CN110217558B - Resettable material conveying device based on intelligent machine vision - Google Patents

Resettable material conveying device based on intelligent machine vision Download PDF

Info

Publication number
CN110217558B
CN110217558B CN201910506117.XA CN201910506117A CN110217558B CN 110217558 B CN110217558 B CN 110217558B CN 201910506117 A CN201910506117 A CN 201910506117A CN 110217558 B CN110217558 B CN 110217558B
Authority
CN
China
Prior art keywords
layer
plate
node
impurities
ring
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
Application number
CN201910506117.XA
Other languages
Chinese (zh)
Other versions
CN110217558A (en
Inventor
张振东
张运国
张志江
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Anyang Xinju Environmental Protection Equipment Manufacturing Co ltd
Original Assignee
Anyang Xinju Environmental Protection Equipment Manufacturing Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Anyang Xinju Environmental Protection Equipment Manufacturing Co ltd filed Critical Anyang Xinju Environmental Protection Equipment Manufacturing Co ltd
Priority to CN201910506117.XA priority Critical patent/CN110217558B/en
Publication of CN110217558A publication Critical patent/CN110217558A/en
Application granted granted Critical
Publication of CN110217558B publication Critical patent/CN110217558B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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 invention discloses a chain plate type material conveying device based on machine vision, which comprises: the device comprises a chain plate type conveyor belt, an impurity detection device and a control device, wherein the chain plate type conveyor belt is used for conveying materials possibly containing impurities to a target position and comprises a transmission chain, a material plate surface and a fixed shaft; the impurity detection device is arranged above the chain plate type conveyor belt to detect the materials on the chain plate type conveyor belt and judge whether the materials at the corresponding positions contain impurities or not; the control device is used for controlling the firing pin execution mechanism positioned below the surface of the material plate to press the button for turning the turnable plate at the corresponding position of the chain plate type conveyor belt according to the position information of the impurities and the impurity removal instruction, the locking mechanism fails, and the turnable plate at the corresponding position turns downwards under the action of gravity to remove the impurities at the position.

