CN110217558A - Material transmitting device is resetted based on intelligence machine vision - Google Patents
Material transmitting device is resetted based on intelligence machine vision Download PDFInfo
- Publication number
- CN110217558A CN110217558A CN201910506117.XA CN201910506117A CN110217558A CN 110217558 A CN110217558 A CN 110217558A CN 201910506117 A CN201910506117 A CN 201910506117A CN 110217558 A CN110217558 A CN 110217558A
- Authority
- CN
- China
- Prior art keywords
- layer
- impurity
- chain
- plate
- node
- 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.)
- Granted
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65G—TRANSPORT OR STORAGE DEVICES, e.g. CONVEYORS FOR LOADING OR TIPPING, SHOP CONVEYOR SYSTEMS OR PNEUMATIC TUBE CONVEYORS
- B65G43/00—Control devices, e.g. for safety, warning or fault-correcting
- B65G43/08—Control devices operated by article or material being fed, conveyed or discharged
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65G—TRANSPORT OR STORAGE DEVICES, e.g. CONVEYORS FOR LOADING OR TIPPING, SHOP CONVEYOR SYSTEMS OR PNEUMATIC TUBE CONVEYORS
- B65G47/00—Article or material-handling devices associated with conveyors; Methods employing such devices
- B65G47/34—Devices for discharging articles or materials from conveyor
- B65G47/38—Devices for discharging articles or materials from conveyor by dumping, tripping, or releasing load carriers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0014—Image feed-back for automatic industrial control, e.g. robot with camera
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65G—TRANSPORT OR STORAGE DEVICES, e.g. CONVEYORS FOR LOADING OR TIPPING, SHOP CONVEYOR SYSTEMS OR PNEUMATIC TUBE CONVEYORS
- B65G2203/00—Indexing code relating to control or detection of the articles or the load carriers during conveying
- B65G2203/04—Detection means
- B65G2203/041—Camera
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/06—Recognition of objects for industrial automation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Robotics (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of chain-plate type material conveying device based on machine vision, it include: chain-plate type conveyer belt, impurity detection device, control device, chain-plate type conveyer belt for will likely the material transmission containing impurity to target position, chain-plate type conveyer belt includes driving chain, material plate face and fixing axle；Impurity detection device is arranged above chain-plate type conveyer belt, to be detected and be judged whether the material of corresponding position contains impurity to the material on chain-plate type conveyer belt；Control device is used to be removed according to the location information and impurity of impurity and instruct, control is located at the button of the turnable plate overturning of the striker executing agency pressing chain-plate type conveyer belt corresponding position below material plate face, lockable mechanism failure, the turnable plate of corresponding position are downwardly turned under the effect of gravity to remove the impurity of the position.
Description
Technical field
The present invention relates to intelligent Manufacturing Technology fields, in particular to a kind of chain-plate type material based on machine vision
Transmission device.
Background technique
Machine vision is fast-developing branch of artificial intelligence and mode identification technology.Generally,
Machine vision is exactly to replace human eye and human brain with machinery equipment to measure and judge.NI Vision Builder for Automated Inspection is adopted by vision
Acquisition means (video camera or camera) will be ingested target and be converted into picture signal, send dedicated image processing system to, obtain
To the image information of target subject, the information such as pixel distribution and brightness, color further according to image, be transformed into automatic identification and
Abstract mathematics information required for automatic control algorithm, automatic identification and control system carry out various operations to these signals and come
Clarification of objective is extracted, and then field device is controlled according to the result of differentiation and executes corresponding order.
In Chemical Manufacture, frequently encounter by chain-plate type conveyer belt to reaction unit transmission need be added it is powdered
Material often needs to isolate in powder material in powder material containing the mixed impurity in midway before reaction unit is added
Impurity, present way are usually artificial separation, and efficiency is lower.
Summary of the invention
The present invention provides a kind of chain-plate type material conveying device based on machine vision, deposits in the prior art to overcome
At least one technical problem.
