CN104961009B - Many elevator in parallel operation control method for coordinating based on machine vision and system - Google Patents
Many elevator in parallel operation control method for coordinating based on machine vision and system Download PDFInfo
- Publication number
- CN104961009B CN104961009B CN201510279503.1A CN201510279503A CN104961009B CN 104961009 B CN104961009 B CN 104961009B CN 201510279503 A CN201510279503 A CN 201510279503A CN 104961009 B CN104961009 B CN 104961009B
- Authority
- CN
- China
- Prior art keywords
- elevator
- ladder
- floor
- head
- machine vision
- 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
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B1/00—Control systems of elevators in general
- B66B1/02—Control systems without regulation, i.e. without retroactive action
- B66B1/06—Control systems without regulation, i.e. without retroactive action electric
Landscapes
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Elevator Control (AREA)
- Indicating And Signalling Devices For Elevators (AREA)
Abstract
The invention discloses a kind of many elevator in parallel operation control method for coordinating based on machine vision and system, comprise the following steps: the image of camera collection inside and outside elevator is converted into gray-scale map, and carries out binary conversion treatment；Carrying out HOG feature extraction, obtain characteristics of human body's sample, construct two Fuzzy neural classifiers, utilize the two Fuzzy neural classifiers trained that it is carried out two-value judgement, the human body ruled out by grader carries out statistics and is added, and draws number inside and outside ladder；By to the demographics inside and outside elevator, consider each floor and wait residual capacity and distance time ladder position in ladder number, each elevator ladder, use traffic signal coordination that elevator is run and be scheduling.The present invention effectively reduces the lifting that elevator is unnecessary, reduces the abrasion of elevator, extends elevator service life, and reduces the waiting time of boarding personnel, and on the basis of energy-conservation, the life making people is more convenient.
Description
Technical field
The present invention relates to a kind of many elevator in parallel operation control method for coordinating based on machine vision and system.
Background technology
Along with the development in city, the requirement of living standard is improved constantly by people, and the intellectuality of building is the most important.No matter
It is that elevator has become one of requisite vehicles in residential quarter or at office block.
And along with the increase of people's current density, elevator can be taken the most efficiently for the ease of people, building has many
Portion's elevator in parallel operation.When people are when waiting elevator, often it can be allowed to connect together simultaneously by several elevators in parallel to save time
Receiving instruction, which portion takes advantage of soon which elevator.But this way is while facilitating some people, also increase other passenger-in-elevators
Waiting time, and increase the abrasion of elevator, make the elevator life-span reduce, fault increases.
Therefore, it is necessary to carry out the research and development of many elevators parallel connection coordinated operation control method and device, to reach to reduce elevator
Operation energy consumption, personnel's waiting time, increase the effect in elevator life-span.
On the other hand, along with the reduction of the cost of photographic head, it is widely applied in building and elevator.Pass through elevator
The video image of inside and outside photographic head shooting carries out the security monitoring of elevator operation and has been obtained for extensively applying, but the most not yet
See the research how using these video images to realize in terms of many elevator in parallel operation coordinate control.
The basis realizing the coordination control of many elevator in parallel operation is to carry out demographics according to the image of shooting inside and outside ladder to send out
Existing, present stage, the process carrying out demographics based on video image has Hough loop truss algorithm, Harr method etc..But due to ladder
Inside and outside video image often definition is low and is highly prone to the impact of the interference factors such as light.In this case, often use
Personnel's method of counting often precision based on video image of rule is the highest.
Summary of the invention
The present invention is to solve the problems referred to above, it is proposed that a kind of many elevator in parallel operation based on machine vision are coordinated to control
Method and system, the present invention carries out image procossing to the video image that the video camera inside and outside elevator shoots, and uses Hog feature extraction
Algorithm and two Fuzzy neural network classifiers carry out accurate metering to personnel amount inside and outside ladder；Count situation according to personnel, combine
Conjunction considers that residual capacity and the distance of distance time ladder position in ladder number, each elevator ladder waited by each floor, uses traffic signal coordination
Elevator is run and is scheduling, it is achieved ViewSonic's ladder, reduce elevator energy consumption and abrasion；Communicated by CAN, it is achieved many
The coordination of elevator in parallel operation, makes elevator Effec-tive Function at any time.
To achieve these goals, the present invention adopts the following technical scheme that
A kind of many elevator in parallel operation control method for coordinating based on machine vision, comprises the following steps:
(1) image of camera collection inside and outside elevator is converted into gray-scale map, carries out binary conversion treatment, and it is special to carry out HOG
Levy extraction；
(2) according to obtaining characteristics of human body's sample, construct two Fuzzy neural classifiers, utilize the two Fuzzy god trained
Categorized device carries out the two-value judgement of head and non-head to it, and the human body ruled out by grader carries out statistics and is added, draws
The inside and outside number of ladder；
(3) by the demographics inside and outside elevator, consider each floor and wait residual capacity in ladder number, each elevator ladder
And distance time ladder position, use traffic signal coordination that elevator is run and be scheduling；
(4) communicate, it is achieved the coordination that elevator runs.
In described step (1), method particularly includes: first the coloured picture of photographic head shooting inside and outside elevator is transformed into gray-scale map,
And use Gamma correction method to be standardized；On the basis of gray-scale map, calculate each pixel gradient magnitude and gradient direction comes
Capture profile information, is divided into cell factory by original image, adds up the rectangular histogram of each cell factory；Cell factory is combined into
Big block, normalized gradient rectangular histogram in block；The histogram vectors that all pieces interior be combined into a big Hog feature to
Amount.
Described step (1) method particularly includes: first the coloured picture of photographic head shooting inside and outside elevator is transformed into
Gray-scale map, and use Gamma correction method to be standardized；Each pixel (x, y) place's ladder is calculated on the basis of gray-scale map
Degree size And gradient directionCapturing profile information, wherein (x y) represents input picture to H respectively
Middle pixel (x, y) pixel value at place；Original image is divided into cell factory, adds up the rectangular histogram of each cell factory；Thin
Born of the same parents' unit is combined into big block, normalized gradient rectangular histogram in block；The histogram vectors that all pieces interior is combined into one greatly
Hog characteristic vector, just obtained all feature X=(x of human body head and non-head1,x2,…xn)T, for for two patterns
Stick with paste neural classifier the classification learning of human body head with non-head is used.
In described step (2), concrete grammar includes:
(2-1) training sample set with the number of people Yu inhuman labeling head is chosen；
(2-2) parameter of stochastic generation two Fuzzy membership function layer, and defeated according to training sample set computation rule layer
Go out matrix；
(2-3) optimal value of rules layer and the interval weight vector sum training sample set estimation of output layer is set, exports two
The input/output model of Fuzzy neural classifier.
In described step (2-1), choose one with the training sample set of the number of people Yu inhuman labeling head, be designated asWherein Xi=(xi1,xi2,…xin)TFor i-th sample characteristics, ti∈{0,
1}, 1 is expressed as people's head contour, and 0 is inhuman head contour, and the class label of N number of sample is combined into a vector T=[t1,…,tN
]T。
In described step (2-2), the parameter of stochastic generation two Fuzzy membership function layer, and calculate according to training sample set
The output matrix of rules layer:
WhereinK=1 ..., M,WithIt is respectively the kth type-2 fuzzy sets divided for jth feature to closeMembership function up and down.
In described step (2-3), rules layer with the interval weight vector of output layer is
Optimal value according to training sample set estimation β isWherein H+Moore-Penrose generalized inverse for output matrix H
Matrix, can be to human body head and non-head feature X=(x1,x2,…xn)TRealize two Fuzzy neural classifiers defeated of classification
Entering output model is
In described step (3), its method particularly includes:
(3-1) running status of every layer and each elevator, each elevator and the people of each floor in statistics building are marked
Number and residual capacity；
(3-2) essential state data of current elevator is obtained, according to elevator operation, the inside and outside demographics of ladder and residue
Capacity coordinates group's ladder.
Described step (3-1) is particularly as follows: the floor that can run is total to N shell, and elevator number is S, and the basic status of elevator i is (ki,
Oi,Fi), wherein FiFor elevator i place floor, ki{ 1,0 ,-1} are running status to ∈, and 1 represents ascending for elevator, and 0 represents that elevator stops
Leaning on ,-1 represents that elevator is descending, Oi=Ui-EiFor residual capacity, E thereiniIt is existing people in the ladder obtained by Vision Builder for Automated Inspection
Number, UiFor elevator i heap(ed) capacity, floor FjEssential state data be (Kj,Ej), wherein EjIt is that this layer waits ladder number, KjFor this
Ladder state set waited by floor, forIn one, 1 indicates that personnel are up, and-1 indicates personnel descending.
In described step (3-2), its method particularly includes: according to all elevator operations, each floor is waited ladder number etc. and is entered
Row coordinates group's ladder, as a example by elevator i sends ladder, (1) if now Oi=Ui, in elevator, unmanned boarding, makes ki=0, i.e. elevator stops fortune
OK；If detecting, upward signal makes ki=1, if downstream signal being detected, make ki=-1, elevator brings into operation；If Oi≠Ui, this
Time elevator be in running status；When elevator i runs, according to ki∈KjWhether, determine and its up-downgoing state consistency
Time ladder floor, and select the floor F nearest with itj, this floor wait ladder number EjFurther determine whether to this floor group electricity
Ladder i；
(2) suppose to floor FjSend elevator i1',i2',…im', if group's ladder quantity is inadequate, i.e. meet conditionAndTime then send elevator i to floor Fj；Continue to hold in return step (1) after sending ladder
OK, when sending enough elevator i1',i'2,…i'mTo floor FjTime, if FjIt is top of building N-1 layer or bottom inverse
Two layers, elevator then returns step (1) and continues executing with；Otherwise at KjMiddle rejecting kiAfter proceed optimizing next time.
In described step (4), elevator main controller is positioned at elevator(lift) machine room, receives the video image letter in each floor, car
Breath, installs floor controller near the elevator panel of each floor, and floor controller is led to master controller by CAN
News, what floor controller was responsible for receiving master controller sends ladder order, controls group's ladder, complete to call elevator.
Many elevator in parallel operation coordinated control system based on said method, including camera system, Hog feature extraction mould
The inside and outside personnel's counting module of block, ladder, multi-parallel elevator intelligent group ladder module, communication module based on CAN and elevator controlling
System；
Wherein, described camera system includes multiple photographic head, and photographic head is respectively arranged in elevator and every floor elevator
Mouthful, collector's information is transferred to Hog characteristic extracting module；
Described Hog characteristic extracting module, is used for utilizing Hog feature extraction algorithm extraction HOG feature as descriptor, draws
The characteristic vector of image is described；
Personnel's counting module inside and outside described ladder, is used for building and training two Fuzzy neural network classifiers, utilizes training
Good grader identifies head and non-head image, counts number inside and outside ladder；
Described multi-parallel elevator intelligent group ladder module, for counting situation according to personnel, considers each floor and waits ladder people
In elevator ladder several, each, residual capacity and distance wait the distance of ladder position, use traffic signal coordination to run elevator and are scheduling,
Realize ViewSonic's ladder, by communication module based on CAN, group's ladder instruction is transferred to apparatus for controlling elevator；
Described apparatus for controlling elevator, for performing the elevator operating instruction that multi-parallel elevator intelligent group ladder module is assigned, control
The operation of elevator processed.
The invention have the benefit that
(1) a kind of method that the present invention proposes new accurate metering, by the Hog feature extraction algorithm in machine learning
And two Fuzzy neural network classifier personnel amount inside and outside ladder has been carried out accurate metering；
(2) counting situation according to personnel, when waiting ladder personnel and pressing pectus elevator, master controller is by considering each building
Layer waits residual capacity and the distance of distance time ladder position in ladder number, each elevator ladder, uses traffic signal coordination to run elevator
It is scheduling, it is achieved ViewSonic's ladder, reduces elevator energy consumption and abrasion, save the waiting time of people；
(3) communicated by CAN, it is achieved the coordination of many elevator in parallel operation, owing to the ladder of sending of the present invention realizes
Circuit is to be connected in parallel with floor panel button；Both of which can realize elevator is sent ladder application；The control strategy of the present invention
Do not affect the operation of original elevator.
Accompanying drawing explanation
Fig. 1 be the present invention based on machine vision two Fuzzy neural network classifier schematic diagram；
Fig. 2 is the two Fuzzy neutral net schematic diagrams of the present invention；
Fig. 3 is the coordination group ladder flow chart of the present invention；
Fig. 4 is that the intelligence group ladder of the present invention controls topological diagram；
Fig. 5 is that the ladder of sending of the present invention realizes circuit diagram；
Fig. 6 is the present invention and floor panel interface circuitry schematic diagram；
Fig. 7 is that the multi-parallel elevator of the present invention runs system general diagram.
Detailed description of the invention:
The invention will be further described with embodiment below in conjunction with the accompanying drawings.
As it is shown in fig. 7, a kind of building many elevator in parallel operation control method for coordinating based on machine vision and device, right
Elevator paired running technology before has carried out substantial amounts of improvement, is the succession to it and development.This invention to elevator inside and outside
The video image of video camera shooting carries out image procossing, uses Hog feature extraction algorithm and two Fuzzy neural network classifiers
Personnel amount inside and outside ladder is carried out accurate metering；Count situation according to personnel, consider each floor and wait ladder number, each elevator ladder
Interior residual capacity and distance wait the distance of ladder position, use traffic signal coordination to run elevator and are scheduling, it is achieved ViewSonic
Ladder, reduces elevator energy consumption and abrasion；Communicated by CAN, it is achieved the coordination of many elevator in parallel operation, make elevator in office
When wait all Effec-tive Function.
Coordinate to control for realizing the many elevator in parallel operation of building, present invention is primarily based on three below module:
Module 1: the inside and outside personnel's counting module of ladder based on Hog feature extraction algorithm and two Fuzzy neural network classifiers
In ladder and the photographic head installed above of every floor lift port, the video image arrived according to camera collection, utilize
Hog feature extraction algorithm extraction HOG feature, as descriptor, draws the characteristic vector describing image, then trains two Fuzzies
Neural network classifier, and identify head and non-head image with the grader trained, count number inside and outside ladder.Algorithm
Specific as follows:
1, the step of personnel HOG feature extraction inside and outside elevator:
First the coloured picture of photographic head shooting inside and outside elevator is transformed into gray-scale map, and uses Gamma school
Execute and be standardized；Each pixel (x, y) place's gradient magnitude is calculated on the basis of gray-scale map And gradient directionCapturing profile information, wherein (x y) represents in input picture H respectively
Pixel (x, y) pixel value at place；Original image is divided into cell factory, adds up the rectangular histogram of each cell factory；Cell
Unit is combined into big block, normalized gradient rectangular histogram in block；The histogram vectors that all pieces interior be combined into one big
Hog characteristic vector, has just obtained all feature X=(x of human body head and non-head1,x2,…xn)T, for for two Fuzzies
The classification learning of human body head with non-head is used by neural classifier.
, illustrate as outline meanwhile, use Gamma correction method to be standardized arranging a kind of choosing of simply technical staff
Selecting, it is clear that on basic correction method choice, those skilled in the art need not pay creative work and completes corresponding
Conversion.
2, according to the sample of the number of people Yu inhuman labeling head, the structure of two Fuzzy neural classifiers is carried out:
Two Fuzzy neural classifiers use the identification of the two Fuzzy neural fusion numbers of people and non-number of people sample, this point
Class device structure, can be effective as it is shown in figure 1, be made up of input layer, two Fuzzy membership function layers, rules layer and output layer four layers altogether
In conjunction with ability and the self-learning capability of neutral net of two fuzzy systems anti-noise jammings, obtain the classification knot of superior performance
Really.
Following method is used to build for the number of people two Fuzzy neural classifiers with the non-number of people of classifying:
(1) first, choose one with the training sample set of the number of people Yu inhuman labeling head, be designated as
Wherein Xi=(xi1,xi2,…xin)TSpecial for i-th sample
Levy, ti{ 0,1}, 1 is expressed as people's head contour to ∈, and 0 is inhuman head contour.The class label of N number of sample is combined into a vector T=
[t1,…,tN]T；
(2) parameter of stochastic generation two Fuzzy membership function layer, and according to the output of training sample set computation rule layer
Matrix
WhereinK=1 ..., M,
WithIt is respectively the kth type-2 fuzzy sets divided for jth feature to closeMembership function up and down.
(3) rules layer with the interval weight vector of output layer isAccording to instruction
The optimal value practicing sample set estimation β isWherein H+Moore-Penrose generalized inverse matrix for output matrix H.
(4) thus, can be to human body head and non-head feature X=(x1,x2,…xn)TTwo Fuzzies realizing classification are neural
The input/output model of grader is
3, counting.The ladder inside and outside video image input shot by photographic head, is covered by the rectangular window less than image to be detected
Lid constantly slides in image, extracts the characteristic vector of overlay area pixel, utilizes the two Fuzzy nerve classification trained
Device carries out the two-value judgement of head and non-head to it, and the human body ruled out by grader carries out statistics and is added, draws inside and outside ladder
Number.
Module 2: multi-parallel elevator intelligent based on machine vision group ladder module
This module counts situation according to personnel, consider each floor wait ladder number, each elevator ladder in residual capacity and away from
From the distance of time ladder position, use traffic signal coordination that elevator is run and be scheduling, it is achieved ViewSonic's ladder, reduction elevator energy consumption
And abrasion.
Assuming the floor N shell altogether that building elevator can run, elevator number is S.The basic status of elevator i is (ki,Oi,Fi),
Wherein FiFor elevator i place floor, ki{ 1,0 ,-1} are running status to ∈, and 1 represents ascending for elevator, and 0 represents elevator parking ,-1 table
Show that elevator is descending, Oi=Ui-EiFor residual capacity, E thereiniIt is existing people in the ladder obtained by Vision Builder for Automated Inspection in module 1
Number, UiFor elevator i heap(ed) capacity.Floor FjEssential state data be (Kj,Ej), wherein EjIt is that this layer waits ladder number, KjFor this
Ladder state set waited by floor, forIn one, 1 indicates that personnel are up, and-1 indicates personnel descending.
According to all elevator operations, each floor time ladder number etc. coordinates sends ladder.As a example by elevator i sends ladder, tool
Body algorithm is as follows:
Obtain elevator i current essential state data { ki,Oi,Fi}。
(1) if now Oi=Ui, in elevator, unmanned boarding, makes ki=0, i.e. elevator is out of service；If upward signal being detected
Make ki=1, if downstream signal being detected, make ki=-1, elevator brings into operation.If Oi≠Ui, elevator now is in running shape
In state.When elevator i runs, according to ki∈KjWhether, determine the time ladder floor with its up-downgoing state consistency, and select with
Its nearest floor Fj, machine vision module calculate this floor and wait ladder number EjFurther determine whether to this floor group electricity
Ladder i.
(2) by with master controller communication, it is assumed that to floor FjSend elevator i1',i'2,…i'mIf group's ladder quantity is not
Enough, i.e. meet conditionAndTime then send elevator i to floor Fj.Returning after sending ladder
Mastery routine continues executing with.When sending enough elevator i1',i'2,…i'mTo floor FjTime, if FjBe top of building N-1 layer or
Layer second from the bottom bottom person, elevator then returns mastery routine and continues executing with；Otherwise at KjMiddle rejecting kiAfter proceed to seek next time
Excellent.
Also, it should be appreciated that, utilize the conversion of other condition modes, as running status use a, b, c} represent, and etc. simple
The exchange of status data expression way, should belong to what those skilled in the art were readily apparent that.
Module 3: communication-cooperation based on CAN
This module is used for intelligence group based on the machine vision ladder hardware circuit realizing being proposed.
Intelligence group ladder system mainly includes master controller, floor controller, sends ladder to realize circuit and CAN.Elevator
Master controller is positioned at elevator(lift) machine room, receives the video image information in each floor, car, intelligent coordinated through algorithm statistical analysis
Group's ladder.Floor controller is installed near the elevator panel of each floor.Floor controller is led to master controller by CAN
News.
The overall topological structure of intelligence group ladder is as shown in Figure 4.
Also, it should be appreciated that, in existing elevator controlling mode, using drives to control is one of them, therefore, and can
To be understood by, it is used herein control circuit and controls elevator simply one selection, and other select, including software program control
System, Control, PLC control, Single-chip Controlling etc. are the alternative that those skilled in the art are readily apparent that.
What floor controller was responsible for receiving master controller sends ladder order, controls send ladder to realize hardware circuit, completes calling electric
Ladder.
Being implemented as of intelligence group's ladder:
Such as, after the ladder order of a certain floor group assigned by master controller, the floor controller of corresponding floor receives order, defeated
Go out high level to close to port Port1, relay adhesive, S1.Complete a call request.
Floor controller realizes the hardware circuit of group's ladder as shown in Figure 5.This circuit is driven two by Phototube Coupling, relay
Divide and constitute.Master controller is assigned after sending ladder instruction, and floor controller PORT port output high level, relay closes.Complete one
The ladder application of secondary group.This circuit employs photoelectric isolation method, preferably eliminates the on-the-spot interference impact on floor controller.Energy
Ladder instruction sent by enough effective master controllers that performs.
Relay contact S1 is to be connected in parallel with the button of floor panel.The call of the present invention can be realized by the Guan Bi of S1
Application.After taking this mode, the control of the present invention does not affect the operation of original electric life controller, can realize calling electricity
Ladder.The paired running that i.e. control method of the present invention and the original controller of elevator can be good.Relay contacts is pressed with floor panel
Bonded as shown in Figure 6.
Although the detailed description of the invention of the present invention is described by the above-mentioned accompanying drawing that combines, but not the present invention is protected model
The restriction enclosed, one of ordinary skill in the art should be understood that on the basis of technical scheme, and those skilled in the art are not
Need to pay various amendments or deformation that creative work can make still within protection scope of the present invention.
Claims (10)
1. many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that: comprise the following steps:
(1) image of camera collection inside and outside elevator is converted into gray-scale map, carries out binary conversion treatment, and carry out HOG feature and carry
Take；
(2) according to obtaining characteristics of human body's sample, construct two Fuzzy neural classifiers, utilize the two Fuzzy nerves trained to divide
Class device carries out the two-value judgement of head and non-head to it, and the human body ruled out by grader carries out statistics and is added, in drawing ladder
Outer number；
(3) by the demographics inside and outside elevator, consider each floor wait in ladder number, each elevator ladder residual capacity and away from
From waiting ladder position, use traffic signal coordination that elevator is run and be scheduling；
(4) master controller sends group's ladder order to floor controller, controls to send ladder to realize circuit and completes elevator-calling, it is achieved elevator
The coordination run；
Described floor controller is communicated with master controller by CAN, described send ladder realize circuit and floor panel by
Key is connected in parallel.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
In described step (1), step includes: first the coloured picture of photographic head shooting inside and outside elevator is transformed into gray-scale map, and uses Gamma
Correction method is standardized；Each pixel gradient magnitude and gradient direction is calculated to capture profile letter on the basis of gray-scale map
Breath, is divided into cell factory by original image, adds up the rectangular histogram of each cell factory；Cell factory is combined into big block, block
Interior normalized gradient rectangular histogram；The histogram vectors that all pieces interior is combined into a big Hog characteristic vector.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, its
Feature is: described step (1) method particularly includes: first the coloured picture of photographic head shooting inside and outside elevator is transformed into gray-scale map,
And use Gamma correction method to be standardized；Each pixel (x, y) place's gradient magnitude is calculated on the basis of gray-scale mapAnd gradient directionCapturing profile information, wherein (x y) represents in input picture H respectively
Pixel (x, y) pixel value at place；Original image is divided into cell factory, adds up the rectangular histogram of each cell factory；Cell
Unit is combined into big block, normalized gradient rectangular histogram in block；The histogram vectors that all pieces interior be combined into one big
Hog characteristic vector, has just obtained all feature X=(x of human body head and non-head1,x2,…xn)T, for for two Fuzzies
The classification learning of human body head with non-head is used by neural classifier.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
In described step (2), concrete grammar includes:
(2-1) training sample set with the number of people Yu inhuman labeling head is chosen；
(2-2) parameter of stochastic generation two Fuzzy membership function layer, and according to the output square of training sample set computation rule layer
Battle array；
(2-3) optimal value of rules layer and the interval weight vector sum training sample set estimation of output layer is set, exports two patterns
Stick with paste the input/output model of neural classifier.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
In described step (2-1), choose one with the training sample set of the number of people Yu inhuman labeling head, be designated as
Wherein Xi=(xi1,xi2,…xin)TFor i-th sample characteristics, ti∈
{ 0,1}, 1 is expressed as people's head contour, and 0 is inhuman head contour, and the class label of N number of sample is combined into a vector T=[t1...,
tN]T。
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
In described step (2-2), the parameter of stochastic generation two Fuzzy membership function layer, and according to training sample set computation rule layer
Output matrix:
Wherein WithIt is respectively the kth type-2 fuzzy sets divided for jth feature to closeMembership function up and down.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
In described step (2-3), rules layer with the interval weight vector of output layer isRoot
According to the optimal value of training sample set estimation β it isWherein H+Moore-Penrose generalized inverse square for output matrix H
Battle array, can be to human body head and non-head feature X=(x1,x2,…xn)TRealize the input of two Fuzzy neural classifiers of classification
Output model is
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
In described step (3), its method particularly includes:
(3-1) mark the running status of every layer and each elevator, in statistics building each elevator and the number of each floor and
Residual capacity；
(3-2) essential state data of current elevator is obtained, according to elevator operation, the inside and outside demographics of ladder and residual capacity
Coordinate group's ladder.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, is characterized in that:
Described step (3-1) is particularly as follows: the floor that can run is total to N shell, and elevator number is S, and the basic status of elevator i is (ki,Oi,Fi), its
Middle FiFor elevator i place floor, ki{ 1,0 ,-1} are running status to ∈, and 1 represents ascending for elevator, and 0 represents elevator parking, and-1 represents
Elevator is descending, Oi=Ui-EiFor residual capacity, E thereiniIt is existing number in the ladder obtained by Vision Builder for Automated Inspection, UiFor electricity
Ladder i heap(ed) capacity, floor FjEssential state data be (Kj,Ej), wherein EjIt is that this layer waits ladder number, KjScalariform is waited for this floor
State collection, for{ 1}, {-1}, { in-1,1}, 1 indicates that personnel are up, and-1 indicates personnel descending.
A kind of many elevator in parallel operation control method for coordinating based on machine vision, its feature
It is: in described step (3-2), its method particularly includes: according to all elevator operations, each floor time ladder number coordinates sends
Ladder, to elevator i group ladder, (1) if now Oi=Ui, in elevator, unmanned boarding, makes ki=0, i.e. elevator is out of service；If detecting
Upward signal makes ki=1, if downstream signal being detected, make ki=-1, elevator brings into operation；If Oi≠Ui, elevator now is place
In running status；When elevator i runs, according to ki∈KjWhether, determine the time ladder floor with its up-downgoing state consistency,
And select the floor F nearest with itj, this floor wait ladder number EjFurther determine whether to send elevator i to this floor；
(2) suppose to floor FjSend elevator i'1, i'2,…i'mIf group's ladder quantity is inadequate, i.e. meets conditionAndTime then send elevator i to floor Fj；Continue to hold in return step (1) after sending ladder
OK, when sending enough elevator i'1,i'2,…i'mTo floor FjTime, if FjIt is top of building N-1 layer or bottom inverse
Two layers, elevator then returns step (1) and continues executing with；Otherwise at KjMiddle rejecting kiAfter proceed optimizing next time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510279503.1A CN104961009B (en) | 2015-05-27 | 2015-05-27 | Many elevator in parallel operation control method for coordinating based on machine vision and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510279503.1A CN104961009B (en) | 2015-05-27 | 2015-05-27 | Many elevator in parallel operation control method for coordinating based on machine vision and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104961009A CN104961009A (en) | 2015-10-07 |
| CN104961009B true CN104961009B (en) | 2016-10-05 |
Family
ID=54215145
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510279503.1A Active CN104961009B (en) | 2015-05-27 | 2015-05-27 | Many elevator in parallel operation control method for coordinating based on machine vision and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104961009B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106241533B (en) * | 2016-06-28 | 2018-10-30 | 西安特种设备检验检测院 | Elevator occupant's comprehensive safety intelligent control method based on machine vision |
| CN106241534B (en) * | 2016-06-28 | 2018-12-07 | 西安特种设备检验检测院 | More people's boarding abnormal movement intelligent control methods |
| CN106976766B (en) * | 2016-08-23 | 2020-05-26 | 深圳达实智能股份有限公司 | Elevator dispatching method and device |
| CN107273852A (en) * | 2017-06-16 | 2017-10-20 | 华南理工大学 | Escalator floor plates object and passenger behavior detection algorithm based on machine vision |
| CN108217352B (en) * | 2018-01-19 | 2019-11-05 | 深圳禾思众成科技有限公司 | A kind of intelligent elevator scheduling system |
| CN108764468A (en) * | 2018-05-03 | 2018-11-06 | 中国科学院计算技术研究所 | Artificial neural network processor for intelligent recognition |
| CN108675071B (en) * | 2018-05-03 | 2020-01-17 | 中国科学院计算技术研究所 | Cloud cooperative intelligent chip based on artificial neural network processor |
| CN108545561B (en) * | 2018-05-11 | 2020-09-22 | 金昱西 | Elevator control method and elevator control system |
| CN108657896A (en) * | 2018-06-12 | 2018-10-16 | 牛东阳 | The safety monitoring assembly of elevator rope |
| CN109179101A (en) * | 2018-09-07 | 2019-01-11 | 平安科技（深圳）有限公司 | Elevator control method, device, computer equipment and computer readable storage medium |
| CN109626150A (en) * | 2018-11-14 | 2019-04-16 | 深圳壹账通智能科技有限公司 | Elevator concocting method and system |
| CN110040586A (en) * | 2019-04-11 | 2019-07-23 | 哈尔滨理工大学 | A kind of Elevator group control method based on recognition of face |
| CN111724289A (en) * | 2020-06-24 | 2020-09-29 | 山东建筑大学 | Environmental protection equipment identification method and system based on time sequence |
| CN111747247B (en) * | 2020-07-01 | 2022-10-28 | 广州赛特智能科技有限公司 | Method for taking elevator by robot |
| CN113697620A (en) * | 2021-09-27 | 2021-11-26 | 湖南桅灯智能科技有限公司 | Elevator group control method and device based on visual identification |
| CN115303901B (en) * | 2022-08-05 | 2024-03-08 | 北京航空航天大学 | Elevator traffic flow identification method based on computer vision |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002302348A (en) * | 2001-04-04 | 2002-10-18 | Mitsubishi Electric Corp | Elevator control device |
| CN100595121C (en) * | 2007-12-10 | 2010-03-24 | 中国科学院计算技术研究所 | Target place optimized dispatching method and system |
| CN102126653B (en) * | 2011-01-06 | 2013-09-11 | 南京航空航天大学 | Intelligent Multiple -elevator system based on vision |
-
2015
- 2015-05-27 CN CN201510279503.1A patent/CN104961009B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104961009A (en) | 2015-10-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104961009B (en) | Many elevator in parallel operation control method for coordinating based on machine vision and system | |
| CN106966298B (en) | Assembled architecture intelligence hanging method based on machine vision and system | |
| CN106586738B (en) | A kind of elevator with multiple compartments dispatching method of view-based access control model detection | |
| CN109292579B (en) | Elevator system, image recognition method and operation control method | |
| CN105384015A (en) | Elevator control system based on human face recognition and intelligent recommendation | |
| CN205328387U (en) | Elevator control system based on face identification recommends with intelligence | |
| CN100595121C (en) | Target place optimized dispatching method and system | |
| CN107187966B (en) | A kind of multi-section elevator control method and system based on number testing result | |
| CN113682908B (en) | Intelligent scheduling method based on deep learning | |
| CN106553941A (en) | A kind of intelligent elevator group control method and system | |
| CN110963209A (en) | Garbage sorting device and method based on deep reinforcement learning | |
| CN102009879A (en) | Elevator automatic keying control system and method, face model training system and method | |
| CN108217352A (en) | A kind of intelligent elevator dispatches system | |
| WO2020238661A1 (en) | Elevator scheduling method, device, computer apparatus, and storage medium | |
| CN107628490A (en) | The recognition of face Intelligent elevator runtime and operation method of large-scale office building | |
| CN103425884B (en) | A kind of fuzzy prediction method of subway transfer channel traffic state | |
| CN107703937A (en) | Automatic Guided Vehicle system and its conflict evading method based on convolutional neural networks | |
| CN109045676A (en) | A kind of Chinese chess identification learning algorithm and the robot intelligence dynamicization System and method for based on the algorithm | |
| CN110415267A (en) | A kind of online thermal infrared target identification device of low-power consumption and working method | |
| CN106933585A (en) | A kind of self-adapting multi-channel interface system of selection under distributed cloud environment | |
| CN109829390A (en) | A kind of elevator intelligent scheduling system and method based on deep learning | |
| CN115147488A (en) | Workpiece pose estimation method based on intensive prediction and grasping system | |
| Chou et al. | Improving elevator dynamic control policies based on energy and demand visibility | |
| CN108675071B (en) | Cloud cooperative intelligent chip based on artificial neural network processor | |
| CN112149588B (en) | Intelligent elevator dispatching method based on pedestrian attitude estimation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20171201 Address after: 300202 Xiang year square 1-2-1004 of Dongjiang road and Neijiang Road, Hexi District, Tianjin Patentee after: The Mahdi (Tianjin) Engineering Design Institute Co. Ltd. Address before: Feng Ming Licheng District Lingang Development Zone, 250101 Ji'nan Road, Shandong Province Patentee before: Shandong Construction University |
|
| TR01 | Transfer of patent right |