the 13th floor), so the 10th floor is the 10th floor. Total H180 for elevators with floors
It is necessary to obtain the evaluation value of the pet rat, record it all, and carry out identification procedures. Therefore, the authors
We decided to simplify it to the extent that it can be used practically in order to express all the characteristics of traffic flow. First, regarding the direction, let's summarize this in 7 A]
! All the floors on the same Yervis floor J1 phrase II
We decided to conduct an analysis assuming that the ICs are on separate floors. In other words, the number of congestion detection intermediate floors i to be detected is 1. The same effect can be achieved by setting data for changing floors in each direction (t) and C. Also, the size of array variable 11 is elevator -1l.
ll I Revised from the stress exerted on the river 'C Light load direction (submerged floor (hereinafter abbreviated as floor; floors with no directionality in 11 are abbreviated as floor)) has little xinxing.In other words, if several floors with a negative weight of −11 are considered as elements, to
Therefore, for convenience of explanation, consider 1.The value of the array variable +1 is 1.2゜3.
shall be. As a result, the 80 evaluation values described above can be reduced to 12. That is, the evaluation field value of each round can be calculated using the following equations (4) to (9). (The inth car concentration floor in the time period [-Δ1, 1+Δ[]...(A) W:'(
t) = time period 1 - Δ1, 1 - t Δt]... (5) From equations (4) and (5), it is determined that passengers are on the busiest floor l1l (Rino). Mix?<A degree V n (') k''2F M+3. V
, (L) =-=time□ii) (t-old, t-1-
Persons disembarking from the first floor at Δt]...
...(8) (7) From formula (8), the congestion degree V of the +1st exiting customer concentration floor In(ri is as follows. trl, F is the number of service floors of the elevator (here, (Although the number of service floors for each direction is used, it is also possible to use the number of service floors for each direction.) The authors finally show the floor concentration function V to further simplify In place of the equations (6) and (9), the following equations (li+) and (11) are adopted, and the 1f-F [+lIi value of 8 saris is expressed as traffic?111.011 points i , ': J element. In the improvement formula, n represents the concentration level (φ) for each floor. Typical traffic lights at the time of dispatch)
The concentration level @ for each floor is illustrated, and the curve f U P I
Nt? , J, the ratio of passengers in the car on the 'U L' direction side is 111111', Ii! f I) N I
N + rJ, I) Hayabusa ratio of the number of passengers in the 1-inch direction'
4"<', j・- (J, each indicates the distribution situation of each floor, etc.).Song, gland f IJ i' OrJ
``1.' + and the curve f])Nl)U'J-'!
(・"t - 几一 slant upward F1 force direction and downward direction (1111
3 to the power of 8)
The sum of the ratio of passengers entering and exiting the car is 100%, and from a macro perspective, the number of IJ people in January is equal to the number of people shown in the traffic report. be equal. PQ 3 Iy, I (l is a graphical representation of the ratio of boarding and exiting cars to I2 and medium-high floor llj ■ 田人i ratios, and the curve f VIN &, j, the curve f V for the number of passengers in the east car)
I) U T indicates the degree of distribution and distribution of the number of people leaving the car to the main floor. The traffic demand in 12~'3) can be decomposed into the following elements by recognizing its characteristics. (1) The traffic density function C(t) is 120 people and 15 minutes in C(tZ~'tA)... ...
(121C (β2 people - Lg) 200 people 15 minutes --
---β13) (2) Floor concentration function V of passengers and personnel
8(t) is 60'-1-30' in VF'(t!~IP)
+4500 ...04) (3) Floor concentration function of people leaving the car VR(t) is VR('t - ts)
"17"-1-162-l-14'+13'+12'
-+-11'-1-7'-1-4'+154.5 ・
・・・・・・(15)(4) Power cage lumped rank function 1n(t
) is In(t, ~l5)=$01. $02. $00...
・(Country (5) Descending basket intensive rank function I: (t) September 1=1
1 straight bow with a distribution of ~3 (t2~t, l = $04. $07. $06 ・
...07), the floors of β1 and 41~ are as shown in the lumped rank function l (Kiritsu vl &, 1:1: Since it cannot be called a concentrated floor, the order is 110th floor In. Traffic 1i, which is a characteristic of traffic demand at the time of dispatch
't is close to the maximum in the previous 6 predetermined periods (such as 1 day), and on a specific floor (lobby) (In-, t=$
02) and 61 are the majority of the money, and (VS
(1) large as -4500) and i) N direction shift 1
iiiIJ staff is extremely small (1n (severely $8 to $81
The II'η characteristic that 3 does not exist can be practically completely expressed by eight functions representing the values of the five elements mentioned above. Next, there are various types of traffic that occur depending on the time of day and day of the week. We will explain how to recognize y hit YC and extract it as +yt< significant one-time mode in that building. First, the overall control principle of an elevator will be explained using the procedure shown in FIG. In addition, Figure 4 C shows the program
It does not illustrate the operation flow of the node circuit or the node circuit.