In order to achieve the above object, the embodiment of the present invention provides a kind of chain-plate type material transmission dress based on machine vision
It sets, comprising:
Chain-plate type conveyer belt, impurity detection device, control device, in which:
The chain-plate type conveyer belt for will likely the material transmission containing impurity to target position, the chain-plate type passes
Sending band includes driving chain, material plate face and fixing axle, and the driving chain is metallic closed chain, described in two sides
The mounting hole fixed for described fixing axle one end is equipped on driving chain, the both ends of each fixing axle pass through the peace
Hole is filled to be arranged on the driving chain；The material plate face includes multiple mutually independent turnable plates, is arranged in same root
Turnable plate in the fixing axle is centrosymmetric pattern arrangement, and the loading end of each turnable plate is for carrying powder
Last material；Each turnable plate is provided with support ring and lock ring, and the upward plane of the turnable plate is that carrying is flat
Face is provided with the support ring and the lock ring, institute in the turnable plate one side edge parallel with the fixing axle
It states support ring and lock ring fixation is threaded through in the fixing axle；
The support ring is the annulus of inside and outside double-layer structure, and the outer layer of the support ring is fixed on the turnable edges of boards
The internal layer fixation of edge, the support ring is threaded through in the fixing axle, is relatively outer between the internal layer and outer layer of the support ring
The smooth rubbing surface of boundary's sealing；
The lock ring is the annulus of interior, domestic and abroad three-decker, the outer layer of the lock ring be fixed at it is described can
Returning face plate edge, the outer layer inner ring of the lock ring are flute profile circle, the middle layer of the lock ring be hidden in internal layer and outer layer it
Between unidirectional controllable lockable mechanism, the lock ring internal layer fixation is threaded through in the fixing axle, the internal layer of the lock ring
Side is equipped with the button that can make lockable mechanism failure, when the lockable mechanism fails, the lock ring under the effect of gravity
Outer layer rotated relative to internal layer, drive the turnable plate to rotate down so that the material carried thereon is poured onto the material
Waste material layer below plate face；
The impurity detection device is arranged above the chain-plate type conveyer belt, on the chain-plate type conveyer belt
Material is detected and judges whether the material of corresponding position contains impurity, is specifically included: by camera to operation to target
The material of detecting domains is shot, and material image to be detected is obtained；The processor module of the impurity detection device is according in advance
The neural network detection model of building detects the material image to be detected to judge in the material image to be detected
Material whether contain impurity, when judging result is to contain impurity in material, obtained according to from the material image to be detected
The driving chain on mark information and the material plate face lateral acquisition of information impurity location information, and it is raw
It removes and instructs at impurity, the location information of the impurity and impurity removal instruction are sent to the control device；
The control device is used for according to the location information of the impurity and impurity removal instruction, and control is located at institute
State the striker executing agency below material plate face press the chain-plate type conveyer belt corresponding position turnable plate overturning by
Button, the lockable mechanism failure, the turnable plate of corresponding position are downwardly turned under the effect of gravity to remove the impurity of the position
It removes.
Optionally, the neural network detection model includes input layer, output layer and hidden layer, swashing between layers
Encourage function are as follows:
σ (x)=max (- 0.01x, x) -0.01, x ∈ R,
The first convolutional layer is provided with after the input layer, the convolution window of first convolutional layer is 3x3, described
Each node of first convolutional layer is only connected with 3x3 node of the input layer corresponding position；First convolutional layer is every
The weight of the corresponding 3x3 connection of a node is respectively defined as w1-1, w1-2 ..., w1-9 by the sequence of row-column, and described
Each node on first convolutional layer is all the same in the weight of corresponding position；
The second convolutional layer is provided with after first convolutional layer, the convolution window of second convolutional layer is 5x5, institute
The each node for stating the second convolutional layer is only connected with 5x5 node of the first convolutional layer corresponding position；The volume Two
The weight of the corresponding 5x5 connection of each node of lamination is respectively defined as w2-1, w2-2 ..., w2-15 by the sequence of row-column,
And each node of second convolutional layer is all the same in the weight of corresponding position；
Third convolutional layer is provided with after second convolutional layer, the convolution window of the third convolutional layer is 3x3,
Each node of the third convolutional layer is only connected with 3x3 node of the second convolutional layer corresponding position；The third
The weight of the corresponding 3x3 connection of each node of convolutional layer is respectively defined as w3-1, w3-2 ..., w3-9 by the sequence of row-column,
And each node on the third convolutional layer is all the same in the weight of corresponding position；
It is provided with resampling layer after the third convolutional layer, if the size of the third convolutional layer is d × d, then institute
State resampling layer having a size ofThe corresponding 16 nodes connection of each node of the resampling layer, and it is each
The weight of connection is fixed as
The hidden layer is located at after the resampling layer, and the hidden layer shares two layers, is that two full connections are hidden
Layer, the number of nodes of each hidden layer are equal to the pixel number of the rectangular characteristic of resampling layer output；
The number of nodes of the output layer is 1, and for indicating whether input picture contains impurity, in training set, 0 is indicated
Free from foreign meter, 1 indicates to contain impurity；
The training set of the neural network detection model includes multiple groups training data, training data described in every group include: through
Pretreated pretreatment sample image and corresponding identification label are crossed as training set while inputting neural network, training obtains
The weight of each layer of neural network.
Optionally, the region below the camera is located on the chain-plate type conveyer belt and is set as detection identification burst, it is described
Detection identification burst is provided with 4 optical markings, is respectively placed in conveyor belt two sides, each 2 of every side, 4 optical markings are passing
It send and constitutes a rectangular area in band plane, the rectangular area corresponds to the spectral discrimination region in acquired image.
This application describes a kind of machine vision method for detecting impurities neural network based, this method will contain impurity
Material be placed on conveyer belt, and on a moving belt select one at determinating area to be detected, with video camera be aligned determinating area
It is shot；When impurity enters determinating area with conveyer belt, contain in the image of neural network algorithm identification video camera shooting
Impurity, and send a signal to conveyor equipment and contaminant removal equipment；Conveyer belt stops transmission, and contaminant removal equipment starting will
Material and impurity in determinating area remove together；No longer include when neural network algorithm detects in video camera determinating area
When impurity, restarting signal is sent to conveyor equipment, conveyer belt restarting.
The inventive point of the embodiment of the present invention includes:
1, the powder material on plate chain type conveyer belt is identified by machine vision in the embodiment of the present invention, works as inspection
It can be controlled before material is sent to target position according to the location of impurity when measuring in powder material containing impurity
Corresponding turnable plate overturning avoids bringing impurity into subsequent work into so that impure material to be poured onto the waste material layer of lower section
Sequence；Meanwhile the exclusion process of impurity does not need manual intervention, improves the automatization level of material transmission process.