Description

Resettable material conveying device based on intelligent machine vision
Technical Field
The invention relates to the technical field of intelligent manufacturing, in particular to a chain plate type material conveying device based on machine vision.
Background
Machine vision is a branch of the rapid development in the fields of artificial intelligence and pattern recognition. In summary, machine vision is to use a machine device to replace human eyes and brains for measurement and judgment. The machine vision system converts the shot target into image signal by a vision collecting device (a video camera or a camera), transmits the image signal to a special image processing system to obtain the image information of the shot target, converts the image signal into abstract mathematical information required by an automatic identification and automatic control algorithm according to the information of pixel distribution, brightness, color and the like of the image, and the automatic identification and control system performs various operations on the signal to extract the characteristics of the target so as to control the field equipment to execute corresponding commands according to the judged result.
In chemical production, often meet through the powdered material that chain slat type conveyer belt needs to add to the reaction unit conveying, often contain the impurity of mixing in midway in the powdered material, need isolate the impurity in the powdered material before adding reaction unit, present practice is manual separation generally, and efficiency is lower.
Disclosure of Invention
The invention provides a chain plate type material conveying device based on machine vision, which is used for overcoming at least one technical problem in the prior art.
In order to achieve the above object, an embodiment of the present invention provides a chain slat type material conveying device based on machine vision, including:
chain slat type conveyer belt, impurity detection device, controlling means, wherein:
the chain plate type conveyor belt is used for conveying materials possibly containing impurities to a target position, and comprises a transmission chain, a material plate surface and fixed shafts, the transmission chain is a metal closed chain, mounting holes for fixing one end of each fixed shaft are formed in the transmission chains on two sides, and two ends of each fixed shaft are arranged on the transmission chains through the mounting holes; the material plate surface comprises a plurality of mutually independent turnover plates, the turnover plates arranged on the same fixed shaft are arranged in a central symmetry mode, and the bearing surface of each turnover plate is used for bearing powder materials; each turnable plate is provided with a support ring and a locking ring, the upward plane of the turnable plate is a bearing plane, the support ring and the locking ring are arranged on one side edge of the turnable plate parallel to the fixed shaft, and the support ring and the locking ring are fixedly arranged on the fixed shaft in a penetrating manner;
the supporting ring is a circular ring with an inner layer structure and an outer layer structure, the outer layer of the supporting ring is fixed on the edge of the turnover plate, the inner layer of the supporting ring is fixedly arranged on the fixed shaft in a penetrating mode, and a smooth friction surface which is sealed relative to the outside is arranged between the inner layer and the outer layer of the supporting ring;
the locking ring is a circular ring with an inner layer structure, a middle layer structure and an outer layer structure, the outer layer of the locking ring is fixedly arranged at the edge of the turnable plate, the inner ring of the outer layer of the locking ring is a toothed ring, the middle layer of the locking ring is a one-way controllable locking mechanism hidden between the inner layer and the outer layer, the inner layer of the locking ring is fixedly arranged on the fixed shaft in a penetrating mode, one side of the inner layer of the locking ring is provided with a button capable of enabling the locking mechanism to fail, when the locking mechanism fails, the outer layer of the locking ring rotates relative to the inner layer under the action of gravity, and the turnable plate is driven to rotate downwards to dump materials borne by the turnable plate to a waste material layer below the surface of the material plate;
impurity detection device sets up link plate formula conveyer belt top is in order to right material on the link plate formula conveyer belt detects and judges whether the material that corresponds the position contains impurity, specifically includes: shooting the material running to a target detection domain through a camera to obtain an image of the material to be detected; the processor module of the impurity detection device detects the image of the material to be detected according to a pre-constructed neural network detection model to judge whether the material in the image of the material to be detected contains impurities, and when the judgment result shows that the material contains impurities, the processor module acquires position information of the impurities according to mark information on the transmission chain and transverse information of the surface of the material, which are acquired from the image of the material to be detected, generates an impurity removal instruction, and sends the position information of the impurities and the impurity removal instruction to the control device;
the control device is used for controlling the firing pin execution mechanism located below the surface of the material plate to press the button for turning the turnable plate at the corresponding position of the chain plate type conveyor belt according to the position information of the impurities and the impurity removing instruction, the locking mechanism fails, and the turnable plate at the corresponding position turns downwards under the action of gravity to remove the impurities at the position.
Optionally, the neural network detection model includes an input layer, an output layer, and a hidden layer, and the excitation function between the layers is:
σ(x)＝max(-0.01x，x)-0.01，x∈R，
a first convolution layer is arranged behind the input layer, the convolution window of the first convolution layer is 3x3, and each node of the first convolution layer is connected with only 3x3 nodes at the corresponding position of the input layer; the weights of 3x3 connections corresponding to each node of the first convolutional layer are respectively defined as w1-1, w1-2, … and w1-9 in row-column order, and the weights of each node of the first convolutional layer at the corresponding position are the same;
a second convolutional layer is arranged behind the first convolutional layer, the convolution window of the second convolutional layer is 5x5, and each node of the second convolutional layer is connected with only 5x5 nodes at the corresponding position of the first convolutional layer; the weights of 5x5 connections corresponding to each node of the second convolutional layer are respectively defined as w2-1, w2-2, … and w2-15 in row-column order, and the weights of each node of the second convolutional layer at the corresponding position are the same;
a third convolutional layer is arranged behind the second convolutional layer, the convolution window of the third convolutional layer is 3x3, and each node of the third convolutional layer is connected with only 3x3 nodes at the corresponding position of the second convolutional layer; the weights of 3x3 connections corresponding to each node of the third convolutional layer are respectively defined as w3-1, w3-2, … and w3-9 in row-column order, and the weights of each node of the third convolutional layer at the corresponding position are the same;
a resampling layer is arranged behind the third convolution layer, and if the size of the third convolution layer is dxd, the resampling layer has a size of
Figure GDA0002373811070000041
Each node of the resampling layer is connected with 16 nodes corresponding to the node, and the weight of each connection is fixed
Figure GDA0002373811070000042
The hidden layer is positioned behind the resampling layer, the hidden layer has two layers which are two fully-connected hidden layers, and the number of nodes of each hidden layer is equal to the number of pixels of the rectangular feature output by the resampling layer;
the node number of the output layer is 1, the node number is used for indicating whether the input image contains impurities, in the training set, 0 indicates that the input image does not contain the impurities, and 1 indicates that the input image contains the impurities;
the training set of the neural network detection model comprises a plurality of groups of training data, and each group of training data comprises: and (3) taking the preprocessed sample images and the corresponding identification marks as training sets and inputting the training sets into the neural network at the same time, and training to obtain the weight of each layer of the neural network.
Optionally, a region of the chain plate type conveyor belt below the camera is set as a detection identification section, the detection identification section is provided with 4 optical markers, the detection identification section is respectively arranged at two sides of the conveyor belt, 2 optical markers are respectively arranged at each side, 4 optical markers form a rectangular region on the plane of the conveyor belt, and the rectangular region corresponds to an image determination region in the acquired image.
The application describes a machine vision impurity detection method based on a neural network, which comprises the steps of placing a material containing impurities on a conveyor belt, selecting a to-be-detected judgment area on the conveyor belt, and shooting the judgment area by using a camera; when the impurities enter the judgment area along with the conveyor belt, the neural network algorithm identifies that the images shot by the camera contain the impurities, and sends signals to the conveyor belt equipment and the impurity removal equipment; stopping conveying by the conveyor belt, starting the impurity removing equipment, and removing the materials and the impurities in the judging area; when the neural network algorithm detects that the camera judges that the impurity is not contained in the area any more, a restarting signal is sent to the conveyor belt equipment, and the conveyor belt is restarted.
The invention of the embodiment of the invention comprises the following steps:
1. in the embodiment of the invention, the powder materials on the plate chain type conveying belt are identified through machine vision, when the powder materials are detected to contain impurities, the corresponding turnable plates are controlled to turn over according to the positions of the impurities before the materials are conveyed to the target position, so that the materials containing the impurities are poured into a waste material layer below, and the impurities are prevented from being brought into the subsequent working procedures; meanwhile, the impurity removing process does not need manual intervention, and the automation level of the material conveying process is improved.
2. The material plate surface of the chain plate type conveyor belt in the embodiment of the invention is composed of a plurality of mutually independent turnable plates, each turnable plate is arranged on a fixed shaft through a support ring and a locking ring, the support ring is a circular ring with an inner layer structure and an outer layer structure, the outer layer of the support ring is fixed on the edge of the turnable plate, the inner layer of the support ring is fixedly arranged on the fixed shaft in a penetrating way, and a smooth friction surface which is sealed relative to the outside is arranged between the inner layer and the outer layer of the support ring; the locking ring is a circular ring with an inner layer structure, a middle layer structure and an outer layer structure, the outer layer of the locking ring is fixedly arranged at the edge of the turnable plate, the inner ring of the outer layer of the locking ring is a toothed ring, the middle layer of the locking ring is a one-way controllable locking mechanism hidden between the inner layer and the outer layer, the inner layer of the locking ring is fixedly arranged on the fixed shaft in a penetrating way, one side of the inner layer of the locking ring is provided with a button capable of disabling the locking mechanism, when a material on a certain turnable plate contains impurities, a firing pin executing mechanism positioned below the corresponding turnable plate is controlled to press the corresponding button, the outer layer of the locking ring rotates relative to the inner layer under the action of gravity to drive the corresponding turnable plate to rotate downwards so as to dump the material borne on the turnable plate into the waste material layer below, when the turnable structure is turned, due to the characteristics of powder materials, the material borne on the turnable, has little effect on the material on the surrounding invertable plates.
3. In the embodiment of the invention, the reset mechanism is arranged behind the screening section of the conveying belt, the reset mechanism is a smooth slope which gradually rises along the material conveying direction of the conveying belt and is used for helping the reset of the turned-down independent plate, and when the turned-down turnable plate runs to the position of the reset mechanism, the reset mechanism gradually resets under the contact action of the slope, so that the influence of the turned-down turnable plate on the materials on the surrounding plate surface is avoided.
4. In the embodiment of the invention, the neural network is introduced into the material conveying device, and the classification capability of machine vision is trained according to the characteristics of the powder material and the impurities frequently mixed in, so that the accuracy of automatically identifying the impurities in the powder material is improved, and the application of machine identification in chemical production is expanded.
In the embodiment, when utilizing the chain slat type conveyer belt based on machine vision to carry the powder material, can carry out the conveyer belt material that the camera shot the region and carry out impurity detection, if contain impurity in the material then confirm the board that can overturn at its place according to the positional information of impurity, through the corresponding board upset that can overturn of actuating mechanism control with the material that contains impurity topple over to the storage case of below (waiting to reach and filter through impurity and recycle the material after the volume of settlement), whole process does not need manual operation, the automation level and the efficiency of production have been improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic diagram of a chain slat conveyor belt configuration according to one embodiment of the present invention;
FIG. 2 is a partial schematic view of a drive chain according to one embodiment of the present invention;
FIG. 3a is a schematic view of an end face of a stationary shaft according to an embodiment of the present invention
FIG. 3b is a partial schematic view of a side of a stationary shaft according to one embodiment of the present invention;
FIG. 4 is a schematic view of the split joint of the independent plates on the board surface according to one embodiment of the present invention;
FIG. 5 is a schematic view of a self-contained panel construction according to one embodiment of the present invention;
FIG. 6 is a schematic view of a support ring structure according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a locking ring configuration according to an embodiment of the present invention, wherein the left side is a configuration diagram of the locking mechanism in effect, and the right side is a configuration diagram of the locking mechanism in failure;
FIG. 8a is a diagram illustrating the normal operation of an individual plate of a conveyor belt in accordance with an embodiment of the present invention;
FIG. 8b is a view of the conveyor belt with the individual segments rotated downwardly in accordance with one embodiment of the present invention;
fig. 9 a-9 e are schematic diagrams of a process of assisting the reposition of an independent plate by a smooth slope according to a common embodiment.
Description of reference numerals:
1-a transmission chain; 2-material board surface; 3, fixing a shaft; 4-direction of belt advance; 5-mounting holes on the transmission chain; 6-mounting holes on the end face of the fixed shaft; 7-installing a positioning mark on the locking ring; 8-outside the conveyor belt; 9-a reversible plate; 10-a support ring; 11-a locking ring; 12-a bearing plane; 14-supporting the ring outer layer; 15-support ring inner layer; 16-support ring friction face; 17-locking ring outer layer; 18-locking ring outer layer gear ring; 19-locking ring middle layer locking mechanism; 20-locking ring inner layer; 21-a locking ring button; 22-locking ring friction surface; 24-smooth slope.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
It is to be noted that the terms "comprises" and "comprising" and any variations thereof in the embodiments and drawings of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.
As shown in fig. 1 to 8b, a chain plate type material conveying device based on machine vision according to an embodiment of the present invention includes: chain slat type conveyer belt, impurity detection device, controlling means, wherein:
the chain plate type conveyor belt is used for conveying materials possibly containing impurities to a target position, and comprises a transmission chain 1, a material plate surface 2 and fixed shafts 3, wherein the transmission chain 1 is a metal closed chain, mounting holes for fixing one end of each fixed shaft 3 are formed in the transmission chains 1 on two sides, and two ends of each fixed shaft 3 are fixedly arranged on the transmission chains 1 through the mounting holes; the material plate surface 2 comprises a plurality of mutually independent turnover plates 9, the turnover plates arranged on the same fixed shaft 3 are arranged in a central symmetry mode, and the bearing surface of each turnover plate is used for bearing powder materials; each turnable plate 9 is provided with a support ring 10 and a locking ring 11, the upward plane of the turnable plate 9 is a bearing plane 12, the support ring 10 and the locking ring 11 are arranged on one side edge of the turnable plate 9 parallel to the fixed shaft 3, and the support ring 10 and the locking ring 11 are fixedly arranged on the fixed shaft 3 in a penetrating manner;
the supporting ring 10 is a circular ring with an inner layer structure and an outer layer structure, an outer layer 14 of the supporting ring 10 is fixed on the edge of the turnover plate 9, an inner layer 15 of the supporting ring 10 is fixedly arranged on the fixed shaft 3 in a penetrating manner, and a smooth friction surface 16 sealed relative to the outside is arranged between the inner layer and the outer layer of the supporting ring 10;
the locking ring 11 is a circular ring with an inner layer structure, a middle layer structure and an outer layer structure, the outer layer of the locking ring 11 is fixedly arranged at the edge of the turnover plate 9, the inner ring of the outer layer 17 of the locking ring 11 is a toothed ring 18, the middle layer of the locking ring 11 is a one-way controllable locking mechanism 19 hidden between the inner layer and the outer layer, the inner layer 20 of the locking ring 11 is fixedly arranged on the fixed shaft 3 in a penetrating manner, a button 21 capable of enabling the locking mechanism to fail is arranged on one side of the inner layer of the locking ring 11, and when the locking mechanism fails, the outer layer of the locking ring 11 rotates relative to the inner layer under the action of gravity to drive the turnover plate to rotate downwards so as to dump materials borne on the turnover plate to a waste material layer below the material plate surface 2;
impurity detection device sets up link plate formula conveyer belt top is in order to right material on the link plate formula conveyer belt detects and judges whether the material that corresponds the position contains impurity, specifically includes: shooting the material running to a target detection domain through a camera to obtain an image of the material to be detected; the processor module of the impurity detection device detects the image of the material to be detected according to a pre-constructed neural network detection model to judge whether the material in the image of the material to be detected contains impurities, and when the judgment result shows that the material contains impurities, the processor module acquires position information of the impurities according to the marking information on the transmission chain 1 and the transverse information of the material plate surface 2 acquired from the image of the material to be detected, generates an impurity removal instruction, and sends the position information of the impurities and the impurity removal instruction to the control device;
the control device is used for controlling the firing pin execution mechanism located below the material plate surface 2 to press the button for turning the turnable plate at the corresponding position of the chain plate type conveyor belt according to the position information of the impurities and the impurity removing instruction, the locking mechanism fails, and the turnable plate at the corresponding position turns downwards under the action of gravity to remove the impurities at the position.
In the above embodiment, the conveying surface of the conveying belt is formed by splicing a plurality of independent plates with the same structure, and the independent plates are mounted on the same fixed shaft, the independent plates are arranged on two sides of the fixed shaft in a central symmetry manner, and the bearing surfaces of the independent plates face the outer side 8 of the conveying belt.
The individual panels comprise a reversible plate 9, a support ring 10 and a locking ring 11. The turnover plate 9 is a rectangular flat plate, wherein one upward plane is a bearing plane 12, the support ring and the locking ring are arranged on the same edge of the turnover plate, and the fixing shaft penetrates through the support ring and the locking ring to fix the rectangular flat plate in a horizontal position.
The support ring 10 is a circular ring with an inner layer and an outer layer, the outer layer 14 is fixed on the edge of a bearing plane, the inner layer 15 is fixed on a fixed shaft, and a smooth friction surface 16 sealed relative to the outside is arranged between the inner layer and the outer layer.
The locking ring is an inner, middle and outer layer circular ring, the outer layer 17 of the locking ring is fixed on the edge of a bearing plane, the inner ring of the outer layer of the locking ring is a circle of tooth-shaped ring 18, the middle layer of the locking ring is a one-way controllable locking mechanism 19 hidden between the inner layer and the outer layer, the inner layer 20 of the locking ring is fixed on a fixed shaft, the locking ring is provided with a button 21 with a positioning mark, and a smooth friction surface 22 sealed relative to the outside is arranged between the inner layer and the outer layer.
When the locking mechanism is effective, the independent plate can rotate upwards and can be locked when rotating downwards. In the conveying section, the plate surface is fixed on the fixed shaft by the locking ring 11 and the support ring 10, and the independent plate is kept in a horizontally bearable posture under the action of the locking mechanism 19 in the locking ring. The conveyed materials are placed on the whole plate surface by other devices, and the fixed shaft is pulled by the chain to drive the plate surface to move forward, so that the materials are conveyed.
When the conveying belt passes through the identification area, the position of the sundries is identified through the identification device, the independent plate where the sundries are located is determined, and the identification device sends data of the independent plate where the sundries are located to the control device.
The control device looks up the database according to the data that the recognition device sent, finds the button on the lock ring inlayer that corresponding independent plate corresponds to through actuating mechanism (like the firing pin) trigger button, locking mechanism inefficacy in the lock ring this moment, and corresponding independent plate can overturn downwards under the action of gravity, and debris can drop to the collection department of below. After the sundries are removed, the actuating mechanism can release the button, the locking mechanism takes effect again, the actuating mechanism pushes the corresponding independent plate back to the original position, and the conveying belt restores the conveying state again.
In one embodiment, the neural network detection model comprises an input layer, an output layer and a hidden layer, and the excitation function between the layers is as follows:
σ(x)＝max(-0.01x，x)-0.01，x∈R，
a first convolution layer is arranged behind the input layer, the convolution window of the first convolution layer is 3x3, and each node of the first convolution layer is connected with only 3x3 nodes at the corresponding position of the input layer; the weights of 3x3 connections corresponding to each node of the first convolutional layer are respectively defined as w1-1, w1-2, … and w1-9 in row-column order, and the weights of each node of the first convolutional layer at the corresponding position are the same;
a second convolutional layer is arranged behind the first convolutional layer, the convolution window of the second convolutional layer is 5x5, and each node of the second convolutional layer is connected with only 5x5 nodes at the corresponding position of the first convolutional layer; the weights of 5x5 connections corresponding to each node of the second convolutional layer are respectively defined as w2-1, w2-2, … and w2-15 in row-column order, and the weights of each node of the second convolutional layer at the corresponding position are the same;
a third convolutional layer is arranged behind the second convolutional layer, the convolution window of the third convolutional layer is 3x3, and each node of the third convolutional layer is connected with only 3x3 nodes at the corresponding position of the second convolutional layer; the weights of 3x3 connections corresponding to each node of the third convolutional layer are respectively defined as w3-1, w3-2, … and w3-9 in row-column order, and the weights of each node of the third convolutional layer at the corresponding position are the same;
a resampling layer is arranged behind the third convolution layer, and if the size of the third convolution layer is dxd, the resampling layer has a size of
Figure GDA0002373811070000101
Each node of the resampling layer is connected with 16 nodes corresponding to the node, and the weight of each connection is fixed
Figure GDA0002373811070000102
The hidden layer is positioned behind the resampling layer, the hidden layer has two layers which are two fully-connected hidden layers, and the number of nodes of each hidden layer is equal to the number of pixels of the rectangular feature output by the resampling layer;
the node number of the output layer is 1, the node number is used for indicating whether the input image contains impurities, in the training set, 0 indicates that the input image does not contain the impurities, and 1 indicates that the input image contains the impurities;
the training set of the neural network detection model comprises a plurality of groups of training data, and each group of training data comprises: and (3) taking the preprocessed sample images and the corresponding identification marks as training sets and inputting the training sets into the neural network at the same time, and training to obtain the weight of each layer of the neural network.
In one implementation, a region of the chain plate type conveyor belt below the camera is set as a detection identification section, the detection identification section is provided with 4 optical marks which are respectively arranged at two sides of the conveyor belt, 2 optical marks are respectively arranged at each side, and 4 optical marks form a rectangular region on the plane of the conveyor belt, and the rectangular region corresponds to an image judgment region in the acquired image.
In one embodiment, the camera is mounted above the conveyor belt with a line of sight directed perpendicular to the plane of the conveyor belt, and should be able to completely and properly image the decision region on the conveyor belt. The complete and proper judgment means that the judgment area completely falls into the visual range of the camera and is positioned in the central zone of the visual range as much as possible so as to reduce errors caused by image distortion at the edge of the lens of the camera as much as possible, and the distance between the camera and the conveyor belt ensures that human eyes can clearly distinguish target objects and impurities in the image. A group of optical marks with special identification characteristics are arranged on the conveyor belt and are respectively arranged on two sides of the conveyor belt, two optical marks are arranged on each side, and then the four optical marks form a rectangular area on the plane of the conveyor belt, and the area is a judgment area. When the camera takes an image, the optical mark can be clearly identified in the image, so that a corresponding image judgment area is marked in the image.
In the above embodiment, the input image may also be preprocessed before inputting the image into the neural network. The pretreatment steps are as follows:
obtaining image pixel coordinates of the four optical marks according to pixel positions corresponding to the four optical mark features in the image, and drawing out a corresponding area DI of the judgment area in the image;
dividing the image area DI into several square sub-images, each sub-image having a side length of a multiple of 4 and not less than 16, wherein pixels at the edges which cannot form a square sub-image can be dropped;
and taking each obtained square subgraph as an input of the neural network.
And after the preprocessing is finished, entering a neural network classification step. The neural network classifier consists of an input layer, an output layer and a hidden layer, wherein each layer comprises a plurality of nodes called neurons, the neurons and the connection among the neurons form a neural network, and the network is determined by an excitation function, weights and the connection mode among the neurons.
For example, a single hidden layer neural network comprises three input layer nodes X1, X2, X3, one output layer node y, three hidden layer nodes n1, n2, n 3. It can be seen that if no stimulus function is added, each node is a linear combination of its predecessors.
In this case, the neural network classifier cannot accurately classify the nonlinearly separable data set. This problem can then be solved by adding the stimulus function σ (x) at the hidden layer and output layer nodes.
Therefore, the primary task of using neural network classifiers is to design appropriate excitation functions. In one embodiment of the invention, the excitation function is defined as follows:
σ(x)＝max(-0.01x，x)-0.01，x∈R
the above formula is equivalent to:
Figure GDA0002373811070000121
the excitation function is an improvement of a classical ReLU excitation function, and by designing the excitation function, the efficiency and the performance of a neural network can be effectively considered, so that a better identification effect is achieved, and the excitation function is also an innovation point of the embodiment of the invention. Experiment comparison shows that the accuracy of the neural network in the aspect of image recognition and the time consumption of a single image are obviously improved by the excitation function.
The connection weights between neurons are important parameters of the neural network, and these weights are obtained through training of the neural network. If the neurons in adjacent layers are combined in a full connection mode, namely a mode that any two neurons have one connection, and each connection has an independent weight, the number of parameters needing to be optimized by the neural network increases exponentially with the increase of the layer number,making optimization difficult to achieve. For example, if a 100 × 100 resolution image is used, the number of nodes in the input layer is 100 × 100 — 10 if a fully connected neural network is used4The nodes of the hidden layer also have 100 × 100 ═ 104Since any two nodes are connected, the weighted value will be 104*104＝108And (4) respectively. As can be seen from the example of only one hidden layer, it is very difficult to optimize the parameters of such data amount, and therefore, in the application described in the embodiment of the present invention, the network structure of the neural network also needs to be optimized.
The optimized neural network structure is constructed as follows:
after the input layer (original image, I), convolutional layer C1 is placed, and the convolution window is 3x3, that is, each node of convolutional layer C1 has connection with only 3x3 nodes at the corresponding position of the layer above (input layer I); the weights of the 3 × 3 connections are defined as w1-1, w1-2, …, w1-9 in row-column order, respectively, and the weight of the node on each convolutional layer C1 at the corresponding position is the same.
A convolutional layer C2 is placed behind a convolutional layer C1, the convolutional window is 5x5, and each node of the convolutional layer C2 is connected with only 5x5 nodes at the corresponding position of the upper layer (convolutional layer C1); the weights of the 5 × 5 connections are defined as w2-1, w2-2, …, w2-15 in row-column order, respectively, and the weight of the node on each convolutional layer C2 at the corresponding position is the same.
Convolutional layer C3 was placed behind convolutional layer C2 with a convolutional window of 3x3, and similarly, each node of convolutional layer C3 had connections only to 3x3 nodes at the corresponding positions of the layer above it (convolutional layer C2); the weights of the 3 × 3 connections are defined as w2-1, w2-2, …, w2-9 in row-column order, respectively, and the weight of the node on each convolutional layer C3 at the corresponding position is the same.
A resample layer C4 is placed behind the convolutional layer C3. if the convolutional layer C3 size is d × d, then the resample layer C4 size is d × d
Figure GDA0002373811070000131
Each node of the resampling layer C4 is connected to its corresponding 16 nodes, and the weight is fixed to 1/16. To pairAt a node in resample level C4, assuming its coordinates are (i, j) at resample level C4, its corresponding node in convolutional layer C3 is all 16 nodes whose coordinates are within the (4i,4j) and (4i +3,4j +3) rectangles of convolutional layer C3.
Two fully connected hidden layers C5 and C6 are placed behind the resample layer C4, with each layer having a number of nodes equal to the number of pixels of the rectangular feature output by the resample layer C4.
The number of nodes of the output layer O is 1, indicating whether or not the input image contains impurities. In the training set, 0 indicates no impurity, and 1 indicates impurity.
The embodiment of the invention adopts the optimized neural network, so that the recognition efficiency of the neural network can be effectively improved on the premise of ensuring the recognition rate, which is an innovative point in the embodiment of the invention.
In addition, as shown in fig. 9a to 9e, in an embodiment of the present specification, a reset mechanism is further provided, after the conveyor belt screens a section, a smooth slope 24 for assisting the reset of the turned-down independent plate is directly below the conveyor belt, and the length of the smooth slope is greater than the sum of the widths of two independent plate surfaces; the topmost end of the smooth slope can be provided with an elastic protective layer. Fig. 9 a-9 e are schematic diagrams of a process of a smooth slope assisted individual plate repositioning of a common embodiment, wherein the sequence is chronological from top to bottom.
Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
Those of ordinary skill in the art will understand that: modules in the devices in the embodiments may be distributed in the devices in the embodiments according to the description of the embodiments, or may be located in one or more devices different from the embodiments with corresponding changes. The modules of the above embodiments may be combined into one module, or further split into multiple sub-modules.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (3)

1. A chain slat type material transfer device based on machine vision, characterized in that includes: chain slat type conveyer belt, impurity detection device, controlling means, wherein:
the chain plate type conveyor belt is used for conveying materials possibly containing impurities to a target position, and comprises a transmission chain, a material plate surface and fixed shafts, the transmission chain is a metal closed chain, mounting holes for fixing one end of each fixed shaft are formed in the transmission chains on two sides, and two ends of each fixed shaft are arranged on the transmission chains through the mounting holes; the material plate surface comprises a plurality of mutually independent turnover plates, the turnover plates arranged on the same fixed shaft are arranged in a central symmetry mode, and the bearing surface of each turnover plate is used for bearing powder materials; each turnable plate is provided with a support ring and a locking ring, the upward plane of the turnable plate is a bearing plane, the support ring and the locking ring are arranged on one side edge of the turnable plate parallel to the fixed shaft, and the support ring and the locking ring are fixedly arranged on the fixed shaft in a penetrating manner;
the supporting ring is a circular ring with an inner layer structure and an outer layer structure, the outer layer of the supporting ring is fixed on the edge of the turnover plate, the inner layer of the supporting ring is fixedly arranged on the fixed shaft in a penetrating mode, and a smooth friction surface which is sealed relative to the outside is arranged between the inner layer and the outer layer of the supporting ring;
the locking ring is a circular ring with an inner layer structure, a middle layer structure and an outer layer structure, the outer layer of the locking ring is fixedly arranged at the edge of the turnable plate, the inner ring of the outer layer of the locking ring is a toothed ring, the middle layer of the locking ring is a one-way controllable locking mechanism hidden between the inner layer and the outer layer, the inner layer of the locking ring is fixedly arranged on the fixed shaft in a penetrating mode, one side of the inner layer of the locking ring is provided with a button capable of enabling the locking mechanism to fail, when the locking mechanism fails, the outer layer of the locking ring rotates relative to the inner layer under the action of gravity, and the turnable plate is driven to rotate downwards to dump materials borne by the turnable plate to a waste material layer below the surface of the material plate;
impurity detection device sets up link plate formula conveyer belt top is in order to right material on the link plate formula conveyer belt detects and judges whether the material that corresponds the position contains impurity, specifically includes: shooting the material running to a target detection domain through a camera to obtain an image of the material to be detected; the processor module of the impurity detection device detects the image of the material to be detected according to a pre-constructed neural network detection model to judge whether the material in the image of the material to be detected contains impurities, and when the judgment result shows that the material contains impurities, the processor module acquires position information of the impurities according to mark information on the transmission chain and transverse information of the surface of the material, which are acquired from the image of the material to be detected, generates an impurity removal instruction, and sends the position information of the impurities and the impurity removal instruction to the control device;
the control device is used for controlling the firing pin execution mechanism located below the surface of the material plate to press the button for turning the turnable plate at the corresponding position of the chain plate type conveyor belt according to the position information of the impurities and the impurity removing instruction, the locking mechanism fails, and the turnable plate at the corresponding position turns downwards under the action of gravity to remove the impurities at the position.
2. The chain slat type material conveying device of claim 1, wherein the neural network detection model comprises an input layer, an output layer and a hidden layer, and an excitation function between the layers is as follows:
σ(x)＝max(-0.01x，x)-0.01，x∈R，
a first convolution layer is arranged behind the input layer, the convolution window of the first convolution layer is 3x3, and each node of the first convolution layer is connected with only 3x3 nodes at the corresponding position of the input layer; the weights of 3x3 connections corresponding to each node of the first convolutional layer are respectively defined as w1-1, w1-2, … and w1-9 in row-column order, and the weights of each node of the first convolutional layer at the corresponding position are the same;
a second convolutional layer is arranged behind the first convolutional layer, the convolution window of the second convolutional layer is 5x5, and each node of the second convolutional layer is connected with only 5x5 nodes at the corresponding position of the first convolutional layer; the weights of 5x5 connections corresponding to each node of the second convolutional layer are respectively defined as w2-1, w2-2, … and w2-15 in row-column order, and the weights of each node of the second convolutional layer at the corresponding position are the same;
a third convolutional layer is arranged behind the second convolutional layer, the convolution window of the third convolutional layer is 3x3, and each node of the third convolutional layer is connected with only 3x3 nodes at the corresponding position of the second convolutional layer; the weights of 3x3 connections corresponding to each node of the third convolutional layer are respectively defined as w3-1, w3-2, … and w3-9 in row-column order, and the weights of each node of the third convolutional layer at the corresponding position are the same;
a resampling layer is arranged behind the third convolution layer, and if the size of the third convolution layer is dxd, the resampling layer has a size of
Figure FDA0002091897370000031
Each node of the resampling layer is connected with 16 nodes corresponding to the node, and the weight of each connection is fixed
Figure FDA0002091897370000032
The hidden layer is positioned behind the resampling layer, the hidden layer has two layers which are two fully-connected hidden layers, and the number of nodes of each hidden layer is equal to the number of pixels of the rectangular feature output by the resampling layer;
the node number of the output layer is 1, the node number is used for indicating whether the input image contains impurities, in the training set, 0 indicates that the input image does not contain the impurities, and 1 indicates that the input image contains the impurities;
the training set of the neural network detection model comprises a plurality of groups of training data, and each group of training data comprises: and (3) taking the preprocessed sample images and the corresponding identification marks as training sets and inputting the training sets into the neural network at the same time, and training to obtain the weight of each layer of the neural network.
3. The chain slat type material conveying device according to claim 1, wherein a region of the chain slat type conveyor belt below the camera is set as a detection identification section, the detection identification section is provided with 4 optical marks respectively disposed at two sides of the conveyor belt, 2 optical marks are respectively disposed at each side, and 4 optical marks form a rectangular region on the plane of the conveyor belt, and the rectangular region corresponds to an image determination region in the acquired image.
CN201910506117.XA 2019-06-12 2019-06-12 Resettable material conveying device based on intelligent machine vision Active CN110217558B (en)

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 CN110217558A (en) 2019-09-10
CN110217558B true 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)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Citations (8)

* Cited by examiner, † Cited by third party
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

Patent Citations (8)

* Cited by examiner, † Cited by third party
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

Also Published As

Publication number Publication date
CN110217558A (en) 2019-09-10

Similar Documents

Publication Publication Date Title
CN110217558B (en) Resettable material conveying device based on intelligent machine vision
Davis et al. The classification of construction waste material using a deep convolutional neural network
JP7132701B2 (en) Tire image recognition method and tire image recognition device
US11361418B2 (en) Transfer learning based capsule endoscopic images classification system and method thereof
JP5257274B2 (en) MOBILE BODY DETECTING DEVICE, MOBILE BODY DETECTING METHOD, AND COMPUTER PROGRAM
CN108038424B (en) Visual automatic detection method suitable for high-altitude operation
US20080089560A1 (en) Known face guided imaging method
WO2018123607A1 (en) Upward-facing marker, image processing device, image processing method, and program
Mustafa et al. Agricultural produce sorting and grading using support vector machines and fuzzy logic
JP2019028657A (en) Learned model for building region extraction
CN109271934A (en) System and method for Traffic Sign Recognition
JP2021026697A (en) Scaffold material counting device and method, and scaffold material management system and method
Opitz et al. Object recognition and image segmentation: the Feature Analyst® approach
US20230274538A1 (en) Adaptable Automated Interpretation of Rapid Diagnostic Tests Using Self-Supervised Learning and Few-Shot Learning
Saifullah et al. Non-destructive egg fertility detection in incubation using SVM classifier based on GLCM parameters
CN110533629A (en) A kind of detection method and detection device of Bridge Crack
CN107671007B (en) Intelligent sorting system for detecting pearl shapes in batches based on machine vision
CN110288009B (en) Chain plate type material screening and conveying method based on neural network
JP2011240257A (en) Rotating oval body and entire-surface image inspection device of spherical body
RU2727858C2 (en) Training device, paper sheet identification device and paper sheet identification method
JP4812083B2 (en) Sorting method and apparatus for iriko etc.
Šćekić et al. Image-based parking occupancy detection using deep learning and faster r-cnn
Sanchez et al. Automatic character recognition in porcelain ware
KR20230082836A (en) System for supporting fruit sort based on artificial intelligence
Sigdel et al. Automatic classification of protein crystal images

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