It provides conceptual step-by-step instructions that explain the entire learning process. 1zu Building opening day (111tT NN OP 10
) Traffic 919 requirements are scheduled in advance from the Intelligent IM terminal, etc.;
Input from a bow Y board, etc. The points to be input at this time are the schedule time (DEPLE N0T) as shown in Figure 14.
291) and floor 11, which is expected to be crowded due to boarding and alighting.
No table on 1st floor! Enter '292, this 1.
As a result, at least the middle floor outside congestion 1 has been separated, so the evaluation values for each element of the above-mentioned 5 4 ft F, t 8 tP, and TE are created as follows. (11C(AMI+1-AMI(1,30)=135
...(18) E'Z' jTl No song specified, corresponds to intermediate level α4] 35 [15 minutes per person]
Set. (21IR(/MVIIO-AMlo, 301=$82
．． $05゜$85 Scream...(1!
j(3) Ill(AMIO-AMIO,30)=$f
32. $05゜$85...
...9I Since there were no instructions for getting on or getting off, I entered the same floor for both.In another case, when only 1 bitter was manually operated, I entered both directions. do. fno II V 8 (AMIO-AMIo, 30)
=1700 =-=I71) f! i) V”l
ArX411)-AJilo, 30)-1700-Q
'JHere,' there is no constant 1)1 of the %VC concentration, and the three i, lt: :I・(1st floor f: Since it is specified, the total value corresponding to the concentration fW level β2 is Selection 1700
And so. Since the lr mode shown in Figure 1, such as when dispatched or during lunch, is generally likely to occur, it may be possible to store it in the factory exit iii, sometimes it C) M, etc. This can be done only in buildings where you want to achieve optimum control from the first day. Next, elevator-fttll 'l' for Km! j Automatic setting of enemy mode (procedure P 30 ) will be explained. The detailed procedure of this part is shown in Figure 5. For example, the procedure for detecting traffic demand for one day or one week3) P31
), -J2 written 5fiJt (D feature element, zunaji traffic flj C (rito, 't' of traffic flow! i-dispersed element function all element functions all operations,) 32 ), record (f
At this time, for example, the traffic demand is calculated every 7.5 minutes (according to Tsunoda, 192 groups per day, 13 per week)
44 rivers, and approximately 10 i(,H) of non-volatile memory is required to record the feature element function values of the 8 animals/group mentioned above. This requires a high-speed performance mode.Furthermore, it is necessary to express the characteristics when the number of users is small.
It's impossible. Therefore, as shown in Fig. 1, when the traffic volume is small, good synchronization is used, and when the traffic exit 1- is large, it is detected with a short fixed time 1..., then the timer 11 to record for -7. The number decreases. For example, if the characteristics are detected when there are 11 Tl of predetermined personnel exchanges, there will be about 48 pairs per day. However, for a predetermined period of time, for example 3
If more than () minutes have passed, the feature determination is performed at the time of -L'. [i! i1 When this method is used, time data indicating the time width must also be recorded in pairs with the characteristic data. All the above procedures are repeated for a predetermined period of time, for example, one day, and a portion of the traffic demand consisting of tens or hundreds of sets is analyzed and recorded. Next, set the new IP detection mode and check whether the IP address is correct or not.
Find the 'r\r modified mode extraction function for '1' drawing. Ma?゛First, - lp of the 8th century already calculated - key rice function i
: , f' + Raisari (1 shift r'jH A 1 turn. First, find the traffic shown in Figure 2 (d level α6 to αI4- (hand 111 + 'f P 35). 11 Next, calculate this traffic volume level. Create and convert each set of traffic volume function (j(su?i) into 1i level gate CVC, and above, the traffic demand for 111 minutes is calculated by the 8 elements 1jl, 1x) consisting of multiple u911. You can create an array (hand 11il'j P 36
). 81) In Figure 2, the actual Mli L fc traffic I4 level function CV (+) (2) number of values is 7 ($06 ~ $00)
However, it is also possible to reduce the number to 4. Create a frequency distribution of the values of the centrality function V (t) of Gyukaku and floor, and [■2 (Curve of 1/V+i-, create, level β1
~ Provide a distribution level consisting of β4 or L (hand 111't
P37). This 1. Rougher: To floor 1 Song f4 "111↓" function ■
V(t) and restrict fJfi 'lJ'1 of each 'rk'O'!i element combination to some extent. Then, in order to extract all the representative 'ljN? Create an array expressing the detected traffic demand by a coarse function ( ) ) 37 ).Furthermore, lr!? characteristic mode extraction function PS m'C coarse /1'! Find each +-fine mode (Step 1.1
38). Il+ is the number assigned to the timing mode that is recognized as the same. For example, it can be determined using the following α1 formula. T'5h)='L'(In) (k I(CV(I
)) -1-k t (VV(,,l) -1-VV(
m) l l-Q, 'j Note that the feature mode m l should be marked in the order in which it was detected. The function '1'm is the number of times or accumulated time that the same feature mode as the feature mode I11 or a mode similar to ld is detected. In the event that the number of 6-loss modes exceeds a predetermined number, the characteristic mode in which '[m is small and was drawn earliest] is included in the characteristic modes that are close to other races. Also, when recognizing the characteristic mode of times? The determination may be made based on whether or not the i-th Hayabusa middle-order function J, and Lumped rank function ■Vn is $0
For example, it may be determined that IVl to lv and l all match (step J137). Multiple feature mode extraction functions PS n+ obtained above
-f, phase! Compared to L, among the new feature modes h=I
A Select one group or all the upper groups of X and create a new elevator system? 1ill Ml Temporarily register as a characteristic mode in 'l' (f order P39). At this time, if there are no or few feature modes set in advance, we will extract more, or we will always extract the maximum number of feature modes that can be set for production 1, and we will extract 11 in a faster number. ￥
Make sure to complete the setting of 敞mode. Meat, provisional S, the floor of the characteristic mode to be recorded medium level function VV (t) and traffic ti level function CV (s), not l, but the original function The value of is registered.As a result, the subsequent identification becomes iE I+' (Ii.) From the above, the mode of the traffic characteristics that existed in the traffic demand between that elevator. However, the present invention is not limited to this method, and several dozen feature modes for prediction are set in advance, and the method described below has an advantage in that it is possible to extract a large number of feature modes. It is also possible to configure a configuration in which the feature mode f that occurs frequently is selected and set separately as a feature to be used for elevator control. In this way, the newly extracted temporary registered tr, "i Collection mode and already registered cows +j
Feature mode priority function P, which calculates the priority of fine modes,
1. ' 0 and the feature mode generation case (procedure I) 40) in which the feature mode is completely generated by modifying the registered feature shell prime function value of the registered feature mode - [This will be explained using FIG. 6. Note that the registered feature mode element function includes the already-described feature mode element function CVm of the detected traffic demand. In addition to im and m, a periodic function 'rT'1m for learning all elements that can be repeated periodically and a time function Ti1t for learning elements that occur at a regular time on a daily basis are added. First, for each element function of the Ni set of feature modes created in @IK procedure I) 33, Ml
Registered set/ii (Learn which of the salient points P'rn is closest to which has a multidimensional vector expressed by a mode element function. Furthermore, the weighting between each element is a constant k! ~ 1 (performed by 5. Detected traffic?, ', 4 The salient point [1n vector created by the vector by the feature mode element function of V3! is the vector. The salient point F?n can also be described in the same way, and the distance between the two points is scalar 1'
rlPm,l(lj:l,!((move 1)
1L'tP 4 1), mixed? For the explanation of the principle, only the first order of the +'fl Rin middle order is evaluated here, and the formula is shown in a simplified manner, but in reality, iJ 2 ,
It is better to add the third concentrated floor to the K/r value by lightening the way If. Also, the floor number difference (difference value = In-I, T,)
It is also possible to change equation (A) so that fl, "j" is "0" when there is a match, and "1" is -1-i when there is a mismatch. Registered IV with the closest suffocation Pm!
jIa mode number... is determined and recorded as the closest registered feature mode (Step 1) 42). M I N (J,”, ,+, P,,!,・
・■゛1. u+) ・・・・・・・・・46 or more steps fK: I+ = 1 to II: N, Twist to 1: Execute the characteristic element function of the traffic demand detected one after another and record it with 71 (Step 1.'42). Next, for example, find the integrated value of the number of times and time selected for each registered 7i potato team model snowflake m) and use this as the wr value function φ′1゛m for all registered feature modes (Procedure P441゜this). : fF filli L, depending on the frequency of occurrence, but it is better to use the sum of the products of the vectors of Traffic 717. The following K /I' mode extraction priority function φPm All registrations from the equation Find the completed feature mode. φPm = (1 kg) φPm + 6 x φ'I'm ・-
-H Remove the plurality of feature modes at the lowest level of this value (Step 11 [j P 45 ). Therefore, the four stomachs i recorded earlier (1 because it cannot take a larger value than modeshi); i: easy to be removed. The mode function Pm (I
Learn and set all the functions of each element of l, 'il 叡 12) (Step 111''46) a Example 1.1 Traffic function C,, j +; 1 traffic 1\l two-level function (: ■In is the element value of the feature mode that is extracted in the current predetermined period (one day or one week) /hi%R mode I11, or is the element value of the feature mode Il +
Traffic determined by i'! ; Element values calculated from Kaname's Kazume average and registered special 6! With the element value of the mode), we performed the same learning (Cho 11 JI no Zuichi Sakujo H1!) 1 practice as the formula to generate the function value of Traffic... Regarding floors):
1. ji↓2. For the third rank function 1, the 13-event is also selected in descending order of frequency of occurrence. Also, the time-related functions II + 13 m and T 11 m
Similarly to the (to) formula, find the exponent below/by no means (Step P47)
・In addition, for synchronization, manage all the individual time from the previous detection until the next detection, and learn the characteristics that are repeated in two different synchronizations by learning about multiple synchronizations that occur frequently. I can do it. Further, by individually learning a plurality of times that occur frequently in the first eye and recording them as a time function 1'll(+n), more accurate predictive control becomes possible. Based on the feature modes (hereinafter simply referred to as features) that have been extracted by learning and setting the element functions, the 11-character traffic information is recorded according to step t' 50 in Figure 4. Based on that data, you can learn the optimal control parameters for each Pj7R mode, check the capacity of the actual machine if it is a computer, or calculate the capacity of the actual machine or elevator.
Or record the selection results of the optimal program in a simulation using a large-scale computer at the building monitoring computer or central maintenance center (f order 1' 60
). Next, the characteristic modes of the current traffic demand are explained in steps ■) 31 and 13.
32, and feature pattern recognition learning is performed using a method similar to steps P41 and 1]42. However, here I will capture some examples of improvements that are a little negative. In other words, the previous battle did not include the time element, but in the actual 121 (turn) elevator - tlrlJ diameter 1 series, especially the ?11' M Iult fltll mll elevator - the control arco. IJ the tIll parameter
The delay from J change 5th until it actually functions is thought to be several minutes, as the average period between elevator laps 113 and 113 is 120 seconds, and it generally stabilizes after 10 minutes.) There is Eri. So % '11r enemy recognition 1ifil! '! It is suitable to add temporal continuity as an evaluation element to the formula (c) or formula 30. For example, in (c) 2, the scalar of feature 1 that was selected last time and is currently in upper 1/heter control... Only the term P(su, +n) k (P(t),,n) l
If c 11 is taken as a value smaller than the coefficient kg't71, the previously selected and current elevator control special mode can be easily recognized. Second
Furthermore, it is also effective to detect features that are repeated at the same time every day as early as possible using similar means. For example, as shown in Figure 1, when the dispatch time is flat (if the dispatch peak occurs at 8:15 every day), extract it as P(In), which is one of the characteristics, and calculate it as the time element function 1'l1m. L-C0
If 8o 15 is recorded, for example,
It is the characteristic term of the pig.ItJ'f;C12 Switch to the formula to identify the characteristic mode, t-', I', 1. rnrn l <1(1(71"
'...I) (Step P 70) I jumped early by the past day (Chizuki l own 1) 50 and P2O)
Adapted to dispatch based on output amount L1 data [- can be operated in a negative mode. For example, if k7 is 15 minutes, then if the current time t is within ±15 minutes of the time indicated by the value o815 of the return function 1" Ll m indicating the predicted time learned up to the past day, it will be 11"! j
The value of the term I′(t),” is a scalar sleeve 1)(t)
, becomes smaller than m and when the times match, l −k ,
C) It becomes easier to be selected. It! identified by the feature mode identification (procedure L' 70) according to ``2'' below. J characteristic mode full analysis, if the learned one-time moment Pm is approximately approximated by formula (2), formula 040, or formula (c) (is smaller than the predetermined value), then
According to the restriction 1 data (data created in steps P50 and 1160) determined by 47 l'rn, the vehicle rotation Hill is 11. Due to the change in the layout of the building, the building's layout has changed, and the opening of the building has changed.
A station whose directivity exceeds a predetermined value not only has the first characteristic Pm, but also relatively nearby multiple characteristic ri! ! ! We can identify all the features and use the system (1) by interpolating the r data,
(Way in response to the approaching IW traffic report) (':J to O11
Execute all steps J' 60 and control the elevator by using the obtained (step 1.
” 75). (If there is a sushi event that will only take place on that day, please enter the time and the contents of row 11 in advance by pressing the switch or key indicating the floor and control type.
l> Input from the Y board, etc., and immediately j1) (= what control 11
1. Determine whether to perform 1 f (allocation of 2 priority services, door opening tights/extension, non-stop, code registration permission), and if it is determined that the specified time period has entered, priority will be given to the driving style and driving situation determined by step 1) 75. Then, at least part of the payment and event reservation details are given and the driver is ordered to drive (step 80). Through these procedures, the operating method (algorithm) that is predicted to always provide optimal control and its control 1 (1 constant (
Control the elevator operation using parameters (parameters).
P90). In addition, schedules for changes in transportation demand and costs due to changes in building layouts and exhibitions that will be held for one month (11i) are set on the KIUY board, etc., and the set contents are manually transferred to the IFT. I:' 2 o MR, press the same as explained above and temporarily set it as a new feature mode,
Speed up your learning (Step 1) 95). We believe that you have understood the procedure for raw material l of the present invention shown in Figure 1a4. Below, we will explain the overall procedure.
enact. (]) 1000 After step 5 is completed, generally shij, root P9
Returning to the feature mode setting from 5-21C, in buildings where adaptive learning control cannot be implemented due to factors other than aiming for the maximum elevator mtl-i, such as usability and morale, set route J:' 95-1. It is also possible, for example, that only when instructed by the scale setting device)
) It is better to proceed to 95-2. There's always a better way! Proceed to rate P 95-2, but restrict active operations in step P40 and step ■) 60, and proceed to step 1) 80 and &J: I' 90 device all indeligients jl! IA terminal, Crt T
Displays GakuR results in a format that is easy for humans to judge,
Eleven Bake Tube]) 11 people confirmed what they learned! i
In addition, active learning control such as registration is executed. (2) Figure 4 shows the learning control procedure, and the actual operation is as follows. For example, collection and detection of traffic demand included in step I) 30 and step P 7 (+ feature mode identification and step 1
1 to 'Y 117 s and step 90 elevator-driving side? 1°1111 yo parallel 1) [Executed. ! However, the operation control of the elevator is generally always at jull +1. ! I was asked to operate the system in parallel with the other procedures, and I set the system to operate in parallel with the other procedures (1
It is necessary to create For example, computer-controlled 1
In the case of step 1', which requires a large amount of processing 1c such as 40, 1) 60, and J]75, priority is given to the elevator-t control blog by assigning a lower task l/c to the task in step P90. This can be realized by executing the learning control during execution 1-1 during the free time. In addition, as another method, the learning control section can be used as a separate computer to perform parallel processing.・Next, the effects of the table in which all the principles of the present invention are implemented will be explained with reference to FIGS. 7 and 8. Figure 7 shows one of the buildings with many visitors during H, such as the Broadcasting Center.
I'm a visitor in case I = J traffic demand change song #i! C! Illustrated. The learning process when there is no schedule reservation on 1'+11 is shown in 111th, 21-1, and 3l1.
The following was shown. Symbols 1) J,, W are the time functions TJ1. shows the size of
The first point is zero, and it gradually expands, as l)
r defeat is identified time period [) n is visitor traffic M 91
It met, and a solid 1-so ζ was ejected. In addition, the symbol PL Igu shows the detection status of the feature element function for extracting and setting all modes (71-,) of the visitor. Showing the movement status of the floors when visitors to the 81st ν IK are included. ']'1
8 and fJIN [J'J'l 8yjiyp, I, 'Ca
-b, I-I L K Match fz special defeat element function is learned. Finally, a first specific embodiment of the present invention will be explained using the circuit diagrams shown in FIGS. 9 to 11 and the recorded data shown in FIGS. 12 and 13. The elevator operation control system 110 (executes step 90) is composed of the pole call registration 110 to the elevator drive device 115, and each of these control IIR++ block circuits 111 to 117ff: The devices that realize it are as follows. It may be a known technique (for example, Japanese Patent Application No. 851-57927 shows a device for detecting the number of passengers getting on and off from an in-car type @ detection device). For example, a door opening/closing system? 1'l device 11
4 includes the open, fj button, light, and operation signal of the device. 9
) K is detected. As a result of being input to the traffic demand detection circuit 130 and learned - [elevator which is finally fed back - example form of operation control! 2! Selection circuit 1
The only difference is that it has means for switching the (III control algorithm and control constants) by the signal from 7o (procedure P75). Demand 1.) 130 is a traffic demand special mode identification circuit 150 (details shown in FIG. 6).
1 Old report learning circuit I for each feature according to the identified features (50 is a traffic demand record 1 that records data for each feature
i51 Mfl, 1) 1 (i 1 A n and service status (elm operating number, pole call duration, door open time, indoor noise level, n's power, prank, car refusal, 'rlr, source 1・l) 'j King, temperature, etc.) Recording circuit diagram 6
1 B n and Fl to record learning results such as detected time and same JV, d1,? Intermediate MF MI2 recording circuit l〕1
Accumulation and learning of data collected by 61Cn (hand f
li1P 50) 'fc. Is there a time signal light for other circuits? There is a control circuit 140 which is used to control the operation of each circuit. Also relevant are procedures I) 20 and P'80 and P95 control). There is a reservation/setting circuit 190, and necessary data is recorded in a control mode recording circuit 192, a time Ne recording circuit 191, and a flifJ target registration circuit 193 according to input contents. The energy saving level command circuit 181, the service level command circuit 182, and at! 71.7 There is a target value setting circuit 180 consisting of a Xt command circuit 183 and a human power device 184 which serves as an input means for these, and the input is mainly input to the collateral type ts;J+1 selection circuit 170, and to the elevator-1liIJ bait 1 It is reflected. From the 101 yen 1 and the 11th VC, the Ministry of Transport, which is particularly relevant to the present invention, will explain the essential one-time microextraction. The input signal lines 111 to L117 of the elevator 11i11'i unloading yarn 110 etc. are used to monitor data regarding elevator operation and traffic demand, and the circuit 1) 131 collects data 1 (does this. The start time at this time is Gold Circuit J) Record at 1311'. The traffic demand at the current level is calculated by the circuit 133 every actual traffic interval (several minutes) using the traffic 1 blue report old measurement circuit 131. In other words, the data of the circuit 131, the time recorded in the entire circuit fJ1311', and the I
l, 'p Jll c', -v4% f Find h
i+i with H1 measurement transit time! It can be found by This (Il-shaped 1(〆l '?l-, #, Q, using the arithmetic circuit 134 having a constant of about 11,9, calculates the traffic demand for arc time I1) from the signal line 1) from the 134 button. Output. This signal is mainly used for one-time fine recognition by the circuit 157 (in step 11701, it is used in conjunction with f to stabilize feature identification. The circuit 132 is old data). 1310 Operates based on the size and time elapsed trap, FIG. The output of the traffic demand element value calculation request circuit 152 shown in Figure A is activated on January 52, circuit J) 131 and 1) 131 'r (7) Holds the contents and current time. . The data in circuit D 1.31 is cleared, the current time clock is reset to circuit D131T, and the next traffic demand detection (including other information/sampling) is completed. do. The signal D132 containing the traffic demand detected in this manner becomes input data to the intersection 1ffi demand one feature extraction (hand 1111' [P 30 ) section made up of circuits 151 to 156 in FIG. 10. First, the circuit 151 outputs the signal 1,) 1 a little later than the signal 1) 152 every time a new traffic demand is detected.
52-2, the function of the traffic demand characteristic (a element) is calculated, and the deviation is recorded by the circuit 153 (procedure) '32.1
133). Next, create traffic m and level I to create the evaluation function (
Step 1) 35) Find a function for evaluating the control and the size of the element (Step P 36 ) Executed by the entire control circuit 154. This control capacity 1 is the current traffic m (data No. 14 No. 10
34) is low and a predetermined period of time has elapsed, for example, or within the last few hours. At this time, the curve C2' shown in FIG.
Ill? +l'llf A circuit 159 was provided for the purpose of increasing the speed. Also, transportation? Distribution circuit 153VC every time I7 is detected
kl, data as shown in FIG. 12 is recorded. In other words, the recorded data per time of No. 8 [1] 153 n&
;i: 13 flIIit is shown, which is sequentially recorded by a strip-shaped or functionally annular recording circuit. Furthermore, the layer discriminant function ', L'' which was not mentioned in the principle section
Capture for M + n and i' M 2 n . Even for the same transportation demand, its characteristics change depending on the customer group and time of day. For example, in the morning, the robot moves quickly, but in the evening, it moves slowly, and you may miss the elevator that is the hardest. This situation can be expressed as an environmental function T M 1 m
shall be. In addition, the 1L rates such as those with many children and light weights, those with many mischievous calls, and those with tun chairs and VL calls are treated as customer group functions 'J' M 2 and 17-C. Konoshira's passenger)? 4 discriminant functions also become part of the stress on the elevator IJill size, so these are used as /lf characteristic elements.
This is a good addition. Features are extracted from the data recorded as described above (circuit 155), and the extracted and learned % characteristics are recorded data i) 156 m f as one set shown in the rglZ diagram.
The table with the number of feature modes MN (here 1l118) is the feature mode record data I) based on the scale settings.
158 and the extracted and learned feature mode record data])
156 and -C. And schedule settings and C, 11, AM+IguOM
It is assumed that data initialized by K is stored and these l feature modes are not erased by automatic learning. Although an example of the present invention has been described above, the present invention is not limited to this, and for example, the circuits 151 to 156 and the circuits shown in FIG. 1
59, the part that performs the special L "change mode setting and 4]j configuration can be independently operated with the elevator-1li1ffn1 device. t in the case of equipment that does
All of the above-mentioned feature mode setting and generation are processed by a separate digital computer. Q, Bill? Processing by a computer installed in the monitor, elevator control room, etc. 111
It is also possible to have a 1 sting-no configuration. Furthermore, tyo/1i41iii s53-143513
The VC displayable l1-type elevator maintenance equipment allows this equipment Kt, f: Connection 1-, elevator-
It is also possible to create a characteristic mode for elevator control that takes into account all changes in traffic demand, sets a new % characteristic mode, and generates a characteristic mode that has already been registered. [Effects of the Invention] As explained above, according to the present invention, it is possible to accurately extract the characteristic mode of the elevator traffic M, and it is possible to accurately extract the characteristic mode of the elevator traffic M, and to detect changes in the elevator intersection ′311τ1 due to changes in the building environment. It has become a suppressor that can adapt and generate distinctive features.

[Brief explanation of the drawing]

1 to 8 are diagrams for explaining the principle according to an embodiment of the present invention, FIG. 9 1 to 11 are configuration diagrams of an embodiment of a traffic demand characteristic detection device according to the present invention, and FIG. 12 and FIG. 13 is a diagram illustrating the recorded butara music. 110... Elevator control l system, 130... Traffic information detection circuit, 140... Time signal generation circuit, 150
...Traffic demand 1 hour fine mode identification circuit, 157...
Traffic demand characteristic identification circuit, 160... 1 timing mode information learning circuit, 170... control 1 form selection circuit, 15
1... Traffic demand feature element value calculation circuit, 155...
Elevator control characteristic mode setting circuit, 156...
・Ele [Yokozoku Korei (α) 7-sword (b) rF Minahide O Dokansaki-ρ\Kokuzukoroya Rider Minabz zKl vt
- V (V) Figure 70 No. 1 Door q 72 D152 3 Product procedure amendment (method) ilF, 'l' l: Long' Self-attachment Mr. Kazuo Sugi 11 f'l
Table 7 of Showa 5th Fi 1 - Wait + f'l lf'fi No. 133373
Name of the invention Elevator - Transport WII essential feature detection device supplementary 11 for each 1.1'1 1. Marunouchi
1'115a] issue' It, :'+ll 14
Ceremony? $ shiso per day 1 degree Name of company representative: 3 111 Katsu Shigeyo Osamu Hitoi... 11 East I: and a brief description of the drawings in the specification. Supplementary details 1. Submit a "document certifying authority of representation" as shown in the attached sheet. 2. In the detailed statement, page 47, line 13, change ``desu.'' to ``, 1st
FIG. 4 is an elevator schedule setting diagram. ” is corrected.

Claims (1)

[Claims] 1. An elevator that provides multi-floor service, is provided with means for detecting all the traffic m and y of the elevator, and in the elevator that is controlled f11 (I) according to the traffic demand button, 1 for elevator-traffic demand characterized by comprising means for setting a characteristic mode of traffic demand for elevator control MTL, and means for generating all of the control characteristic modes according to the traffic demand detected by the above NC. Time fineness detection device. 2. %Y [Claim 1, paragraph 9:8.-The above! [The fine mode setting means is an elevator configured to arbitrarily set and set a plurality of modes in advance. 3. In claim 1, the upper nc generation means includes a means for extracting a characteristic mode from the detected traffic N points, and a fiiffed elevator-intersecting one point. 4. In claim 3, the feature mode setting means sets the extraction mode of the extraction means to '1!j for control.
Elevator-traffic demand characteristic detection device configured to set failure mode pressure rL 5, % Ft'r In claim 3, the generating means includes the flilJ item corresponding to the extraction mode Ic R1 of the extraction means. 14) An elevator-traffic demand feature detection device configured to generate all feature modes for use. 6. In claim 3, when the above-mentioned mode does not correspond to the above-mentioned control feature mode, the generation means converts the current nuclear extraction mode into a new control feature mode? ?
The /lf sign detection device of the next elevator traffic demand is configured to be set to fine mode. 7. Range of /l=♀S'Fnff In the first term, the generating means includes means for recognizing characteristic data h3 of the detected traffic demand, and means for recording the recognized characteristic data. , this recorded characteristic data is "an elevator equipped with means for extracting a characteristic mode of traffic demand for a predetermined period; The demand detection means includes means for storing the traffic information data of the elevator, and means for outputting the accumulated data when the accumulated data of the predetermined information reaches a predetermined number m. 9'P'r ITI'Mi'J In the first term of the desired range, the traffic demand detection means detects the traffic (
A data accumulation means for accumulating information data, a %' accumulation time measuring means for measuring the accumulation time, and at least a "distribution accumulation, fI?" A traffic demand detection tine generation means that determines the cycle of detecting the traffic demand from the data obtained, and calculates the density of the traffic demand from the accumulated data and the above fttE accumulation time, and uses it as part of the traffic N detection data. Elevator-traffic γfl feature detection device located 1 meter away from detection data output means. io, trH′ An elevator-traffic demand characteristic detecting device configured to express characteristics of traffic demand based on the following. 11, I [\N'F iiV Quest (D 1llix
I! , (] m 7 JJi Ic 1;-vh,
The feature recognition means recognizes the characteristics of the intersection 11T1nFf by at least an element recognition means representing the overall traffic W, a four-element recognition means representing the state of passenger movement between each floor, and a plurality of these elements. A device for detecting fl'V signs of elevator traffic and irl demand, which is configured to detect. 12,4′! i' Scope of claims! In item ¥11, the element recognition means representing the movement state of the passenger is an elevator-traffic demand characteristic detection device that determines a distribution function of frequency of use for each floor and uses it as the value of the element. 13 In claim 11, the above-mentioned passenger transfer! Element δ3 representing the lil+ state The identification means detects a floor that is used more frequently than other floors, and uses the floor number for identifying the detected floor as the value of the above element.Elevator-traffic demand 14. Scope of Patent Rf1 In Clause 7, the traffic wind detection means shall be configured to output an output when at least a predetermined traffic exceeds [ft e]. The above-mentioned one-time data recording means records new l-feature recognition data after outputting the l-feature recognition data of the traffic demand output by the detector I, and -CIf according to the traffic demand density.
A device for detecting characteristics of elevator-traffic demand that changes the cycle of fff-data recording. 15, 'P! j #'F' range 7th term 1,
The recognition data recording means is configured to record only for a predetermined period of time, and is configured to continue to generate a traffic demand characteristic mode even after the tK characteristic mode is set. 16,4: r Fl'l' In claim 7,
The above-mentioned recognition data recording means is used to record the recognized traffic demand.
11 or time fK: corresponds to recognition data; elevator-traffic demand characteristic detection device 7 configured to record
'j. 17, /l'! ? In claim 7, the recognition data recording means performs recognition of the transportation demand if the data indicating the feature recognized in the feature recognition data is insufficient to be extracted by the feature extraction means. Elevator configured not to record data - t of traffic demand
P! J symptom detection device @. 18. In claim 7, the characteristic mode extracting means extracts the amount of data included in the traffic demand at the elevator after the recognition data of the traffic demand for a predetermined period is recorded by the characteristic data recording means. In order to extract the timing mode that exists in it ('1 yearn for, frequently used I, the features that occur in reality are set as feature modes - [411 + elevator created with 1 so as to be ill - t features of traffic demand + color IJ: l device. 19. 'I'!, ?I/1 in claim 18
Then, the IF characteristic mode extracting means calculates for each characteristic the total number of hours of traffic demand for which the characteristics indicated by the recorded characteristic data match or are similar, and has a period longer than a predetermined time period. Detection of features of IT1 demand configured to extract features, and extracting one tabular feature from among the many features of the traffic demand for the elevator mentioned above. Device. 2. In claim 7, the feature mode extraction means comprises an initial element function for each characteristic mode. Performance Wf of the evaluation function from the evaluation formula for characteristic mode extraction
: An elevator-traffic demand feature detection device configured to extract a predetermined number of feature modes from A1a of the feature mode extraction evaluation function. 21.7MF Claim No. 20 provides that the feature extraction evaluation formula includes at least function terms relating to the saliency of the % squad, the time when special C break occurs, and the continuity of lr\hui. A device for detecting lpf symptoms of commuting demand. 22.Special Bo "H Blue Request Range @ Item 4, the generation means is configured to set the mode continuously extracted by the extraction Sensari for a predetermined period as the control feature mode. Demand feature detection device. 23. /l', YFl'F In claim 3, the generating means performs a complete comparative evaluation of the control feature mode and the extraction mode, and changes the settings of the control feature mode. As shown in (“11? Elevator-traffic demand characteristic detection device.
Elevator - transportation demand 'l? Fine detection device.