2, the material plate face of the chain-plate type conveyer belt in the embodiment of the present invention is by multiple mutually independent turnable board groups
At, and plate each may be reversed and be arranged in fixing axle by support ring and lock ring, support ring is inside and outside double-layer structure
Annulus, the outer layer of support ring are fixed on turnable edges of boards edge, and the internal layer fixation of support ring is threaded through in fixing axle, support ring
It is the smooth rubbing surface of relatively extraneous sealing between internal layer and outer layer；Lock ring is the annulus of interior, domestic and abroad three-decker, locking
The outer layer of ring is fixed at turnable edges of boards edge, and the outer layer inner ring of lock ring is flute profile circle, and the middle layer of lock ring is to be hidden in
The internal layer fixation of unidirectional controllable lockable mechanism between internal layer and outer layer, lock ring is threaded through in fixing axle, the internal layer of lock ring
Side is equipped with the button that lockable mechanism can be made to fail, and when the material on plate, which may be reversed, in some contains impurity, control, which is located at, to be corresponded to
The striker executing agency below plate may be reversed and press corresponding button, corresponding lockable mechanism fails, under the effect of gravity lock ring
Outer layer rotated relative to internal layer, drive corresponding turnable plate to rotate down so that the material carried thereon is poured onto the useless of lower section
In the bed of material, in overturning, due to the characteristic of powder material, the material carried on plate is only may be reversed in this by above-mentioned turnable structure
It is poured onto the waste material layer of lower section, the material on plate may be reversed to surrounding and have little effect.
3, in the embodiment of the present invention conveyer belt screen section after be provided with resetting-mechanism, the resetting structure be along
The smooth slope that conveyer belt transmission material direction gradually rises, is used to help the independent plate block reset turned down, when that turns down turns over
When rotating plate is run to resetting structure position, is gradually resetted under the contact action with slope, avoid the turnable plate turned down to week
Material on coaming plate face impacts.
4, neural network is introduced into material transmitting device in the embodiment of the present invention, is mixed into according to powder material and often
The characteristic of impurity is trained the classification capacity of machine vision, improves the accuracy of impurity in automatic identification powder material,
Application of the machine recognition in Chemical Manufacture is expanded.
In the above-described embodiments, when being sent using the chain-plate type conveyer belt based on machine vision to powder material progress bolt, i.e.,
Can to carry out video camera shooting area conveyer belt material carry out defects inspecting, if in material containing impurity if according to impurity
Location information determines the turnable plate where it, and plate overturning may be reversed accordingly by actuating mechanism controls will contain impurity
Material is poured onto the material storing box (recycling by contaminant filter to material after the amount of setting to be achieved) of lower section, entire mistake
Journey does not need manual operation, improves the gentle efficiency of Automated water of production.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will to embodiment or
Attached drawing needed to be used in the description of the prior art is briefly described, it should be apparent that, the accompanying drawings in the following description is only
Some embodiments of the present invention, for those of ordinary skill in the art, without creative efforts, also
Other drawings may be obtained according to these drawings without any creative labor.
Fig. 1 is the chain-plate type conveyer structure schematic diagram of one embodiment of the invention；
Fig. 2 is the driving chain partial schematic diagram of one embodiment of the invention；
Fig. 3 a is the fixing axle end view of one embodiment of the invention
Fig. 3 b is the partial schematic diagram of the fixing axle side of one embodiment of the invention；
Fig. 4 is that the plate face independence plate of one embodiment of the invention splices schematic diagram；
Fig. 5 is the independent plate structure schematic diagram of one embodiment of the invention；
Fig. 6 is the supporting ring structure schematic diagram of one embodiment of the invention；
Fig. 7 is the lock ring structural schematic diagram of one embodiment of the invention, wherein when left side is come into force for lockable mechanism
Structure chart, right side are structure chart when lockable mechanism fails；
Fig. 8 a is the state diagram when conveyer belt independence plate of one embodiment of the invention works normally；
Fig. 8 b is the state diagram when conveyer belt independence plate of one embodiment of the invention is rotated down；
Fig. 9 a- Fig. 9 e is the process schematic of the smooth slope auxiliary independent plate block reset of embodiment altogether.
Description of symbols:
1- driving chain；2- material plate face；3- fixing axle；4- conveyer belt direction of advance；Installation on 5- driving chain
Hole；The mounting hole of the fixed axial end of 6-；7- lock ring installs telltale mark；On the outside of 8- conveyer belt；Plate may be reversed in 9-；10- support
Ring；11- lock ring；12- load plane；14- support ring outer layer；15- support ring internal layer；16- support ring rubbing surface；17- locking
Ring outer layer；18- lock ring outer layer gear ring；19- lock ring middle layer lockable mechanism；20- lock ring internal layer；21- lock ring button；
22- lock ring rubbing surface；24- smooth slope.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art under that premise of not paying creative labor it is obtained it is all its
His embodiment, shall fall within the protection scope of the present invention.
It should be noted that term " includes " and " having " and their any changes in the embodiment of the present invention and attached drawing
Shape, it is intended that cover and non-exclusive include.Such as contain the process, method, system, product of a series of steps or units
Or equipment is not limited to listed step or unit, but optionally further comprising the step of not listing or unit, or can
Selection of land further includes the other step or units intrinsic for these process, methods, product or equipment.
As shown in FIG. 1 to FIG. 8 b, the chain-plate type material conveying device based on machine vision of one embodiment of the invention,
It include: chain-plate type conveyer belt, impurity detection device, control device, in which:
The chain-plate type conveyer belt for will likely the material transmission containing impurity to target position, the chain-plate type passes
Sending band includes driving chain 1, material plate face 2 and fixing axle 3, and the driving chain 1 is metallic closed chain, positioned at two sides
The mounting hole fixed for described 3 one end of fixing axle is equipped on the driving chain 1, the both ends of each fixing axle 3 pass through
The mounting hole is fixed on the driving chain 1；The material plate face 2 includes multiple mutually independent turnable plates
9, the turnable plate being arranged in fixing axle 3 described in same root is centrosymmetric pattern arrangement, and each turnable plate holds
Section is for carrying powder material；Each turnable plate includes that plate 9, support ring 10 and lock ring 11 may be reversed, described
It is load plane 12 that the upward plane of plate 9, which may be reversed, is set in the turnable plate 9 one side edge parallel with the fixing axle 3
It is equipped with the support ring 10 and the lock ring 11, the support ring 10 and the fixation of the lock ring 11 are threaded through the fixation
On axis 3；
The support ring 10 is the annulus of inside and outside double-layer structure, and the outer layer 14 of the support ring 10 is fixed on and described can turn over
The fixation of internal layer 15 at 9 edge of rotating plate, the support ring 10 is threaded through in the fixing axle 3, the internal layer of the support ring 10 and outer
It is the smooth rubbing surface 16 of relatively extraneous sealing between layer；
The lock ring 11 is the annulus of interior, domestic and abroad three-decker, and the outer layer of the lock ring 11 is fixed at institute
Turnable 9 edge of plate is stated, 17 inner ring of outer layer of the lock ring 11 is flute profile circle 18, and the middle layer of the lock ring 11 is hiding
The fixation of internal layer 20 of unidirectional controllable lockable mechanism 19 between internal layer and outer layer, the lock ring 11 is threaded through the fixation
On axis 3, the internal layer side of the lock ring 11 is equipped with the button 21 that can make the lockable mechanism failure, and the lockable mechanism loses
When effect, the outer layer of the lock ring 11 is rotated relative to internal layer under the effect of gravity, drives the turnable plate to rotate down to incite somebody to action
The material carried thereon is poured onto the waste material layer of 2 lower section of material plate face；
The impurity detection device is arranged above the chain-plate type conveyer belt, on the chain-plate type conveyer belt
Material is detected and judges whether the material of corresponding position contains impurity, is specifically included: by camera to operation to target
The material of detecting domains is shot, and material image to be detected is obtained；The processor module of the impurity detection device is according in advance
The neural network detection model of building detects the material image to be detected to judge in the material image to be detected
Material whether contain impurity, when judging result is to contain impurity in material, obtained according to from the material image to be detected
The driving chain 1 on mark information and the material plate face 2 lateral acquisition of information impurity location information, and
Impurity removal instruction is generated, the location information of the impurity and impurity removal instruction are sent to the control device；
The control device is used for according to the location information of the impurity and impurity removal instruction, and control is located at institute
State the lower section of material plate face 2 striker executing agency press the chain-plate type conveyer belt corresponding position the overturning of turnable plate by
Button, the lockable mechanism failure, the turnable plate of corresponding position are downwardly turned under the effect of gravity to remove the impurity of the position
It removes.
In the above-described embodiments, the transmission plate face of conveyer belt is spliced by multiple mutually isostructural independent plates, installation
In same root fixing axle, the independent plate fixing axle two sides be centrosymmetric pattern arrangement, loading end towards pass
Send the outside 8 of band.
The independent plate includes that plate 9, support ring 10 and lock ring 11 may be reversed.The turnable plate 9 is a rectangle
Plate, wherein a upward plane is load plane 12, the support ring and lock ring are arranged in the same of turnable plate
On edge, fixing axle is passed through from the support ring and lock ring, and rectangular plate is fixed on horizontal position.
Support ring 10 is a kind of inside and outside two layers of annulus, and the outer layer 14 is fixed on load plane edge, the internal layer
15 are fixed in fixing axle, are the smooth rubbing surface 16 of relatively extraneous sealing between ectonexine.
Lock ring is in a kind of, in, outer three layers of annulus, lock ring outer layer 17 is fixed on load plane edge, outside lock ring
Layer inner ring is a circle flute profile circle 18, and lock ring middle layer is a kind of unidirectional controllable lockable mechanism being hidden between internal layer and outer layer
19, lock ring internal layer 20 is fixed in fixing axle, and it is opposite between ectonexine that lock ring, which is equipped with the button 21 with telltale mark,
In the smooth rubbing surface 22 of external world's sealing.
When lockable mechanism comes into force, independent plate can be rotated up, and can be locked when rotating down.Transporting section, plate
Face is locked up ring 11 and support ring 10 is fixed in fixing axle, is acted on by the lockable mechanism 19 in lock ring, and independent plate is kept
The position and posture that level can carry.The material being transported is placed in entire plate face by other devices, is fixed by chain traction
Axis drives plate face to advance, to transport material.
When conveyer belt passes through identification region, the position where sundries is identified by identification device, where determining sundries
Independent plate, the data of the independent plate where identification device to control device transmission sundries.
The data inspection database that control device is sent according to identification device finds the accordingly corresponding locking of independent plate
Button on ring internal layer, and by executing agency (such as striker) trigger button, lockable mechanism fails in lock ring at this time, accordingly
Independent plate can downwardly turn under the effect of gravity, sundries can fall to lower section collection at.Sundries excludes work and completes
Afterwards, executing agency, which can send, holds button, and lockable mechanism comes into force again at this time, executing agency by corresponding independent plate pushed home,
Conveyer belt restores shipping situation again.
In one embodiment, the neural network detection model includes input layer, output layer and hidden layer, layer and layer it
Between excitation function are as follows:
σ (x)=max (- 0.01x, x) -0.01, x ∈ R,
The first convolutional layer is provided with after the input layer, the convolution window of first convolutional layer is 3x3, described
Each node of first convolutional layer is only connected with 3x3 node of the input layer corresponding position；First convolutional layer is every
The weight of the corresponding 3x3 connection of a node is respectively defined as w1-1, w1-2 ..., w1-9 by the sequence of row-column, and described
Each node on first convolutional layer is all the same in the weight of corresponding position；
The second convolutional layer is provided with after first convolutional layer, the convolution window of second convolutional layer is 5x5, institute
The each node for stating the second convolutional layer is only connected with 5x5 node of the first convolutional layer corresponding position；The volume Two
The weight of the corresponding 5x5 connection of each node of lamination is respectively defined as w2-1, w2-2 ..., w2-15 by the sequence of row-column,
And each node of second convolutional layer is all the same in the weight of corresponding position；
Third convolutional layer is provided with after second convolutional layer, the convolution window of the third convolutional layer is 3x3,
Each node of the third convolutional layer is only connected with 3x3 node of the second convolutional layer corresponding position；The third
The weight of the corresponding 3x3 connection of each node of convolutional layer is respectively defined as w3-1, w3-2 ..., w3-9 by the sequence of row-column,
And each node on the third convolutional layer is all the same in the weight of corresponding position；
It is provided with resampling layer after the third convolutional layer, if the size of the third convolutional layer is d × d, then institute
State resampling layer having a size ofThe corresponding 16 nodes connection of each node of the resampling layer, and it is each
The weight of connection is fixed as
The hidden layer is located at after the resampling layer, and the hidden layer shares two layers, is that two full connections are hidden
Layer, the number of nodes of each hidden layer are equal to the pixel number of the rectangular characteristic of resampling layer output；
The number of nodes of the output layer is 1, and for indicating whether input picture contains impurity, in training set, 0 is indicated
Free from foreign meter, 1 indicates to contain impurity；
The training set of the neural network detection model includes multiple groups training data, training data described in every group include: through
Pretreated pretreatment sample image and corresponding identification label are crossed as training set while inputting neural network, training obtains
The weight of each layer of neural network.
It is located at the region below the camera in a kind of implementation, on the chain-plate type conveyer belt and is set as detection and knows
Other section, the detection identification burst is provided with 4 optical markings, is respectively placed in conveyor belt two sides, each 2 of every side, 4 light
It learns label and constitutes a rectangular area in conveyer belt plane, the rectangular area corresponds to the spectral discrimination in acquired image
Region.
In one embodiment, video camera is mounted on above conveyer belt, and sight perpendicular alignmnet conveyer belt plane should energy
Enough determinating areas completely, suitably shot on conveyer belt.It is so-called completely, suitably to refer to that determinating area falls completely within video camera
Visual range in, and as far as possible be located at visual range central area, it is abnormal to reduce camera lens edge image as far as possible
Become bring error, should ensure that human eye can clearly tell the object and impurity in image at a distance from video camera and conveyer belt.
Setting has one group of optical markings of specific identification feature on a moving belt, is respectively placed in conveyor belt two sides, each two of every side,
Then four optical markings constitute a rectangular area in conveyer belt plane, which is determinating area.Video camera shooting
When image, optical markings can be identified clearly in the picture, to mark corresponding spectral discrimination region in the picture.
In the above-described embodiments, before image is inputted neural network, input picture can also be pre-processed.
Pre-treatment step is as follows:
According to the corresponding location of pixels of four optical markings features in image, the image slices of four optical markings are obtained
Plain coordinate marks the corresponding region DI of determinating area in the picture；
Image-region DI is divided into several square subgraphs, the multiple that the side length of each subgraph is 4, and not
Less than 16, the pixel that wherein edge cannot form square subgraph can be cast out；
Input by each square subgraph obtained, as neural network.
After pretreatment, into neural network classification step.The neural network classifier, by input layer, output layer
It is formed with hidden layer, each layer includes several nodes, and referred to as neuron, the connection between neuron, neuron constitutes mind
Through network, which is determined by the connection type between excitation function, weight and neuron.
For example, the neural network of a single hidden layer includes X1, X2, X3, one output node layers of three input layers
Y, three hiding node layer n1, n2, n3.As it can be seen that each node is its foreword node if being added without excitation function
Linear combination.
In this case, neural network classifier just can not accurately classify to the data set of Nonlinear separability.In
It is that can solve this problem by the way that excitation function σ (x) is added in hidden layer and output node layer.
It therefore, the use of the top priority of neural network classifier is exactly to design suitable excitation function.At the present invention one
In embodiment, it is as follows to define excitation function:
σ (x)=max (- 0.01x, x) -0.01, x ∈ R
Above formula is equivalent to:
The excitation function is an improvement of classics ReLU excitation function, can be effectively simultaneous by designing the excitation function
The efficiency and performance for caring for neural network, reach better recognition effect, this is also an innovative point of the embodiment of the present invention.It is logical
Cross the accuracy rate and single image time-consuming that Experimental comparison shows that above-mentioned excitation function makes neural network in terms of image recognition
There is apparent performance boost.
Connection weight between neuron is the important parameter of neural network, the training that these weights pass through neural network
It obtains.If being to have a connection between any two neuron in a manner of " full connection " between the neuron of adjacent layer
Mode combine, and every connection has an independent weight, then the neural network number of parameters that needs to optimize will be with
The number of plies increase and exponential increase, make optimization become difficult to achieve.For example, the image of a 100*100 resolution ratio, such as
Fruit uses the neural network of full connection type, and input layer number is 100*100=104, and the node of hidden layer also has 100*
100=104, due to there is connection between any two node, weighted value will have 104*104=108.Only one is hidden
Layer example can find out, optimize such data volume parameter be it is very difficult, therefore, described in the embodiment of the present invention
Application in, it is also desirable to the network structure of neural network is optimized.
The neural network structure for constructing optimization is as follows:
Convolutional layer C1 is placed afterwards in input layer (original image, I), and convolution window is 3x3, i.e. each of convolutional layer C1
3x3 node of the node only with one layer of (input layer I) corresponding position thereon has connection；The weight that this 3x3 is connected is by row-
The sequence of column is respectively defined as w1-1, w1-2 ..., w1-9, and the node on each convolutional layer C1 is in the weight of corresponding position
It is all the same.
Convolutional layer C2 is placed after convolutional layer C1, convolution window is 5x5, and each node of convolutional layer C2 is only and thereon
The 5x5 node of one layer of (convolutional layer C1) corresponding position has connection；The weight that this 5x5 is connected is distinguished by the sequence of row-column
It is defined as w2-1, w2-2 ..., w2-15, and the node on each convolutional layer C2 is all the same in the weight of corresponding position.
Convolutional layer C3 is placed after convolutional layer C2, convolution window is 3x3, and similarly, each node of convolutional layer C3 is only
There is connection with the 3x3 node of one layer of (convolutional layer C2) corresponding position thereon；The weight that this 3x3 is connected is suitable by row-column
Sequence is respectively defined as w2-1, w2-2 ..., w2-9, and the node on each convolutional layer C3 is homogeneous in the weight of corresponding position
Together.
A resampling layer C4 is placed after convolutional layer C3, if convolutional layer C3 size is d × d, resampling layer C4 ruler
It is very little to beThe corresponding 16 nodes connection of each node of resampling layer C4, and weight is fixed as 1/16.For
In some node of resampling layer C4, it is assumed that its coordinate is (i, j) in resampling layer C4, then its corresponding section in convolutional layer C3
Point is coordinate in all 16 nodes that convolutional layer C3 is in (4i, 4j) and (4i+3,4j+3) rectangular extent.
Two full connection hidden layer C5 and C6 are placed after resampling layer C4, it is defeated that every node layer number is equal to resampling layer C4
The pixel number of rectangular characteristic out.
Output layer O node number is 1, indicates whether input picture contains impurity.In training set, 0 indicates free from foreign meter, 1
Expression contains impurity.
Above-described embodiment can effectively promote nerve using the neural network after optimization under the premise of ensureing discrimination
The recognition efficiency of network, this is an innovative point in the embodiment of the present invention.
In addition, being additionally provided with resetting-mechanism as shown in Fig. 9 a- Fig. 9 e in one embodiment of this specification, transmitting
After band screening section, conveyer belt is arranged right below the smooth slope 24 for helping the independent plate block reset turned down, smooth
The length on slope is greater than the sum of the width of two independent plate faces；Smooth slope top is also provided with the protective layer of elasticity.Figure
9a- Fig. 9 e is the process schematic of the smooth slope auxiliary independent plate block reset of embodiment altogether, wherein from top to bottom temporally
Sequence sorts.
Those of ordinary skill in the art will appreciate that: attached drawing is the schematic diagram of one embodiment, module in attached drawing or
Process is not necessarily implemented necessary to the present invention.
Those of ordinary skill in the art will appreciate that: the module in device in embodiment can be described according to embodiment
It is distributed in the device of embodiment, corresponding change can also be carried out and be located in one or more devices different from the present embodiment.
The module of above-described embodiment can be merged into a module, can also be further split into multiple submodule.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；To the greatest extent
Present invention has been described in detail with reference to the aforementioned embodiments for pipe, those skilled in the art should understand that: it is still
It can modify to technical solution documented by previous embodiment or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of the embodiment of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (3)
1. a kind of chain-plate type material conveying device based on machine vision characterized by comprising chain-plate type conveyer belt, impurity
Detection device, control device, in which:
The chain-plate type conveyer belt for will likely the material transmission containing impurity to target position, the chain-plate type conveyer belt packet
Driving chain, material plate face and fixing axle are included, the driving chain is metallic closed chain, the driving chain positioned at two sides
On be equipped with the mounting hole fixed for described fixing axle one end, the both ends of each fixing axle are existed by mounting hole setting
On the driving chain；The material plate face includes multiple mutually independent turnable plates, and the fixing axle described in same root is arranged
On turnable plate be centrosymmetric pattern arrangement, the loading end of each turnable plate is for carrying powder material；Each
The turnable plate is provided with support ring and lock ring, and the upward plane of the turnable plate is load plane, described to may be reversed
The support ring and the lock ring, the support ring and the lock are provided in the plate one side edge parallel with the fixing axle
Only ring fixation is threaded through in the fixing axle；
The support ring is the annulus of inside and outside double-layer structure, and the outer layer of the support ring is fixed on the turnable edges of boards edge, institute
The internal layer fixation for stating support ring is threaded through in the fixing axle, is relatively extraneous sealing between the internal layer and outer layer of the support ring
Smooth rubbing surface；
The lock ring is the annulus of interior, domestic and abroad three-decker, and the outer layer of the lock ring is fixed at the turnable plate
The outer layer inner ring at edge, the lock ring is flute profile circle, and the middle layer of the lock ring is the list being hidden between internal layer and outer layer
To controllable lockable mechanism, the internal layer fixation of the lock ring is threaded through in the fixing axle, and the internal layer side of the lock ring is set
There is the button that can make lockable mechanism failure, when the lockable mechanism fails, the outer layer of the lock ring under the effect of gravity
It is rotated relative to internal layer, drives the turnable plate to rotate down so that the material carried thereon to be poured onto below the material plate face
Waste material layer；
The impurity detection device is arranged above the chain-plate type conveyer belt, with to the material on the chain-plate type conveyer belt into
Row detects and judges whether the material of corresponding position contains impurity, specifically includes: by camera to operation to target detection domain
Material shot, obtain material image to be detected；The processor module of the impurity detection device according to constructing in advance
Neural network detection model detects the material image to be detected to judge the material in the material image to be detected
Whether impurity is contained, when judging result is to contain impurity in material, according to obtaining from the material image to be detected
The location information of the lateral acquisition of information impurity of mark information and the material plate face on driving chain, and generate impurity and go
Except instruction, the location information of the impurity and impurity removal instruction are sent to the control device；
The control device is used for according to the location information of the impurity and impurity removal instruction, and control is located at the material
Striker executing agency below plate face presses the button of the turnable plate overturning of the chain-plate type conveyer belt corresponding position, the lock
Locking mechanism failure, the turnable plate of corresponding position are downwardly turned under the effect of gravity to remove the impurity of the position.
2. chain-plate type material conveying device according to claim 1, which is characterized in that the neural network detection model packet
Include input layer, output layer and hidden layer, excitation function between layers are as follows:
σ (x)=max (- 0.01x, x) -0.01, x ∈ R,
The first convolutional layer is provided with after the input layer, the convolution window of first convolutional layer is 3x3, the first volume
Each node of lamination is only connected with 3x3 node of the input layer corresponding position；The each node of first convolutional layer
The weight of corresponding 3x3 connection is respectively defined as w1-1, w1-2 ..., w1-9, and the first volume by the sequence of row-column
Each node on lamination is all the same in the weight of corresponding position；
It is provided with the second convolutional layer after first convolutional layer, the convolution window of second convolutional layer is 5x5, described second
Each node of convolutional layer is only connected with 5x5 node of the first convolutional layer corresponding position；Second convolutional layer is every
The weight of the corresponding 5x5 connection of a node is respectively defined as w2-1, w2-2 ..., w2-15 by the sequence of row-column, and described
Each node of second convolutional layer is all the same in the weight of corresponding position；
It is provided with third convolutional layer after second convolutional layer, the convolution window of the third convolutional layer is 3x3, described the
Each node of three convolutional layers is only connected with 3x3 node of the second convolutional layer corresponding position；The third convolutional layer
The weight of the corresponding 3x3 connection of each node is respectively defined as w3-1, w3-2 ..., w3-9, and institute by the sequence of row-column
The each node stated on third convolutional layer is all the same in the weight of corresponding position；
It is provided with resampling layer after the third convolutional layer, it is if the size of the third convolutional layer is d × d, then described heavy
Sample level having a size ofThe corresponding 16 nodes connection of each node of the resampling layer, and each connection
Weight is fixed as
The hidden layer is located at after the resampling layer, and the hidden layer shares two layers, is two full connection hidden layers, each
The number of nodes of hidden layer is equal to the pixel number of the rectangular characteristic of resampling layer output；
The number of nodes of the output layer is 1, and for indicating whether input picture contains impurity, in training set, 0 is indicated without miscellaneous
Matter, 1 indicates to contain impurity；
The training set of the neural network detection model includes multiple groups training data, and training data described in every group includes: by pre-
The pretreatment sample image of processing and corresponding identification label as training set while inputting neural network, and training obtains nerve net
The weight of each layer of network.
3. chain-plate type material conveying device according to claim 1, which is characterized in that be located on the chain-plate type conveyer belt
Region below the camera is set as detection identification burst, and the detection identification burst is provided with 4 optical markings, is respectively placed in biography
Band two sides are sent, each 2 of every side, 4 optical markings constitute a rectangular area, the rectangular area pair in conveyer belt plane
It should be in the spectral discrimination region in acquired image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910506117.XA CN110217558B (en) | 2019-06-12 | 2019-06-12 | Resettable material conveying device based on intelligent machine vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910506117.XA CN110217558B (en) | 2019-06-12 | 2019-06-12 | Resettable material conveying device based on intelligent machine vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110217558A true CN110217558A (en) | 2019-09-10 |
| CN110217558B CN110217558B (en) | 2020-04-10 |
Family
ID=67816621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910506117.XA Active CN110217558B (en) | 2019-06-12 | 2019-06-12 | Resettable material conveying device based on intelligent machine vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110217558B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111745742A (en) * | 2020-06-30 | 2020-10-09 | 广州聪慧青贸易有限公司 | Novel die-cutting machine |
| CN113895925A (en) * | 2021-09-28 | 2022-01-07 | 新沂市大自然木业有限公司 | Production transmission structure for intelligent wood processing |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB190615582A (en) * | 1906-07-10 | 1907-04-25 | Babcock & Wilcox Ltd | Improvements in Conveyors. |
| GB1238464A (en) * | 1967-11-14 | 1971-07-07 | ||
| US3735857A (en) * | 1970-01-30 | 1973-05-29 | K C S Charlottenlund Aasted | Mechanism for delivering goods from continuously operating chocolate moulding plants |
| DE202008017880U1 (en) * | 2008-05-15 | 2010-09-23 | MIAW YEOU METAL INDUSTRY CO., LTD., Ta-Ya Hsiang | Chain plate arrangement of a metal chip conveyor |
| CN202807697U (en) * | 2012-10-06 | 2013-03-20 | 黑龙江迪尔制药机械有限责任公司 | Rotating shaft turn-over plate delivery device used for tunnel type drying machine |
| CN103404950A (en) * | 2013-08-30 | 2013-11-27 | 山东理工大学 | Vegetable dehydrator with material clearing devices on lower portion |
| EP3192754B1 (en) * | 2016-01-13 | 2018-06-20 | Befesa Aluminio, S.L. | Metal plate conveyor for hot slag |
| CN208779887U (en) * | 2018-08-23 | 2019-04-23 | 姜少华 | Multi-layer panel-turnover foodstuff drying device |
-
2019
- 2019-06-12 CN CN201910506117.XA patent/CN110217558B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB190615582A (en) * | 1906-07-10 | 1907-04-25 | Babcock & Wilcox Ltd | Improvements in Conveyors. |
| GB1238464A (en) * | 1967-11-14 | 1971-07-07 | ||
| US3735857A (en) * | 1970-01-30 | 1973-05-29 | K C S Charlottenlund Aasted | Mechanism for delivering goods from continuously operating chocolate moulding plants |
| DE202008017880U1 (en) * | 2008-05-15 | 2010-09-23 | MIAW YEOU METAL INDUSTRY CO., LTD., Ta-Ya Hsiang | Chain plate arrangement of a metal chip conveyor |
| CN202807697U (en) * | 2012-10-06 | 2013-03-20 | 黑龙江迪尔制药机械有限责任公司 | Rotating shaft turn-over plate delivery device used for tunnel type drying machine |
| CN103404950A (en) * | 2013-08-30 | 2013-11-27 | 山东理工大学 | Vegetable dehydrator with material clearing devices on lower portion |
| EP3192754B1 (en) * | 2016-01-13 | 2018-06-20 | Befesa Aluminio, S.L. | Metal plate conveyor for hot slag |
| CN208779887U (en) * | 2018-08-23 | 2019-04-23 | 姜少华 | Multi-layer panel-turnover foodstuff drying device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111745742A (en) * | 2020-06-30 | 2020-10-09 | 广州聪慧青贸易有限公司 | Novel die-cutting machine |
| CN111745742B (en) * | 2020-06-30 | 2022-01-07 | 荣艺电子科技(重庆)有限公司 | Novel die-cutting machine |
| CN113895925A (en) * | 2021-09-28 | 2022-01-07 | 新沂市大自然木业有限公司 | Production transmission structure for intelligent wood processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110217558B (en) | 2020-04-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Akram et al. | Automatic detection of photovoltaic module defects in infrared images with isolated and develop-model transfer deep learning | |
| CN110217558A (en) | Material transmitting device is resetted based on intelligence machine vision | |
| Tilocca et al. | Detecting fabric defects with a neural network using two kinds of optical patterns | |
| CA2976769C (en) | Model-based methods and apparatus for classifying an interferent in specimens | |
| CN109461149A (en) | The intelligent checking system and method for lacquered surface defect | |
| WO2018188023A1 (en) | Methods and apparatus for determining label count during specimen characterization | |
| CN108109137A (en) | The Machine Vision Inspecting System and method of vehicle part | |
| CN107944504B (en) | Board recognition and machine learning method and device for board recognition and electronic equipment | |
| CA3030226A1 (en) | System and method for combined automatic and manual inspection | |
| CN107123107A (en) | Cloth defect inspection method based on neutral net deep learning | |
| CN108010025B (en) | Switch and indicator lamp positioning and state identification method of screen cabinet based on RCNN | |
| CN109271934A (en) | System and method for Traffic Sign Recognition | |
| CN107850761A (en) | Sub-micron wafer aligned | |
| CN109165541A (en) | Coding method for vehicle component in intelligent recognition rail traffic vehicles image | |
| EP3916637A1 (en) | Apparatus and method for detecting elements of an assembly | |
| JP6970550B2 (en) | Defect inspection system and defect inspection method | |
| CN106600552A (en) | Texture image completion method and device | |
| CN110288009A (en) | Chain-plate type material neural network based screens transfer approach | |
| CN103177435A (en) | Apple surface non-redundancy information image processing method based on machine vision | |
| AU2013314014B2 (en) | Moving Object Detection Method | |
| CN111104976B (en) | Blue algae coverage calculating method based on time sequence images | |
| Chetoui et al. | Object detection model-based quality inspection using a deep CNN | |
| Mandapaka et al. | Application of automated quality control in smart factories-a deep learning-based approach | |
| JP4812083B2 (en) | Sorting method and apparatus for iriko etc. | |
| CN111356895A (en) | Detection device and method for producing electronic device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |