CN105984766B - The cluster management system of elevator - Google Patents
The cluster management system of elevator Download PDFInfo
- Publication number
- CN105984766B CN105984766B CN201510085112.6A CN201510085112A CN105984766B CN 105984766 B CN105984766 B CN 105984766B CN 201510085112 A CN201510085112 A CN 201510085112A CN 105984766 B CN105984766 B CN 105984766B
- Authority
- CN
- China
- Prior art keywords
- calling
- evaluation
- distribution
- machine
- carriage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
It is of the invention that a kind of processing high efficiency for the distributive judgement for having used RTS is provided, can also select the cluster management system that most suitable carriage makes the elevator of its response as distribution carriage as early as possible even if the number of units increase of carriage.In the cluster management system of the elevator involved by present embodiment, group management control apparatus (21) possesses distribution request distribution department (32), evaluation result acquisition unit (33) and distributive judgement portion (34).Distribution request distribution department (32) issues distribution request to each machine control device, each machine control device is performed the part for being used for that the calling of new ladder riding hall to be distributed to the distributive judgement of some of each carriage and handled with the generation of new ladder riding hall calling.Evaluation result acquisition unit (33) obtains the evaluation result obtained respectively in each machine control device according to distribution request.Distributive judgement portion (34) carries out overall merit to the evaluation result of each machine control device, determines finally to distribute the carriage of new ladder riding hall calling.
Description
Technical field
The application goes out to be willing to the 2014-184364 (applyings date with Japanese Patent：On October 9th, 2014) based on, according to the Shen
It please enjoy priority.The application includes the entire disclosure of which by referring to this application.
Embodiments of the present invention are related to the cluster management system of the elevator of the operating of more carriages of control.
Background technology
In the cluster management system of elevator, if producing new ladder riding hall calling, by be registered in exhaling for each carriage
(ladder riding hall calls and car call) and car position etc. is made to carry out Calculation Estimation value by each carriage for the evaluation function of parameter.And
And the carriage in each carriage with optimal evaluation of estimate is determined as distributing carriage.So-called " distribution carriage " is that distribution ladder riding hall is exhaled
The carriage cried.
In addition, in the case of there are more carriages, the carriage of 11 is referred to as " number machine " sometimes.Also, will distribution
The carriage of ladder riding hall calling is known as " distribution carriage " or " Distribution Unit ".
Herein, the mode as the distributive judgement for carrying out ladder riding hall calling, it may be considered that be referred to as RTS (Real Time
Scheduling, real-time schedule table) new paragon.RTS be calling of the calling for considering currently to be assigned with to future influence with
The mode of the operation of each carriage is predicted the influence of calling that the whole for completing distribution does not respond, is evaluated in the calling in future.It is logical
Cross using the RTS to calculate the evaluation of estimate of each carriage, so as to select most suitable distribution carriage.
The content of the invention
However, above-mentioned RTS is predicted by the operation that each carriage is repeated in each interim allocation model of calling.Therefore, count
Calculation amount becomes very large, and expends the time untill most suitable distribution carriage is selected.
It is preferably within 1 second in general, being operated into the time for selecting distribution carriage from ladder riding hall button.Distribution is calculated
Work of the meeting to other tasks that works long hours of method adversely affects, so being needed in the case where using RTS efficient
Processing step.
Even if problem to be solved by this invention is to provide the processing high efficiency for the distributive judgement for having used RTS, sedan-chair
The number of units increase in railway carriage or compartment, which can also select most suitable carriage as quickly as possible, makes the group of its elevator responded as distribution carriage
Management system.
The cluster management system of elevator involved by present embodiment, possesses group management control apparatus and is filled with the control of multiple machine
Put, control the operation of more carriages by above-mentioned each machine control device under the control of above-mentioned group management control apparatus.
Above-mentioned group management control apparatus possesses distribution request distribution department, evaluation result acquisition unit and distributive judgement portion.Distribution
Distribution department is asked, with the generation of new ladder riding hall calling, distribution request is issued to each machine control device, makes described each number
Machine control device, which performs, is used for handle the distributive judgement of some that each carriage is distributed in the new ladder riding hall calling
A part.Evaluation result acquisition unit, obtains the evaluation obtained respectively in each machine control device according to the distribution request
As a result.The evaluation result of each machine control device of the distributive judgement portion to being obtained by the evaluation result acquisition unit integrates
Evaluate and determine finally to distribute the carriage of the new ladder riding hall calling.
According to the cluster management system of the elevator of said structure, can use the processing of the distributive judgement of RTS it is efficient,
Also select most suitable carriage as quickly as possible even if the number of units increase of carriage makes its response as distribution carriage.
Brief description of the drawings
Fig. 1 is the figure for the interim allocation model for representing RTS.
Fig. 2 is the figure for representing to apply the processing step in the case of A algorithm in RTS.
Fig. 3 is the figure for representing to apply the interim allocation model in the case of A algorithm in RTS.
The figure for the step of Fig. 4 is the centralized Control for representing RTS.
Fig. 5 is the figure for representing the A algorithm of RTS being scattered in the appearance of each machine.
The figure for the step of Fig. 6 is the decentralised control for representing RTS.
The figure for the step of Fig. 7 is scattered+centralized Control (not the applying A algorithm) for representing RTS.
Fig. 8 is the figure of the idea of interim allocation model (only itself number machine) for representing to contemplate in control panel.
Fig. 9 is the figure represented in each machine only using itself number machine as the interim allocation model of object.
Figure 10 is the figure represented in the idea for sharing the interim allocation model (using whole machine as object) that disk is contemplated.
Figure 11 is represented in shared figure of the disk using whole machine as the interim allocation model of object.
Figure 12 is the processing step of the A algorithm of the shared disk side in the scattered+centralized Control (applying A algorithm) for represent RTS
Figure.
Figure 13 is the processing step of the A algorithm of the shared disk side in the scattered+centralized Control (applying A algorithm) for represent RTS
Figure.
Figure 14 is the processing step of the A algorithm of the shared disk side in the scattered+centralized Control (applying A algorithm) for represent RTS
Figure.
Figure 15 is the figure of the structure for the cluster management system for representing the elevator involved by an embodiment.
Figure 16 is that the function of the group management control apparatus in the cluster management system for the elevator for representing the embodiment is formed
Block diagram.
Figure 17 is the frame that the function of number machine control device in the cluster management system for the elevator for representing the embodiment is formed
Figure.
Figure 18 is group management control apparatus and control device in the cluster management system for the elevator for representing the embodiment
The flow chart of processing.
Embodiment
First, for ease of understanding, to RTS (the Real Time used as the distribution control mode of elevator
Scheduling, real-time schedule table) illustrated with A algorithm.
(RTS)
RTS be set the ladder riding hall calling for approx representing to produce in the future imaginary ladder riding hall calling (imagination calling),
Except traffic control off-balancesheet also evaluates the mode of the response for imagination calling.In this case, it is necessary to from not only for Xindeng
The ladder riding hall of note calls (calling that distribution requires) but also which carriage is allocated temporarily to for imagination calling is definite
Rise, carry out the prediction (operation prediction) of run schedule.Therefore, there is the pattern distributed temporarily if the number of packages of imagination calling increases
Numerous property.
In the case where being provided with x imagination calling for c platform carriages, interim allocation model is c(x+1)It is a.
For example, in the cluster management system with 2 carriages (A machines and B machines), imagination calling is being set to 2
In the case of, interim allocation model is 8 as shown in Figure 1.If increasing the number of units of carriage and/or the number of packages of increase imagination calling,
Then pattern quantity can explosively increase.For example, if 12 groups manage and imagination calling is set to 3, then pattern quantity is
20736.
In this way, evaluated in the rts including imagination calling, if so the number of units of carriage becomes more, to definite point
The time is needed untill with carriage.
(basic condition of distributive judgement)
In the rts, with do not respond allocated calling, the calling of distribution request, imagination calling all interim distribution
Pattern is object, obtains the square value of non-response time predicted value, using the square value and commenting as the interim allocation model
Value.
For a certain interim allocation model, having distributed calling has A parts, will distribute when not responding of i of calling for each
Between predicted value be set to rai, the non-response time predicted value of the calling for distribution request is set to rn.In addition, imagination calling has F
Part, r is set to by the non-response time predicted value for calling i relative to each imaginationfi。
The RTS evaluations of estimate E of interim allocation model is expressed as following mathematical expression 1.
... (mathematical expression 1)
The non-response time means the time for being registered in distribution carriage and being responded from ladder riding hall calling, as follows
Obtain.
■ for allocated calling, obtain from ladder riding hall calling be registered in the current elapsed time and estimate from
The sum of this required time (prediction arrival time) untill responding to distribution
Callings (newly generated ladder riding hall calling) of the ■ for distribution request, obtains arriving since then of estimating to interim distribution
Required time (prediction arrival time) untill responding
■ is called for imagination, and the required time untill distribution responds temporarily (is predicted and reached from estimating
Time) in subtract time obtained by the value that was produced after several seconds of imagination calling
In addition, the value of RTS evaluations of estimate is lower, it is considered as better interim allocation model.
By the above method, RTS evaluations of estimate are obtained for all interim allocation models.Moreover, extraction evaluation of estimate becomes most
Small interim allocation model.By the interim allocation model, practically carried out to being assigned with the carriage of calling of distribution request temporarily
Distribution output.This is the basic condition of RTS.
But if merely performing this method, need more processing time.Therefore, by A algorithm described below come
Cut down calculation amount.
(A algorithm)
A algorithm is one of algorithm for being used for realization RTS modes.The A algorithm passes through as the interim distribution of evaluation object
Pattern tapers to promising pattern and is evaluated, so as to cut down calculation amount.In addition, " A " of " A algorithm " is without special meaning
Justice.1 in the polyalgorithm that will merely A, B, C be named as ... is called " A algorithm ".
Fig. 2 is the figure for the processing step for representing to apply the A algorithm in the case of A algorithm in RTS.With with 2 carriages
Exemplified by the cluster management system of (A machines and B machines).
Individually interim allocation model it will be known as " node ".For a certain node, further additional 1 calls and is faced
When distribution and generate new node.New node will be generated referred to as " to be unfolded ".By setting the upper limit to the number for carrying out the expansion,
The time is calculated so as to limit.Explore node as much as possible and can obtain more preferable run schedule, so being managed as group
Performance is favourable.
As shown in Fig. 2, in the feelings being applied to A algorithm in the cluster management system with 2 carriages (A machines and B machines)
Under condition, become following such processing step.
(1) calling (being " the ladder riding hall calling for being actually subjected to distribution " in figure) and vacation for not distributing distribution request temporarily are obtained
Think the run schedule of the state of any calling of calling (being " imaginary ladder riding hall calling " in figure).It is set to " start to save
Point ".
(2) in the case of the calling of distribution request is allocated temporarily to A machines and in the case of being allocated temporarily to B machines,
The run schedule of A, B machine is obtained respectively.For obtained dispatch list, RTS evaluations of estimate are calculated.In addition, RTS here
The computational methods of evaluation of estimate (described in detail later) different slightly from above-mentioned basic skills.
(3) in the interim allocation model for completing evaluation, the node of RTS evaluations of estimate minimum is set to expansion source node.
(4) obtain that further additional 1 imagination calls and distributing temporarily in the case of distribute temporarily from expansion source node
Pattern (node expansion).Whole carriages for that can distribute imagination calling temporarily, generate child node respectively.
In the example of Fig. 2, by " allocation model 1 " is set to expansion source node.It will be exhaled from the expansion source node by the 1st imagination
Making the situation for being allocated temporarily to A machines be set to, " allocation model 3 ", " allocation model is set to by the situation for being allocated temporarily to B machines
4”。
(5) the work predetermined number of times of above-mentioned steps (3) (4) repeatedly.
(6) in the time point terminated repeatedly of above-mentioned steps (5), the node of extraction RTS evaluations of estimate optimal (minimum).
(7) in the node extracted, the carriage of the interim calling for being assigned with distribution request of selection is used as " distribution carriage ".
(computational methods for applying the RTS evaluations of estimate in the case of A algorithm)
In A algorithm, the RTS evaluations of estimate of interim allocation model are compared among the nodes, are elected as evaluation of estimate is small
Source node is unfolded, persistently carries out the processing of node expansion.At this time, between the node as comparison other, evaluation is become sometimes
The number distributed temporarily of the object of non-response time is different.
For example, in fig. 2,2 non-response times are have rated in " start node ".In " allocation model 1 ", " distribution mould
3 non-response times are have rated in formula 2 ", " allocation model 3 ", " have rated 4 non-response times in allocation model 4 ".
The number of packages (number of packages distributed temporarily) for the non-response time that have rated is more, then RTS evaluations of estimate are bigger.Therefore, only
The small node of the number of packages that distributes temporarily is chosen as expansion source node.Therefore, called for the imagination do not distributed temporarily, it is assumed that do not ring
It is identical with " average non-response time " between seasonable, and corrected value is added in RTS evaluations of estimate.
Allocated calling is A parts, will be set to r for each non-response time predicted value for having distributed iai, will for point
The non-response time predicted value of calling with request is set to rn.Imagination calling is F parts, the imagination that will be assigned with temporarily in the node
The number of packages of calling is set to G parts, will be set to r for the non-response time predicted value of each imagination calling ifi。
The RTS evaluations of estimate Ea of interim allocation model is expressed as following mathematical expression 2.
... (mathematical expression 2)
Here, P is " average compensation value ", is set to square of average non-response time.(F-G) represent imagination calling in,
The number of packages that the node does not distribute temporarily.Value as obtained by the value for being somebody's turn to do (F-G) is multiplied by average compensation value P is added to RTS evaluations of estimate
On, so that supplemented with the evaluation of estimate for unrated imagination calling.
In the step of A algorithm, the pattern for the midway also distributed temporarily increase is evaluated.X is being provided with to c platforms carriage
In the case of a imagination calling, the sum of interim allocation model is changed into (c(x+2)- 1) ÷ (c-1) is a.
For example, in the cluster management system with 2 carriages (A machines and B machines), imagination calling is being set to 2
In the case of, interim allocation model is illustrated in figure 3 15.Compared to Figure 1, the pattern of calling is not distributed temporarily comprising a part,
It is envisaged that pattern quantity increase.
In the rts, the processing repeatedly shown in above-mentioned steps (5) is limited to preassigned number.Therefore, do not estimate
To the situation that all interim allocation model is evaluated.But one of the evaluation of estimate by only strictly selecting to be got well
Partial mode is evaluated, so as to obtain the result close to optimal evaluation of estimate in a short time.
The number of work (execution of step (3) (4)) shown in progress above-mentioned steps (5) is defined as " node expansion
Number ".In above-mentioned steps (4), the node that 1 is called and is allocated temporarily to each carriage is generated respectively.Therefore, if carrying out u times
Node is unfolded, then is changed into evaluating the interim allocation model of u × c+1 parts.Here, the number of units of carriage is set to c.It is in addition, a
Also the start node before start node expansion is included in number.
For example, in the case where having 4 carriages and having 3 imagination callings, if 21 minor node expansion are carried out, in A algorithm
In can be evaluated in 341 patterns using 85 patterns as object.To face in contrast, there are 256 in the idea of Fig. 1
When allocation model.Therefore, by the application of A algorithm, can be cut down as the pattern of evaluation object to about 1/3rd.
(applications of the RTS to the cluster management system of elevator)
Next, the situation that RTS is applied in the cluster management system of elevator is illustrated.
The cluster management system of elevator, which has, is all together the control that group manages the overall shared disk (group's host) of control and each machine
Disk.Disk is shared to be connected with can carrying out information exchange with the control panel of each machine.RTS is being applied to the elevator of such structure
Cluster management system in the case of, it is contemplated that (a) centralized Control, (b) decentralised control, (c) be scattered+centralized Control (do not apply A to calculate
Method), (d) it is scattered+centralized Control (applying A algorithm).
(a) centralized Control
Centralized Control is by the intensive A algorithm that RTS is individually performed in group's host, group's host of the information of each machine (carriage)
Method.
The figure for the step of Fig. 4 is the centralized Control for representing RTS.Share disk 11 be group's host, information can be exchanged with it is each
Control panel 12a, 12b ... the connection of number machine.
(1) access evidence is examined and seized with shared
Prior to allocation processing, share disk (group's host) 11 from control panel 12a, 12b ... of each machine obtain car position and/
Or call the run schedule such as the state of registration, the prediction arrival time for reaching each floor.
(2) evaluation result examined and seized with sharing and taken Expert Rules etc.
Disk 11 is shared to transmit the generation of distribution request to control panel 12a, 12b ... of each machine.The control panel of each machine
12a, 12b ... provide the evaluations such as RTS to shared disk 11 and handle required information.
The main project of information described below.
The calling of distribution request is assigned to prediction arrival time in the case of each machine, reaching each floor by ■ temporarily
The evaluation result of ■ Expert Rules
So-called Expert Rules be predict arrival time beyond control targe, comprising for example avoid crowded carriage and/or
It is energy saving etc..
(3) A algorithm is performed with shared disk
Disk 11 is shared individually to perform A algorithm and carry out the postitallation evaluation of each machine.The run schedule of the present situation of each machine,
The state before calling for interim distribution distribution request, can be collected by above-mentioned steps (1), for being assigned with temporarily
The situation of desired calling is distributed, can be collected by above-mentioned steps (2).
While imagination calling is distributed temporarily to each machine while the A algorithm carried out, is individually performed by sharing disk 11.Calculated in A
There is selection in method next by the processing of the node of expansion.In this case, can appoint from the node that whole explorations is completed
The selection of meaning ground.The step of therefore, it is possible to as being illustrated in Fig. 1, performs A algorithm.
(4) distribution carriage is determined with shared disk
Share in disk 11, using the evaluation result of RTS and/or the evaluation result of Expert Rules, determine distribution carriage.Separately
Outside, the evaluation result of Expert Rules obtains in above-mentioned steps (1) from control panel 12a, 12b ... of each machine.
There is following such the strengths and weaknesses in the centralized Control.
(advantage)
■ is concentrated at 1 position and handled, and institute is so as to simply install RTS.
■ can apply A algorithm in the form of former state, so as to the section next to be explored of highest accuracy selection
Point.
■ makes the processing load of control panel 12a, 12b ... of each machine be inhibited.
(shortcoming)
■ processing, which concentrates on, shares disk 11, so cpu load memory usage amount is all very nervous.It cannot expect with 1 piece
Control base board is sufficiently evaluated.
■ is when shared disk 11 is shut down, it is difficult to performs A algorithm with control panel 12a, 12b ... of each machine.
(b) decentralised control
Decentralised control is the method for the A algorithm for disperseing execution RTS with the control panel of each machine.In the example of Fig. 5, in A~C
Control panel 12a~12c of number machine is dispersed with the processing of A algorithm.
Control panel 12a~12c of each machine, predicts the feelings that the calling of distribution request is assigned to itself number machine temporarily first
Operation under condition.At this time, the operation prediction data do not distributed temporarily to other number machine in the case of the calling of distribution request is obtained.
Herein, the starting point for the A algorithm implemented with control panel 12a~12c of each machine is set to " only distribution request
Calling is assigned to the situation of itself number machine temporarily " under allocation model.Moreover, selected by A algorithm next by expansion
Node, can only select to derive from from the allocation model " only the calling of distribution request is assigned to the situation of itself number machine " temporarily under
The pattern gone out.
The figure for the step of Fig. 6 is the decentralised control for representing RTS.Share disk 11 be group's host, information can be exchanged with it is each
Control panel 12a, 12b ... the connection of number machine.
(1) it is predetermined that operation is exchanged between the control panel of each machine
Prior to allocation processing, car position and/or calling are regularly exchanged between control panel 12a, 12b ... of each machine
The run schedule such as the state of registration, the prediction arrival time for reaching each floor.
(2) it is made the fortune calling of distribution request to be assigned to itself number machine temporarily in the case that in the control panel of each machine
Row dispatch list
If producing distribution request, the operation of each machine is made as described below with control panel 12a, 12b ... of each machine
Dispatch list.
■ is made the dispatch list of itself number machine in the case of the calling of distribution request is assigned with temporarily.With interim point
The prediction arrival time of timing can be calculated by existing processing.
In the case of callings of the ■ for not distributing distribution request temporarily, the dispatch list of other number machine is made.Without interim point
The prediction arrival time of timing, is previously obtained by the information exchange in above-mentioned steps (1).
(3) A algorithm is implemented for object with imagination calling in the control panel of each machine
Control panel 12a, 12b ... of each machine, for object, optimal pattern is explored using A algorithm only with imagination calling.
Using from by exploring the RTS evaluations of estimate of evaluation of estimate that obtained optimal interim allocation model calculates as itself number machine.Into
One step, control panel 12a, 12b ... of each machine carry out the evaluation of Expert Rules.RTS evaluations of estimate are concentrated with Expert Rules evaluation of estimate
In shared disk 11.
(4) distribution carriage is determined in shared disk
Share disk 11 to combine RTS evaluations of estimate with Expert Rules evaluation of estimate, determine that the calling of final distribution distribution request is (new
Ladder riding hall call) carriage.
There is following such the strengths and weaknesses in the decentralised control.
(advantage)
The control panel and column processing A algorithm of ■ each machines, so treatment effeciency improves.
■ cannot be participated under the situation of allocation processing even in shared disk, also easily perform RTS.
(shortcoming)
■ also evaluates the run schedule distributed temporarily to other number machine in the case of imagination calling in A algorithm.Therefore,
The control panel needed in each machine keeps the run schedule of whole machine, consumes more storage region.
(c)+centralized Control (not applying A algorithm) is disperseed
Scattered+centralized Control is only to evaluate all interim distribution moulds as object using itself number machine with the control panel of each machine
The run schedule of formula, evaluated with shared disk whole machine whole allocation models method.Herein, without using A algorithm.
That is, above-mentioned (a), (b) method in, by using premised on A algorithm make consider.But in order to cut down calculation amount,
It is also contemplated for different thinkings.The pattern count distributed temporarily seems more, this is because cluster management system to be integrally set to object, is divided
Some calling is allocated temporarily to situation some carriage.
If considered so that carriage is only 1, the calling imagination calling for distribution request, there is no " to which
A's machine distributes " the problem of, and the problem of only exist " whether being distributed to itself number machine ".It is therefore contemplated that after dividing situation
The quantity of interim allocation model is not c(x+1)It is a but 2(x+1)It is a.
Therefore, consider that 2 will be divided into the operation prediction of imagination calling as described below.Processing 1 is decentralized processing,
Processing 2 is centralized processing.
Processing 1：In the control panel of each machine, the calling imagination that distribution request is only made using itself number machine as object is exhaled
The interim allocation model of whole cried, and evaluate the run schedule (operation prediction) of the interim allocation model.
Processing 2：In shared disk, using the evaluation result by the obtained each machine of above-mentioned processing 1, system is managed as group
System is overall, calculates the allocation evaluation value of whole allocation models of whole machine.
The figure for the step of Fig. 7 is scattered+centralized Control (not the applying A algorithm) for representing RTS.It is group's host to share disk 11,
It is connected with control panel 12a, 12b ... of each machine while information can be exchanged.
(1) run schedule when the control panel of each machine is made the interim distribution of itself number machine
If produce distribution request, control panel 12a, 12b ... of each machine is made receive distribution request in the case of
Run schedule.In addition, receive the run schedule before distribution request, can be in advance in control panel 12a, 12b ... of each machine
In be ready to.
(2) the whole allocation models of amount evaluation of itself number machine are pressed in the control panel of each machine
Control panel 12a, 12b ... of each machine is made expression and the calling of distribution request and imagination is called interim distribution still
The interim allocation model of whole of itself number machine, evaluation and the interim corresponding traffic control of allocation model are not allocated temporarily to
Table.
The idea of the interim allocation model contemplated in control panel (only itself number machine) is shown in Fig. 8.If imagination is called
X parts are set to, then interim allocation model presses each's machine respectively there are 2(x+1)It is a.
Without considering the run schedule in the stage of imagination calling, completion is made in above-mentioned steps (1), so making herein
Run schedule in the case of being distributed into additional imagination calling and temporarily, and evaluated.As only with itself number machine
For the example of the interim allocation model of object, there are 8 as shown in Figure 9 in the case of being 2 in imagination calling.The pattern quantity
The number of units of carriage is not depended on.
Control panel 12a, 12b ... of each machine will utilize the above method obtained and be commented with the corresponding distribution of all mode
Value is sent to shared disk 11.
(3) evaluation result of each machine is synthesized in shared disk, and evaluates all interim allocation models
Allocation evaluation value of the disk 11 using control panel 12a, 12b ... the receiving from each machine is shared, evaluation is whole as system
The interim allocation model of body.
The idea of the interim allocation model contemplated in shared disk (whole machine object) is shown in Figure 10.This is distributed temporarily
Pattern is the pattern for representing to distribute to which calling which carriage, and there are c as a whole(x+1)It is a.
It it is 2 in imagination calling for example, in the cluster management system with 3 carriages (A machines, B machines, C machines)
In the case of, share the interim allocation model as shown in figure 11 for evaluating 27 in disk 11.At this time, each machine can be utilized by sharing disk 11
Control panel 12a, 12b ... in the evaluation result obtained, readily calculate the evaluation of estimate of each interim allocation model.
That is, in Figure 11 " in the case of synthesis model 8 ", with reference to the independent pattern of Fig. 9.Moreover, by being combined to A machines
Pattern 5, to B machines integrated mode 2 and to C machines integrated mode 3, calculate and comment while the run schedule of each machine can not be made
Value.It is, the evaluation to all interim allocation models can be implemented with less calculation amount.
(4) final distribution carriage is determined in shared disk
Disk 11 is shared to combine the evaluation of estimate of above-mentioned all interim allocation model and utilize the evaluation of estimate obtained by Expert Rules,
Determine the carriage of the calling (new ladder riding hall calling) of final distribution distribution request.
There is following such the strengths and weaknesses in the scattered+centralized Control.
(advantage)
■ can evaluate more interim allocation models (evaluable all mode) compared with the decentralised control of above-mentioned (b).
In the control panel of ■ each machines, the operation prediction of itself number machine is only carried out, so the consumption of storage region is less.
(shortcoming)
■ handles whole imagination callings, thus calculation amount relative to the number of packages of imagination calling by exponential function and
Increase.Required number is predicted in operation, is 16 times when imagination calling is 3, is 32 times when imagination calling is 4.If
Consider the computing capability of control panel, then it is it is generally acknowledged that difficult using the evaluation comparison of more than 5 imagination callings.
The synthesis that ■ needs to add interim allocation model to shared disk is handled.
■ when shared disk is shut down, in the control panel of each machine one need to undertake decentralized processing with focus on this two
Side, the response for distributing output decline.
(d)+centralized Control (applying A algorithm) is disperseed
As described above, the method that combination combines decentralized processing with centralized processing, it is handled efficient, storage region
Consumption is also few, and advantage is very big.But in the method for above-mentioned (c), all interim allocation models are evaluated with shared disk, so meter
This two side accordingly changes greatly calculation amount with the number of units imagination calling number of packages of carriage.
For example, in 12 cluster management systems, it is necessary to obtain distribution to 20736 in the case of being 3 in imagination calling
Evaluation of estimate.It is highly difficult to handle this number of packages in a short time.Therefore, A algorithm is used in the evaluation processing for sharing disk side.
First, the example of all interim allocation model with reference to shown in Figure 11, again to the control of shared disk 11 and each machine
The task sharing of disk 12a, 12b ... processed illustrate.
Share disk 11 to evaluate whole synthesis models and select to evaluate most excellent synthesis model, determine actual distribution calling
Carriage (distribution carriage).Herein, so-called " synthesis model " means that new calling (calling that distribution requires) such as distributes to No. A
Machine, imagination calling 1 distribute to B machines like that, represent the mould which carriage to distribute the calling of the object of distribution to
Formula.
A machines are being distributed into new calling, imagination calling 1 is distributed to B machines, C machines are distributed into imagination calling 2
In the case of, it is the synthesis model 6 shown in Figure 11.For the synthesis model 6, if separately considering by each machine, then it represents that it is following that
The state of sample.
The new calling of ■ A machines distribution, but imagination calling 12 is not distributed.
■ B machines distribution imagination calling 1, but new calling and imagination calling 2 are not distributed.
■ C machines distribution imagination calling 2, but new calling and imagination calling 1 are not distributed.
If state set is used for Fig. 9, A machines are independent pattern 5, and B machines are independent pattern 3, and C machines are independent mould
Formula 2.Herein, it is not interim that so-called " independent pattern " means to be conceived to 1 carriage, the interim distribution of such as new calling, imagination calling 1
Distribution like that, represent each calling to be distributed or is not assigned to the pattern of itself number machine.
That is, the evaluation of estimate of synthesis model 6, value, right that can be as obtained by combining for A machines by the calculating of independent pattern 5
Calculated in B machines by the value obtained by the calculating of independent pattern 3, for C machines by the value of independent pattern 2 calculating gained.In this way, close
The independent pattern of each machine can be combined to evaluate by becoming the mode.
Next, illustrated for the method evaluated in shared disk 11 using A algorithm.
If as described above, using A algorithm, promising pattern is tapered to using as the interim allocation model of evaluation object
And evaluated, institute is so as to cutting down calculation amount, can determining distribution carriage in a short time.
For example, in the cluster management system with 12 carriages, in the case where imagination calling is 3, Figure 11's is interim
Allocation model (synthesis model) is 20736 as described below.
The carriage that ■ distribution newly calls is 12, so for 12
The carriage of ■ distribution imaginations calling 1 is also 12, so for 12
The carriage of ■ distribution imaginations calling 2 is also 12, so for 12
The carriage of ■ distribution imaginations calling 3 is also 12, so for 12
4 above-mentioned " 12 " all independent, so being as a whole 12 × 12 × 12 × 12=20736.
In above-mentioned idea, it is contemplated that such as " new calling selection A machines, 1 selection B machines of imagination calling, 2 choosing of imagination calling
Select C machines, 3 selection D machines of imagination calling " like that, in the case of its carriage of whole calling selection distribution.
If using A algorithm, a part of evaluation is only carried out, so even if not evaluated all interim allocation models
Result can be deduced.
That is, it is not from the conjunction for being initially ready for determining assigning them to which carriage for whole callings in A algorithm
Become the mode, initially since the state that new calling imagination calling does not distribute temporarily.Moreover, with 11 as needed of one side
Ground adds calling and is allocated while the mode reappraised is handled.Therefore, the synthesis model and A being illustrated in Figure 11
The pattern to be considered in algorithm differs.The synthesis model to be considered in RTS (exploration node), for part calling some not
Determine the carriage of distribution.
Herein, it is described in detail with reference to Figure 12 to Figure 14.
Figure 12 to Figure 14 is the figure for representing to share the processing step of the A algorithm of disk side.Represent with 3 carriages (No. A
Machine, B machines, C machines) cluster management system in, imagination calling is set to example in the case of 2.
(Na) (Nb) in figure ... (Nm) represents the state of the node in each stage.
First, the run schedule of the whole machine for the state do not distributed as new calling and imagination calling, prepares section
Point (Na).
Next, with which branch machine new calling is allocated temporarily to correspondingly, ready node (Nb) (Nc) (Nd).It is right
In the evaluation of estimate of the run schedule of each machine, calculated and completed by the method for Fig. 9.
Herein, illustrated by taking node (Nb) as an example.
■ contemplates the state for being assigned with to A machines and newly calling temporarily.At this time, A machines are changed into the fortune different from node (Na)
Row dispatch list.B machines, the run schedule of C machines are identical with node (Na).
In A algorithm, can also the result calculated by " the independent pattern " of each machine be used to be evaluated.Because completely not
The pattern that the pattern, and/or an only interim distribution part of the interim new calling of distribution and/or imaginary calling call also is contained in individually
Pattern.
In the evaluation of A machines, the evaluation of estimate obtained from the independent pattern 5 of A machines is used.
In the evaluation of B machines, the evaluation of estimate obtained from the independent pattern 1 of B machines is used.
In the evaluation of C machines, the evaluation of estimate obtained from the independent pattern 1 of C machines is used.
(Nb) evaluation of estimate of node entirety becomes above-mentioned 3 evaluations of estimate and upper plus " average compensation value " × 2
The value of product gained.The reason for plus 2 times of mean compensation algorithm is that the situations that 2 imagination callings are distributed together with are compared, node (Nb)
In the state of the calling that is assigned it is 2 few.It is it is, (average to mend by the evaluation of estimate for supplementing 2 average 1 call volumes
Repay value) so as to supplement the evaluations of estimate of 2 amounts of deficiency.By the way that average compensation value is added in evaluation of estimate in advance, so as to then can
Compared with enough evaluations of estimate with imagination to be called to the node for distributing to any's machine.
In average compensation value, the expected value for the incrementss that caused evaluation of estimate is allocated by additional 1 calling is used
(average non-response time square etc.).
If " evaluation of estimate " mentioned here be described in more detail,
So-called " evaluation of estimate of the independent pattern 5 of A machines ", is to distribute to A machines and the imaginary vacation of calling 1 for new calling
Think calling 2 be not assigned to A machines in the case of the single run schedule of A machines evaluation of estimate.The evaluation of estimate be for
Having distributed to upper newly called plus the response of A machines of the sum of square value of predicted value of " non-response time " of the calling of A machines is
The obtained value of predicted value of non-response time only.In addition, the evaluation of estimate is used as " independent pattern by the control panel 12a of A machines
5 " calculate completion.
So-called " evaluation of estimate of the independent pattern 1 of B machines " is all to be not assigned to No. B for new calling imagination calling 1,2
The evaluation of estimate of the run schedule of B machines in the case of machine.The evaluation of estimate is set to for the calling that is assigned to B machines
The sum of the square value of predicted value of " non-response time ".
So-called " evaluation of estimate of the independent pattern 1 of C machines ", is all to be not assigned to No. C for new calling imagination calling 1,2
The evaluation of estimate of the run schedule of C machines in the case of machine.The evaluation of estimate is set to for the calling that is assigned to C machines
The sum of the square value of predicted value of " non-response time ".
As described above, the evaluation of estimate of the node entirety of (Nb) is to distribute to not ringing for whole callings of A, B, C machine
The upper obtained value of product plus average compensation value × 2 of the sum of square value between seasonable.Therefore, the evaluation of estimate of node entirety, energy
Enough combinations are obtained by the evaluation of estimate that the independent pattern of Fig. 9 is obtained and average compensation value.
Other nodes also can be calculated similarly.The main points of calculating are as described below.
■ presses each node, investigates which calling which carriage distributed to.
■ which calling is distributed to the imagination (Figure 11) of which point carriage, be replaced as by each's machine by which
The interim distribution of calling is not still allocated temporarily to the imagination (Figure 10) of itself number machine.
The evaluation of estimate and average compensation value that ■ combinations are obtained by the independent pattern of Fig. 9, obtain the evaluation of estimate of node entirety.
If the evaluation for node (Na) (Nb) (Nc) (Nd) terminates, " the choosing of the characteristic processing as A algorithm is carried out
Select the processing for the node being next unfolded ".Exactly select the evaluation of estimate among the node evaluated before this it is optimal (value is minimum
) node, to the node again, it is contemplated that further called 1 and be allocated temporarily to the situation of a certain carriage and be made ensuing
Node.
But, expansion source will not be become again by having completed the node of expansion.It is for example, complete in node (Na) (Nb) (Nc) (Nd)
Into during evaluation, node (Na) will not become expansion source again.
Among node (Nb) (Nc) (Nd), if for example, the evaluation of estimate of node (Nb) is optimal, go to node (Ne)
(Nf) evaluation of (Ng).Node (Ne) (Nf) (Ng) represents the 1st imagination calling being allocated temporarily to the situations of A machines respectively,
The situation of B machines is allocated temporarily to, is allocated temporarily to the situation of C machines.In addition, node (Ne) (Nf) (Ng) all maintains as former state
" new calling is allocated temporarily to A machines " of node (Nb) this imagination.
If the evaluation for node (Ne) (Nf) (Ng) terminates, next, finding node (Nc) (Nd) (Ne) (Nf)
(Ng) the optimal node of evaluation of estimate among, is set to expansion source node.In this case, node (Nc) (Nd) does not distribute the " the 1st
A imagination calling ", node (Ne) (Nf) (Ng) are assigned " the 1st imagination calling ", and the number of the calling of evaluation is different.But
It is, in the evaluation of estimate of calculate node, correspondingly to adjust what average compensation value was calculated with the number of unrated calling
Number, institute is so as to be mutually compared.
Herein, if for example, node (Nd) is optimal, next by node (Nh) (Ni) (Nj) expansion.If in addition, node
(Nf) it is optimal, then next the node of (Nk) (Nl) (Nm) is unfolded.
Limitation is equipped with to the number of node expansion.Such as if with 4 times for condition, evaluation is as follows：
1st time：The evaluation of node (Nb) (Nc) (Nd)
2nd time：The evaluation of node (Ne) (Nf) (Ng)
3rd time：The evaluation of node (Nh) (Ni) (Nj)
The 4th：The evaluation of node (Nk) (Nl) (Nm)
If the expansion of 4 minor nodes terminates, finally, the node with optimal evaluation of estimate is found.As a result, for example, node
(Nm) it is optimal.At node (Nm), new calling is assigned to A machines, so in this allocation processing, A machines are confirmed as point
With carriage.
In this way, in scattered+central controlled structure, A algorithm is carried out by using group's host, and with following such excellent
Point.
(advantage)
Do not wasted in the processing of evaluation of estimate of independent pattern is obtained.It is, produced in node (Nf) (Nk) (Nm)
The predictions of B machines all use independent pattern 3.At node (Ni), the prediction of B machines also uses independent pattern 3.
In the decentralised control being illustrated in Figure 5, node (Nf) (Nk) (Nm) is evaluated in the control panel 12a of A machines,
Node (Ni) is evaluated in the control panel 12b of C machines.Therefore, it is necessary to the control panel in the control panel 12a of A machines and C machines
This two side of 12c carries out the prediction of the independent pattern 3 of B machines, wastes very much.
In contrast, in the case where using scattered+centralized Control (applying A algorithm), the independent pattern 3 of B machines is commented
Valency carries out in the control panel 12b of B machines in the stage (stage of Fig. 9) of the complete individually pattern of evaluation in advance.Therefore, A calculations are carried out
In the shared disk 11 of method, the evaluation of estimate for calculating the independent pattern completed only is taken out, it is not necessary to repeated as the decentralised control of Fig. 5
Evaluated.
In addition, in the above example, in order to make explanation simple, 3 carriages are set to, the expansion of 4 minor nodes is illustrated,
But it is unfolded more often, can further save the repetition evaluation between each machine in the number of units and/or node of carriage
Waste significantly to cut down calculation amount.
(embodiment)
Hereinafter, the cluster management system of the elevator for the scattered+centralized Control (applying A algorithm) for having used above-mentioned (d) is carried out
Explanation.
Figure 15 is the figure of the structure for the cluster management system for representing the elevator involved by an embodiment, is represented to more electricity
Ladder carries out the structure of group's management.In addition, " elevator " mentioned here is substantially " carriage ", in the case of there are more carriages
Also referred to as " number machine ".
The system possesses group management control apparatus 21 and is connected to number machine control device of the group management control apparatus 21
22a、22b、22c….Group management control apparatus 21 is corresponding to above-mentioned general controls disk (group's host), controls each machine with being all together
Operating.
Number machine control device 22a, 22b, 22c ... is corresponding to above-mentioned shared disk (group's host), controls each machine with being all together
Operating.Number machine control device 22a, 22b, 22c ... is set by each's machine, carries out the fortune of corresponding machine respectively
Turn control.Specifically, the motor (not shown) (hoist engine) for making carriage 23a, 23b, 23c ... lifting work is carried out
Open and close controlling of control and/or door etc..
In addition, group management control apparatus 21 is realized by computer respectively with number machine control device 22a, 22b, a 22c ....
Carriage 23a, 23b, 23c ... lift work by the driving of motor (not shown) (hoist engine) in elevating road
Make.In the interior of carriage 23a, be provided with for specify passenger go to floor go to floor button 24a.Building is gone to by this
The information for going to floor specified by layer button 24a, the car call as A machines are transmitted via a number machine control device 22a
To group management control apparatus 21.For other carriages 23b, 23c ... and same, be respectively arranged with go to floor button 24b,
24c…。
In addition, in the ladder riding hall (the elevator riding Room) of each floor, be configured with for register the ladder riding hall of ladder riding hall calling by
Button 25a, 25b, 25c ....These ladder riding hall button 25a, 25b, 25c ... are connected to group through not shown transmission line cable and manage control
Device 21 processed.Ladder riding hall button 25a, 25b, 25c ... have uplink button and descending button, are configured to the side of going to user
To correspondingly pressing uplink button or descending button.In addition, only there is uplink button in orlop, in the case where the superiors only have
Row button.
In addition, so-called " ladder riding hall calling " is stepped on by the operation of ladder riding hall button 25a, 25b, 25c ... by each floor
The signal of the calling of note, includes registration floor and the information for going to direction.So-called " car call " is by being arranged at cage
The signal for the calling that the interior operation for going to floor button 24a, 24b, 24c ... is registered, includes the information for going to floor.
Herein, in the present embodiment, have in group management control apparatus 21 and number machine control device 22a, 22b, 22c ...
It is ready for use on the work(realized and controlled using the method distribution that i.e. " scattered+centralized Control (applying A algorithm) " is carried out of above-mentioned (d)
Energy.
Figure 16 is the block diagram of the functional structure of the group management control apparatus 21 in the cluster management system for represent elevator.
In group management control apparatus 21, as the function with the corresponding host of the method for above-mentioned (d) (mast) side, tool
Standby calling storage part 31, distribution request distribution department 32, evaluation result acquisition unit 33, distributive judgement portion 34, distribution output section 35.
Calling storage part 31 stores the boarding registered by the operation of ladder riding hall button 25a, 25b, 25c ... of each floor
The Room calls.Registration floor and the information for going to direction are included in ladder riding hall calling.
Distribution request distribution department 32, when being stored with new ladder riding hall calling in calling storage part 31, check mark machine control device
22a, 22b, 22c ... issue distribution request, perform it and are used to new ladder riding hall calling distributing to each carriage 23a, 23b, 23c ...
The distributive judgement of any one processing a part.
At this time, the calling of the required distribution requests of RTS with imagination calling by the number of paying machine control device 22a, 22b,
22c….So-called " calling of distribution request " is the ladder riding hall actually to be distributed calling, that is, is new ladder riding hall calling.
So-called " imagination calling " assumes that the ladder riding hall calling to produce in the future, be consider past operation real result data and/
Or each floor of building traffic need etc. and be made.
Evaluation result acquisition unit 33 obtains what is respectively obtained according to distribution request in number machine control device 22a, 22b, 22c ...
The evaluation result of each machine.
Distributive judgement portion 34 evaluation result of each machine obtained by evaluation result acquisition unit 33 is carried out overall merit with
Determine finally to distribute the carriage that new ladder riding hall calls.In detail, distributive judgement portion 34 first by machine control device 22a,
The evaluation result synthesis of 22b, 22c ....Then, the processing as A algorithm, using among all mode as the interim of evaluation object
Allocation model tapers to promising pattern and is evaluated.Thus, determine finally to distribute the carriage that new ladder riding hall calls.
Distribution output section 35 exports dividing for ladder riding hall calling to the carriage for being determined as distributing carriage by distributive judgement portion 34
With instruction.
Figure 17 is the frame of the functional structure of number machine control device 22a, 22b, 22c ... in the cluster management system for represent elevator
Figure.
In number machine control device 22a, 22b, 22c ..., as with the corresponding slave of the method for above-mentioned (d) (slave)
The function of side, possesses distribution request receiving portion 41, interim allocation model generating unit 42, operation prediction and evaluation portion 43, evaluation result
Transport unit 44.
The distribution that distribution requires receiving portion 41 to receive to be issued by the distribution request distribution department 32 of group management control apparatus 21 please
Ask.At this time, the distribution that distribution requires receiving portion 41 to receive to bring from group management control apparatus 21 when distribution request is issued please
The calling asked is called with imagination.
Interim allocation model generating unit 42, interim distribution is made based on the distribution request received by distribution request receiving portion 41
Pattern.In detail, interim allocation model generating unit 42, the calling based on distribution request only make itself number machine with imagination calling
All interim allocation models are made for object.
Prediction and evaluation portion 43 is run, is made for all interim allocation model being made by interim allocation model generating unit 42
Run schedule (operation prediction) is simultaneously evaluated.
Evaluation result transport unit 44, using the evaluation of estimate of each interim allocation model obtained by operation prediction and evaluation portion 43 as
The evaluation result of itself number machine is conveyed to group management control apparatus 21.
Next, operation of the present embodiment will be described.
Figure 18 is group management control apparatus 21 and control device 22a, 22b, 22c ... in the cluster management system for represent elevator
Processing flow chart.The processing of left side expression group management control apparatus 21 in figure, right side expression machine control device 22a,
The processing of 22b, 22c ....
If the operation of any one for ladder riding hall button 25a, 25b, the 25c ... for passing through each floor is exhaled to be registered with new ladder riding hall
Cry (step A11's be), then group management control apparatus 21 is in the information (registration floor and go to direction) for calling the ladder riding hall
After being stored in calling storage part 31, following such processing is performed.
That is, group management control apparatus 21 issues distribution request by distribution request distribution department 32 first.At this time, please in distribution
Distribution department 32 is asked, traffic needs of each floor based on past operation real result data and/or building etc. are made imagination
Call inventory (step A12).Distribution request distribution department 32, calls the imagination calling of the predetermined quantity on inventory by the imagination and divides
Calling (the new ladder riding hall calling) number of sending jointly to machine control device 22a, 22b, 22c ... (step A13) with request.
In number machine control device 22a, 22b, 22c ..., if receiving distribution request (step by distribution request receiving portion 41 respectively
Rapid B11), then perform following such processing.(divide in addition, receiving the run schedule before distribution request comprising what is do not responded
With calling and/or car position etc.), it is set in advance in the ready dispatch list in control panel 12a, 12b ... side of each machine.
First, number machine control device 22a, 22b, 22c ... are made only with certainly by interim allocation model generating unit 42 respectively
Body machine is the interim allocation model (step B12) of object.In detail, interim allocation model generating unit 42 is made expression and will divide
Calling with request and the interim distribution of imagination calling are not still allocated temporarily to all interim allocation model (references of itself number machine
Fig. 8, Fig. 9).The all interim allocation model being made is paid operation prediction and evaluation portion 43.
Operation prediction and evaluation portion 43 be made with these interim corresponding run schedule of allocation model (operation is predicted), into
Row evaluation (step B13, B14).In detail, as shown in above-mentioned (1) formula, by allocated calling and the calling of distribution request
Interim distribution, the square value for being all set to object, obtaining non-response time predicted value distributed temporarily of imagination calling, calculate this
A little square values and the evaluation of estimate as the interim allocation model.
If the evaluation of estimate (step B15's be) for all interim allocation models, evaluation result transport unit are calculated
These evaluations of estimate are sent to group management control apparatus 21 (step B16) by 44.
In number machine control device 22a, 22b, 22c ..., while perform above-mentioned processing.A number machine control device 22a, 22b,
The obtained evaluation results of 22c ... (for the evaluation of estimate for all interim allocation models being made using itself number machine by object),
Group management control apparatus 21 is paid respectively.
Group management control apparatus 21 obtains the evaluation result (step A14) of each machine by evaluation result acquisition unit 33.By
This, these evaluation results are paid distributive judgement portion 34 by group management control apparatus 21, carry out following such distributive judgement.
That is, evaluation result of the distributive judgement portion 34 based on each machine and determine the calling of final distribution distribution request and (newly multiply
The calling of the terraced Room) carriage.At this time, distributive judgement portion 34 does not evaluate divides temporarily as the obtained whole of the evaluation result of each machine
With pattern, but A algorithm is used, promising pattern will be tapered to as the interim allocation model of evaluation object and evaluated
(step A15).
In detail, as in Figure 12, to illustrated in fig. 14, distributive judgement portion 34 is between node to interim distribution mould
The RTS evaluations of estimate of formula are compared, by the small processing elected expansion source node as, persistently carry out node expansion of evaluation of estimate.According to
The combination of evaluation of estimate and average compensation value that each machine is individually obtained, obtains RTS evaluations of estimate at this time.If the section of predetermined number of times
Point expansion terminates, then the node with optimal evaluation of estimate is found in distributive judgement portion 34, is newly exhaled being assigned among the node temporarily
The carriage cried is ultimately determined to distribution carriage.
In addition, actually the evaluation of estimate as caused by Expert Rules is also combined in addition to RTS evaluations of estimate and selected most suitable
Carriage, but as caused by Expert Rules evaluation of estimate nor is it necessary that.
Distribution output section 35 exports dividing for ladder riding hall calling to the carriage for being determined as distributing carriage by distributive judgement portion 34
With instruction (step A16).For example, in the case where the carriage 23a of the A machines shown in Figure 15 is confirmed as distributing carriage, from group
Management and control device 21 exports control device 22a the distribution instruction of ladder riding hall calling.In control device 22a, according to this point
Control is driven to hoist engine (not shown) with instruction, the registration floor response for making carriage 23a be called in the ladder riding hall.
So according to the present embodiment, using RTS modes cluster management system in, each machine (control device 22a,
22b, 22c ...) disperse to perform the evaluation processing for including the interim allocation model for distributing calling and imagination calling, institute is so as to mitigation
The processing load of group's host (group management control apparatus 21).
Further, each machine is not the allocation model for being made evaluation whole machine respectively, but only using itself number machine as
Object and be made the interim allocation model of evaluation, institute is so as to saving repeats to evaluate the wave of same allocation model between each machine
Take, be efficiently obtained evaluation result.
On the other hand, group's host (group management control apparatus 21), can be selected most suitable using the evaluation result of each machine
Carriage.At this time, promising pattern will be tapered to as the interim allocation model of evaluation object by using A algorithm and carried out
Evaluation, so as to significantly cut down processing time.Therefore, even if the number of units increase of carriage, can also select most as early as possible
Suitable carriage makes its response as distribution carriage.
At least one embodiment as stated above, using the teaching of the invention it is possible to provide the processing of the distributive judgement of RTS will have been used efficient
Change, can also select even if the number of units increase of carriage most suitable carriage as early as possible and make its response as distribution carriage
The cluster management system of elevator.
In addition, several embodiments of the invention is illustrated, but these embodiments are to propose as an example
, it is not intended to limit the scope of invention.These new embodiments can be implemented in a manner of others are various, not depart from hair
In the range of bright purport, various omissions, displacement, change can be carried out.These embodiments and/or its deformation are contained in invention
Scope and/or main idea, and it is contained in the invention described in technical solution and scope equal thereto.
Claims (1)
1. a kind of cluster management system of elevator, possesses group management control apparatus and multiple machine control device, managed in the group
The operating of more carriages is controlled under the control of control device by each machine control device, it is characterised in that：
The group management control apparatus possesses：
Distribution request distribution department, its generation called with new ladder riding hall, distribution request is issued to each machine control device,
Produced as the calling of the distribution request of the new ladder riding hall calling in the future with being envisioned in the distribution of the distribution request
Each machine control device is paid in raw imagination calling, is performed each machine control device and is used for the new ladder riding hall
A part for the distributive judgement of some processing of each carriage is distributed in calling；
Evaluation result acquisition unit, it obtains the evaluation knot obtained respectively in each machine control device according to the distribution request
Fruit；With
Distributive judgement portion, the evaluation result of its each machine control device to being obtained by the evaluation result acquisition unit are closed
Into, by the way that the interim allocation model for being wherein used as evaluation object is tapered to promising pattern and is evaluated, so that it is determined that
The carriage of the new ladder riding hall calling is finally distributed,
Each machine control device possesses：
Distribution request receiving portion, its calling with the distribution request and the imagination calling receive the distribution request in the lump；
Interim allocation model generating unit, its calling based on the distribution request received by the distribution request receiving portion with it is described
Imagination calling is made multiple interim allocation models using itself number machine as object；
Prediction and evaluation portion is run, the operation of its each interim allocation model to being made by the interim allocation model generating unit is pre-
Survey is evaluated；With
Evaluation result transport unit, its using the evaluation of estimate of each interim allocation model obtained by the operation prediction and evaluation portion as
The evaluation result of itself number machine is sent to the group management control apparatus.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2014184364A JP5920674B2 (en) | 2014-09-10 | 2014-09-10 | Elevator group management system |
| JP2014-184364 | 2014-09-10 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105984766A CN105984766A (en) | 2016-10-05 |
| CN105984766B true CN105984766B (en) | 2018-04-17 |
Family
ID=55757470
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510085112.6A Active CN105984766B (en) | 2014-09-10 | 2015-02-16 | The cluster management system of elevator |
Country Status (4)
| Country | Link |
|---|---|
| JP (1) | JP5920674B2 (en) |
| CN (1) | CN105984766B (en) |
| MY (1) | MY174689A (en) |
| SG (1) | SG10201501070UA (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6801050B2 (en) * | 2019-06-12 | 2020-12-16 | 東芝エレベータ株式会社 | Group management control device and group management control system |
| JP7027516B1 (en) * | 2020-12-17 | 2022-03-01 | 東芝エレベータ株式会社 | Group management control device for double deck elevators and group management control method for double deck elevators |
| JP7230104B2 (en) * | 2021-05-26 | 2023-02-28 | 東芝エレベータ株式会社 | Elevator group management control device and elevator group management control method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS61273477A (en) * | 1985-05-29 | 1986-12-03 | 株式会社東芝 | Group controller for elevator |
| JP4949722B2 (en) * | 2006-04-13 | 2012-06-13 | 株式会社日立製作所 | Elevator group management control system |
| JP2011144025A (en) * | 2010-01-15 | 2011-07-28 | Toshiba Elevator Co Ltd | Group management apparatus for elevator |
-
2014
- 2014-09-10 JP JP2014184364A patent/JP5920674B2/en active Active
-
2015
- 2015-02-05 MY MYPI2015700358A patent/MY174689A/en unknown
- 2015-02-12 SG SG10201501070UA patent/SG10201501070UA/en unknown
- 2015-02-16 CN CN201510085112.6A patent/CN105984766B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN105984766A (en) | 2016-10-05 |
| MY174689A (en) | 2020-05-07 |
| JP2016056001A (en) | 2016-04-21 |
| SG10201501070UA (en) | 2016-04-28 |
| JP5920674B2 (en) | 2016-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Multi-agent based real-time production scheduling method for radio frequency identification enabled ubiquitous shopfloor environment | |
| Hao et al. | Implementing a hybrid simulation model for a Kanban-based material handling system | |
| Tang et al. | An improved differential evolution algorithm for practical dynamic scheduling in steelmaking-continuous casting production | |
| Kim et al. | Synchronized production planning and scheduling in semiconductor fabrication | |
| JPH10261122A (en) | Work distribution optimization method | |
| CN105984766B (en) | The cluster management system of elevator | |
| Ioannidis et al. | Fuzzy supervisory control of manufacturing systems | |
| Zhang et al. | Extending augmented Lagrangian coordination for the optimal configuration of cloud-based smart manufacturing services with production capacity constraint | |
| CN108053047A (en) | Cloud resources of production scheduling methods, devices and systems | |
| Yan et al. | Modeling, scheduling and simulation of product development process by extended stochastic high-level evaluation Petri nets | |
| CN105974891A (en) | Mold production process self-adaptive control method based on dynamic spectacular | |
| CN101872378A (en) | Method for modeling complex system on basis of time Petri net and Agent | |
| JP2011159230A (en) | Scheduling support system, scheduling support method, and scheduling support program | |
| Khalid et al. | Modified particle swarm algorithm for scheduling agricultural products | |
| Shukla et al. | An agent-based architecture for production scheduling in dynamic job-shop manufacturing system | |
| Manupati et al. | Intelligent search techniques for network-based manufacturing systems: multi-objective formulation and solutions | |
| Zhang et al. | Vehicle dynamic dispatching using curriculum-driven reinforcement learning | |
| CN113822585A (en) | Intelligent smelting factory informatization management system | |
| Munoz et al. | Distributed approach to group control of elevator systems using fuzzy logic and FPGA implementation of dispatching algorithms | |
| Gonzalez et al. | Operation of pipeless batch plants–II. Vessel dispatch rules | |
| Zou et al. | An effective population-based iterated greedy algorithm for solving the multi-AGV scheduling problem with unloading safety detection | |
| Peng et al. | A collaborative manufacturing execution platform for space product development | |
| Tripathi et al. | Multi agent system in job shop scheduling using contract net protocol | |
| Muñoz et al. | Implementation of dispatching algorithms for elevator systems using reconfigurable architectures | |
| Ming et al. | A New Multi-Agent Model of Workshop Scheduling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1229780 Country of ref document: HK |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |