CN108627520A - A kind of on-line detecting system and method for heterogeneous solid material presentation quality - Google Patents

A kind of on-line detecting system and method for heterogeneous solid material presentation quality Download PDF

Info

Publication number
CN108627520A
CN108627520A CN201810513734.8A CN201810513734A CN108627520A CN 108627520 A CN108627520 A CN 108627520A CN 201810513734 A CN201810513734 A CN 201810513734A CN 108627520 A CN108627520 A CN 108627520A
Authority
CN
China
Prior art keywords
image
data
detection
module
information
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.)
Pending
Application number
CN201810513734.8A
Other languages
Chinese (zh)
Inventor
李龙
陈洪
刘诚学
曹功武
冯绍杰
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.)
BEIJING INSPECT TECHNOLOGY Co Ltd
Original Assignee
BEIJING INSPECT TECHNOLOGY 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 BEIJING INSPECT TECHNOLOGY Co Ltd filed Critical BEIJING INSPECT TECHNOLOGY Co Ltd
Priority to CN201810513734.8A priority Critical patent/CN108627520A/en
Publication of CN108627520A publication Critical patent/CN108627520A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84Systems specially adapted for particular applications
    • G01N21/88Investigating the presence of flaws or contamination

Abstract

The invention discloses a kind of presentation quality on-line detecting systems of heterogeneous solid material, including:Several appearance quality detecting devices, defects detection module, deep learning module and control module;Appearance quality detecting device, the instruction for executing control module transmit heterogeneous solid material to image acquisition areas and carry out Image Acquisition, are additionally operable to the heterogeneous solid material that detection is completed being output to different regions;Defects detection module, for by whether including pixel grey scale pixel set jumpy in detection image to realize the defects detection on measured object surface, forming testing result;Deep learning module, for optimizing defects detection module by the adaptive learning of system;Heterogeneous solid material is transmitted to image acquisition areas, sends out instruction according to testing result and heterogeneous solid material is output to different regions by control module for sending out instruction.The present invention provides detection and information platform for on-line checking, improves detection efficiency and accuracy.

Description

A kind of on-line detecting system and method for heterogeneous solid material presentation quality
Technical field
The present invention relates to the production of heterogeneous solid material, the appearance quality detection technologies of manufacturing field, such as:Carbon block appearance is examined It surveys, the detection of mixed mud appearance, ceramic appearance detection etc., more particularly to a kind of online inspection of heterogeneous solid material presentation quality Examining system and method.
Background technology
The manufacturing of carbon block belongs to the manufacturing field of heterogeneous solid material, before roasting shove charge and after coming out of the stove Presentation quality inspection need to be carried out, the various unqualified carbon blocks such as crackle, unfilled corner, oxidation are discharged.To prevent unqualified carbon block from entering Next procedure, traditional method is that full-time employee is arranged to check carbon block presentation quality naked eyes direct-view in the production line for each enterprise, this On the one hand kind of inspection method increases cost of labor, efficiency is low, and when another aspect worker's work fatigue, which has, checks risk not in place, It reports by mistake;And working environment high temperature with high dust is unfavorable for worker's occupational health of production line;The inspection of traditional detection mode Measured data manually counts, and is supported without big data platform, causes statistical efficiency low, and it is inconvenient that historical data is consulted, and data value has Limit.Especially production scale enlargement and continuous production, online quality control are particularly urgent important.Presentation quality on-line checking Method at home and abroad can directly apply by the experience also without Engineering Projects or maturation or technology.Therefore presentation quality on-line checking side Method still belongs to blank, the problem of being 4.0 automated production urgent need to resolve of current industrial.
Invention content
It is an object of the present invention to solve to detect high labor cost at present, efficiency is low, and false drop rate is high, working environment high temperature Dusty is unfavorable for employee's occupational health;Data manually count, and are supported without big data platform, and efficiency is low, and historical data is consulted The problems such as inconvenience, data value is limited, retrospective is not strong.Realize the online presentation quality of industry internet of intelligent automation The system and method for detection meet country and enterprise's industrial upgrading general orientation.
To achieve the above object, the present invention proposes a kind of presentation quality on-line detecting system of heterogeneous solid material, institute The system of stating includes:Several appearance quality detecting devices, defects detection module, deep learning module and control module;
The appearance quality detecting device, the instruction for executing the control module transmit the heterogeneous solid material Material to image acquisition areas carries out Image Acquisition, is additionally operable to the heterogeneous solid material that detection is completed being output to different regions;
The defects detection module, the image for being acquired according to the appearance quality detecting device, passes through detection image In whether include pixel grey scale pixel set jumpy to realize the defects detection on measured object surface, form detection knot Fruit;
The deep learning module, for optimizing defects detection module by the adaptive learning of system;
The heterogeneous solid material is transmitted to described image acquisition zone, root by the control module for sending out instruction Instruction is sent out according to testing result, and heterogeneous solid material is output to different regions.
As a kind of improvement of above-mentioned apparatus, the detection device of the heterogeneous solid material presentation quality includes:Machinery Mechanism, position sensor, imaging mechanism and image capture module;
The mechanical mechanism is additionally operable to for conveying heterogeneous solid material to be detected to specified position according to inspection It surveys result and heterogeneous solid material to be detected is output to different regions;
The position sensor is mounted at the shooting work area position of described device, for being acquired in detection process The location information of detected material sends the instruction for starting shooting to imaging mechanism when detected material reaches designated position;
Appearance of the imaging mechanism for heterogeneous solid material is taken pictures, and image information is obtained;
Described image acquisition module, the image information for acquiring imaging mechanism, and it is sent to the defects detection module.
As a kind of improvement of above-mentioned apparatus, the defects detection module includes that sample image database is established unit and lacked Fall into detection unit;Sample image database establishes unit, for handling collected sample data, establishes defect sample Image data base;Defect detection unit, the image information for reading described image acquisition module in real time, with sample image data Sample in library is compared, in detection image whether include pixel grey scale pixel set jumpy to realize measured object The defects detection on surface, formed it is final whether qualified testing result.
As a kind of improvement of above-mentioned apparatus, the deep learning module includes adaptive learning unit and defects detection Optimize unit;
The adaptive learning unit is used to establish the feature neural network framework of sample different scale, by inputting sample The decomposition of the different scale characteristic information of this image obtains the image feature information pel of different scale;
The optimization unit of the defects detection is new for constantly inheriting and being added during detection image data accumulation Characteristic information pel, to obtain accurate defect sample image.
As a kind of improvement of above-mentioned apparatus, the control module includes host computer unit, the next controller unit and leads to Interrogate unit;
The host computer unit is used for, and the interaction with the next controller unit in time feeds back to down testing result Level controller cell processing;
The bottom controller unit controls the camera shooting, acquisition is non-for receiving the host computer unit instruction Solid material location information, the action for instructing the mechanical mechanism to execute instruction;
The communication unit, the interaction channel for establishing the host computer unit and the next controller unit.
As a kind of improvement of above-mentioned apparatus, the system also includes:
Data processing module, for all the image collected data in system to be handled and are classified, including it is newly-built Image essential information;Tabulation is established for various features;
Data memory module, it is right for all the image collected data of system to be stored and exported according to format The tabulation of various features is preserved;With
Data management module, the data for system are presented, including data are shown, data statistic analysis, the inquiry of data It is exported with outbound data.
As a kind of improvement of above-mentioned apparatus, the system also includes internet information management modules, including:
Operation instruction unit receives the information of the data management module output by enterprise's Ethernet and equipment state is believed Breath, provides basic data, to establish ERP enterprise resource plan management platforms;With
Information process unit collects the data management module for carrying out information data exchange by workshop Ethernet Creation data, the ERP that data informations at different levels and equipment state are uploaded to operation instruction unit by enterprise's Ethernet puts down Platform, as with external server interactive interface.
The invention also provides a kind of presentation qualities for the heterogeneous solid material realized based on above-mentioned system to examine online Detection method is surveyed, including:
Step 1) control module sends out send instructions to roller-way (1), and the roller-way (1) transmits heterogeneous solid according to instruction Material is to image acquisition areas;
Step 2) the heterogeneous solid material reaches image acquisition areas, and location information is sent to by displacement sensor (3) The control module,
After the step 3) control module receives location information, photographing instruction is sent out;
Step 4) the camera (3) executes photographing instruction, obtains image information and is sent to defect processing module;
Defect processing module described in step 5) reads defect sample image in real time from the sample image database The image information that information butt joint receives carries out defects detection, by judging that detection image includes pixel grey scale picture jumpy Element set forms testing result to realize the defects detection on measured object surface;
If step 6) testing result is zero defect, the next controller unit sends instruction, revolving platform to the revolving platform Carbon block is delivered to normal direction;Otherwise, the next controller unit sends instruction to the revolving platform, and revolving platform is by defective carbon Block is delivered to waste product area;
As a kind of improvement of the above method, the step 5) includes:
Step 5-1) from the data memory module defect sample image information is read in real time;
Step 5-2) mark sample image information in brightness change significantly point be defect point;
Step 5-3) from the appearance delection device sampled image information is read in real time;
Step 5-4) according to step 5-2) the apparent point of brightness change in sample digital picture, to step 5-3) sampling The significant changes of pixel grey scale in image compare;Carry out that the discontinuous, surface direction in depth is discontinuous, material property becomes The comparison of change;
Step 5-5) according to step 5-4) comparison, the surrounding pixel gray scale brought by measured object surface defect drastically becomes The pixel set of change judged, it is consistent significantly to put thresholding with brightness change in sample digital picture, judges measured object surface It is defective;Otherwise zero defect, to realize the defects detection on measured object surface.
As a kind of improvement of the above method, the method further includes:Step 7) carries out the adaptive learning optimization of system Defects detection module, specifically includes:
Step 7-1) basic defects detection is completed, image information is stored in the sample image database;
Step 7-2) according to step 7-1) the obtained image information of defects detection, establish the nerve net of sample different scale Network framework obtains the image feature information of different scale by the decomposition of the different scale characteristic information to input sample image Pel;
Step 7-3) according to step 7-2) the obtained image feature information pel of different scale, divide imaging surface figure Feature, and then identify different defect state, according to the different respective features of defect state and the general character, build surface image Defect characteristic model;
Step 7-4) according to step 7-3) the defect characteristic model established, each being detected on every production line being collected into The surface image information that device generates is classified, and data memory module storage to the sample image database is passed through;
Step 7-5) according to step 7-4) information of sample image database, extract surface image defect model feature sum number Corresponding detecting and alarm is established according to the characteristics of library;
Step 7-6) according to step 7-5) establish detecting and alarm, big data retrieval service is provided, figure is obtained based on retrieval As model data set;
Step 7-7) according to step 7-6) data acquisition system in constantly inherit and be added new characteristic information, carry out deep To make system continue to optimize, the result of optimization forms new sample pattern for degree study, is detected for real-time online.
Advantage of the invention is that:
1, system and method for the invention use industrial machine vision technique, automated control technology, data analysis, artificial The technologies such as intelligence, edge detection, deep learning improve the flexibility and the degree of automation of production, in the case where environmental condition is poor Artificial detection is substituted with machine vision, production efficiency and the degree of automation of production, and machine vision can be greatly improved It is integrated to be easily achieved data information;
2, system and method for the invention are using the industry internet production information system of intelligent automation, the system System and method can accomplish that the defects detection on basis, energy adaptive learning optimization constantly promote accuracy, accomplish that measurement result zero is missed Difference is excluded to judge by accident caused by artificial inertia, is prevented defect carbon block from flowing into using science, the output of accurate data and tracer technique Next procedure;
3, personnel depaly can be reduced in enterprise produces using system and method for the present invention, reduces cost of labor: Improve Quality Inspector's working environment:Quality Inspector in high temperature with high dust environment without working;
4, system and method for the invention can not influence existing with the existing production system slitless connection on-line checking of enterprise Rhythm of production, for carbon block during conveying mobile, this system can carry out Image Acquisition, data calculate analysis and defect inspection Survey, system will according to testing result, and directly guidance adjustment carbon block conveying direction, whole process without human intervention, are realized entirely certainly Dynamic production procedure;
5, the present invention realizes computer internet management, utilizes internet management technology, machine vision technique, artificial intelligence Energy, edge detection, deep learning and automated control technology provide information platform for enterprise production process product testing, make carbon Block appearance quality detection efficiency greatly improves;
6, system and method for the invention have the scalability of good system compatibility and lower cost.
Description of the drawings
Fig. 1 is the on-line detecting system schematic diagram of the present invention;
Fig. 2 is the mechanical device cell schematics 1 of the detection device of the present invention;
Fig. 3 is the mechanical device cell schematics 2 of the detection device of the present invention;
Fig. 4 is the visual identity module diagram of the detection device of the present invention;
Fig. 5 is the network topological diagram of the embodiment of the present invention 2.
Attached drawing identifies
1, roller-way 2, guider 3, position sensor
4, camera 5, revolving platform
Specific implementation mode
A kind of detecting system of heterogeneous solid material presentation quality is carried out in the following with reference to the drawings and specific embodiments detailed It describes in detail bright.
Embodiment 1
As shown in Figure 1, carbon block presentation quality on-line detecting system provided by the invention the system comprises:Presentation quality is examined Survey device, control module, vision module, defects detection module, deep learning module, data processing module, data memory module, The main modulars such as data management module.The appearance quality detecting device includes:Mechanical mechanism, position sensor 3, imaging mechanism And image capture module;
The mechanical mechanism is additionally operable to for conveying heterogeneous solid material to be detected to specified position according to inspection It surveys result and the carbon block detected is output to different regions;
As shown in Figures 2 and 3, the mechanical mechanism includes roller-way 1, guider 2 and revolving platform 5;
The roller-way 1, for transmitting carbon block in the production line;
The guider 2 is surveyed for being located in 1 liang of side rails of roller-way, for avoiding carbon block from being sent out when being moved on roller-way 1 Raw offset;
The revolving platform 5 is located at the bottom of roller-way 1, when testing result is unqualified, air brushing label in fixed position in carbon block, After completing label, the revolving platform 5 drives roller-way 1 to rotate, and the shunting of unqualified carbon block is transmitted to waste product group by roller-way 1, is not entered Production line;When testing result qualification, carbon block is delivered to normal production mode reform by revolving platform.
The position sensor 3 is mounted at the shooting work area position of described device, for being acquired in detection process The location information of detected member carbon carbon block sends the instruction for starting shooting to imaging mechanism when carbon block reaches designated position;
As shown in figure 4, the imaging mechanism includes light source and optical lens and camera, the light source, for providing imaging Required brightness, light source are controllable, using active optical mode, avoid the interference of natural light;The camera is using industrial phase Machine.
The optical lens is essential component in visual identity module, directly affects the quality of image quality, High-resolution camera lens is used in the embodiment of the present invention, ensures image quality.
The imaging mechanism further includes:Picture is stored into designated memory space after the completion of imaging;
Image capture module, the image information for acquiring imaging mechanism, and carbon to be detected is judged according to image information Whether the appearance of block is qualified, and testing result is sent to the mechanical mechanism.
As shown in Figures 2 and 3, described device further includes camera movement adjustment holder, is used for the adjustment of camera camera site.
The control module is the logic control basis of system, including the control to the mechanical mechanism, and it is electrical The data collection and control of element;
The electrical equipment includes the acquisition sensor of various signals, such as position, temperature etc..
The control module includes host computer unit, the next controller unit and communication unit;
The host computer unit is used for, and the interaction with the next controller unit in time feeds back to down testing result Level controller cell processing;
The bottom controller unit controls the camera shooting, acquires carbon for receiving the host computer unit instruction Block location information sends out action command for the mechanical mechanism;
The communication unit is used to establish the interaction channel of the host computer unit and the next controller unit, that is, counts According to instruction transmission channel;The embodiment of the present invention 1 using Industrial Ethernet network as communication network, but not limited to this.
The defects detection module, including sample image database establish unit and defect detection unit;
The sample image database establishes unit, for handling the image collected sample data, establishes sample This image data base;It is automatically performed the storage of sample image, establishes and change corresponding data information;
The defect detection unit will be for that will shoot the sample in obtained image information, with the sample image database It is compared, and is calculated, complete defects detection, and tabulation is carried out for various features, such as:Detection time, whether Defective, defect type, size, local sectional drawing etc.;Formed it is final whether qualified testing result.
The defect detection unit includes the defects detection to measured object presentation quality, and such as crackle aoxidizes, falls rib arrisdefect Deng;
The defect detection unit is used to read image information in real time from the sample image database, and processing calculates, Presentation quality inspection is completed, the various unqualified carbon blocks such as crackle, unfilled corner, oxidation are detected.The defect used in present system Detection technique method is the apparent point of brightness change in reference numbers image, and the significant changes in image attributes usually reflect figure The critical event and variation of picture., material property variation discontinuous including the discontinuous, surface direction in depth and scene lighting become Change.By the pixel set for the surrounding pixel gray scale significant changes that measured object surface defect is brought, to realize measured object surface Defects detection.
The defect inspection method is:
Step 1) reads sample image information in real time from the sample image database;
It is defect point that step 2), which identifies brightness change significantly point in sample image information,;
Step 3) reads sampled image information in real time from image capture module;
Step 4) is according to the apparent point of brightness change in step 2) sample digital picture, in sampled images described in step 3) The significant changes of pixel grey scale compare;Carry out that discontinuous, surface direction in depth is discontinuous, pair of material property variation Than;
According to the comparing result of step 4), the surrounding pixel gray scale brought by measured object surface defect drastically becomes step 5) The pixel set of change judged, it is consistent significantly to put thresholding with brightness change in sample digital picture, judges measured object surface It is defective;Realize the defects detection on measured object surface;
For step 6) according to step 5) testing result, the host computer unit gives the next controller unit to send instruction:If Testing result is zero defect, and the next controller unit sends instruction to the revolving platform, and carbon block is delivered to normal side by revolving platform To;Otherwise, the next controller unit sends instruction to the revolving platform, and defective carbon block is delivered to waste product area by revolving platform.
The deep learning module is the adaptive learning unit comprising system, the optimization unit of defects detection;Its principle For:
1) the feature neural network framework that sample different scale is established according to sample image data, by input sample figure Picture extracts characteristic information by different scale, obtains different scale images information;
2) whole image feature extraction and prediction are based on according to input picture in 1);
1) and 2) 3) according to the multiple dimensioned information of the input picture obtained, feature is extracted, model is generated, in model foundation Afterwards, new characteristic information is constantly inherited and is added during detection image data accumulation, to obtain accurate defect sample This figure improves accuracy rate, optimization algorithm.
Specific method includes:
Step 1) completes basic defects detection, and image information is stored in the sample image database;
The image information that step 2) is obtained according to the defects detection of step 1) establishes the feature nerve net of sample different scale Network framework obtains the image feature information of different scale by the decomposition of the different scale characteristic information to input sample image Pel;
The image feature information pel for the different scale that step 3) is obtained according to step 2) divides the spy of imaging surface figure Sign, and then identify different defect states, according to the different respective features of defect state and the general character, build lacking for surface image Fall into characteristic model;
The defect characteristic model that step 4) is established according to step 3), each detection device on every production line being collected into The surface image information of generation is classified, and data memory module storage to the sample image database is passed through;
Step 5) extracts surface image defect model feature and database according to the information of step 4) sample image database The characteristics of establish corresponding detecting and alarm;
The detecting and alarm that step 6) is established according to step 5) provides big data retrieval service, and image mould is obtained based on retrieval Type data acquisition system;
Step 7) constantly inherits and is added new characteristic information in the data acquisition system according to step 6), carries out depth It practises to make system continue to optimize, the result of optimization forms new sample pattern, is detected for real-time online, for testing result It is shown, inquiry and statistical analysis.
The data processing module is calculated the data in system, treatment classification;
The data processing module, for all the image collected data in system to be handled and are classified, including New image essential information;Tabulation is carried out for various features, such as:Detection time, whether defective, defect type, ruler Very little and local sectional drawing etc.;
The data memory module is that the data of system are stored and exported according to intended manner;
The data memory module stores testing result to designated memory, is integrated to the image data of storage, Report is generated, is specifically included:Data query unit and report generation unit;
The data query unit, for the collected each block of carbon block image to be numbered, convenience sample number According to storage and inquiry.For data query and analysis displaying:Inspection carbon block number, qualification rate, fraction defective, no are crossed per day/week/month Good carbon block defect type and image, bad carbon block defect classification and accounting etc.;
The report generation unit, for by daily/week/moon quality inspection report and account, quality analysis to be crossed inspection carbon block number, closed Lattice rate, fraction defective, bad carbon block defect type and image, the classification of bad carbon block defect and accounting generate report;It can also handle Remainder data in addition to present apparatus logic control.
The data management module includes monitoring, statistics, inquiry, setting etc., and the data for system are presented, including number According to displaying, data statistic analysis, the human-computer interaction functions such as priority assignation.
Embodiment 1 is provided with 4 production lines, including 4 sets of appearance quality detecting devices according to in-situ production environment.
As shown in Figure 2,3, carbon block is moved by roller-way 1 on every production line, is moved to the operating position of camera imaging, According to the collected position signal of position sensor, image collecting device work is sent instructions to;
Measured object carbon block is in transmitting moving process or static shooting, imaging start surface quality detection analysis after the completion.
The communication module of the embodiment of the present invention 1, it is fast between each module, different agreement for being realized by Ethernet TCP/IP Fast smooth data communication ensures that carbon block presentation quality on-line detecting system is reliablely and stablely run.
The communication module has open standards system, and rear end is set based on the open TCP/IP network platforms Meter supports multiple network agreement, convenient for that can be interconnected between each system, intercommunication and cross complaint, it then follows the universal interface standard of specification, Mutual restriction and influence of the system between hardware environment, communication environment, software environment and operating platform is set to minimize.
The embodiment of the present invention 1 has good human-computer interaction interface, and data processing function is powerful, can be set by the demand of enterprise Determine criterion of identification, the standard statistical analysis of setting and output statistical report form etc..
Such as Fig. 5 of embodiment 2, the embodiment of the present invention 2 further include internet informationization management mould on the basis of embodiment 1 Block, including:
Operation instruction unit passes through for establishing ERP enterprise resource plan management platforms for instructing and managing production Data informations at different levels and equipment shape can be monitored online in the data access ERP platforms of production scene by enterprise's Ethernet data interface State, rational allocation resource arrange adjustment production task, provide the management of system optimization, carried for business decision layer and employee in time The management platform of means is run for decision;
Information process unit includes the image procossing of each device for carrying out information data processing, defects detection and point The creation data of production line is carried out image procossing by class, data storage and inquiry etc. by workshop Ethernet, defects detection and point Class and data storage and inquiry etc., and store, data informations at different levels and equipment state are uploaded to by enterprise's Ethernet The ERP platforms of operation instruction unit;Opc server can also be set and interact excuse as external server.
Produced on-site detection unit, including multiple workshops, a plurality of production line, work station is arranged in production line, with existing production System carries out the interaction of key signal, and the data of each production line are acquired by live Ethernet, include the image of camera imaging Information and sensing data etc. are uploaded to information process unit via workshop Ethernet.
This unit using standard Industrial Ethernet network as between the next machine controller and server, camera and server Between communication network, upper (visual services device) and the next machine controller interaction, the next machine controller by I/O and camera, Spot sensor transmission instruction, industrial camera are triggered by the next machine controller, and data are transferred to visual services device by Ethernet Processing.
The system of embodiment 2 has many advantages, such as advance, economy, scalability, opening and ease for use, can substitute Manually, production efficiency and recognition accuracy, good compatibility and lower cost autgmentability are improved, can adapt to data volume, The variation of user demand and customer service rule, while further including the appearance quality detection of green compact carbon block as the extension of system Deng.
Embodiment 3
For the embodiment of the present invention 3 on the basis of embodiment 1, the detecting system further includes operator's management module, packet It includes:
Mode setting unit, for online mode and offline mode to be arranged, on-line mode is and the appearance quality detection Device start and stop synchronize the pattern of putting into operation, on-line checking carbon block quality;Off-line mode be with the appearance quality detecting device not Synchronous start and stop, on-site manual control detection, are used for installation and debugging and plant maintenance;
Display unit is used for human-computer interaction, including site setup touch screen, off-line mode and on-line mode switching, display The same day crosses inspection carbon block quantity, unqualified type and quantity, qualification rate and fault alarm information inquiry;
Priority assignation unit, for setting manager's permission, the display unit can by priority assignation be manager, Modification defect standard can be arranged in program maintenance person and post personnel's three-level, manager's permission, and program maintenance person can enter system and tie up Shield, post personnel can check record report, cannot change standard and enter system;
The production line of different number can be set according to the ambient conditions in plant produced workshop.
It should be noted last that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting.Although ginseng It is described the invention in detail according to embodiment, it will be understood by those of ordinary skill in the art that, to the technical side of the present invention Case is modified or replaced equivalently, and without departure from the spirit and scope of technical solution of the present invention, should all be covered in the present invention Right in.

Claims (10)

1. a kind of presentation quality on-line detecting system of heterogeneous solid material, which is characterized in that the system comprises:Several Appearance quality detecting device, defects detection module, deep learning module and control module;
The appearance quality detecting device, the instruction for executing the control module transmit the heterogeneous solid material extremely Image acquisition areas carries out Image Acquisition, is additionally operable to the heterogeneous solid material that detection is completed being output to different regions;
The defects detection module, the image for being acquired according to the appearance quality detecting device, by being in detection image No includes pixel grey scale pixel set jumpy to realize the defects detection on measured object surface, forms testing result;
The deep learning module, for optimizing defects detection module by the adaptive learning of system;
The heterogeneous solid material is transmitted to described image acquisition zone, according to inspection by the control module for sending out instruction Survey result sends out the heterogeneous solid material that instruction will detect completion and is output to different regions.
2. the presentation quality on-line detecting system of heterogeneous solid material according to claim 1, which is characterized in that described non- The detection device of homogeneous solid material appearance quality includes:Mechanical mechanism, position sensor, imaging mechanism and Image Acquisition mould Block;
The mechanical mechanism is additionally operable to be tied according to detection for conveying heterogeneous solid material to be detected to specified position Heterogeneous solid material is output to different regions by fruit;
The position sensor is mounted at the shooting work area position of described device, tested for being acquired in detection process The location information for surveying object sends the instruction for starting shooting to imaging mechanism when detected material reaches designated position;
Appearance of the imaging mechanism for heterogeneous solid material is taken pictures, and image information is obtained;
Described image acquisition module, the image information for acquiring imaging mechanism, and it is sent to the defects detection module.
3. the presentation quality on-line detecting system of heterogeneous solid material according to claim 1, which is characterized in that described to lack Sunken detection module includes that sample image database establishes unit and defect detection unit;
The sample image database establishes unit, for handling collected sample data, establishes defect sample figure As database;
The defect detection unit, the image information for reading described image acquisition module in real time, with sample image database In sample be compared, in detection image whether include pixel grey scale pixel set jumpy to realize measured object table The defects detection in face, formed it is final whether qualified testing result.
4. the presentation quality on-line detecting system of heterogeneous solid material according to claim 1, which is characterized in that the depth Degree study module includes the optimization unit of adaptive learning unit and defects detection;
The adaptive learning unit is used to establish the feature neural network framework of sample different scale, by input sample figure The decomposition of the different scale characteristic information of picture obtains the image feature information pel of different scale;
The optimization unit of the defects detection during detection image data accumulation for constantly inheriting and being added new spy Reference ceases pel, to obtain accurate defect sample image.
5. the presentation quality on-line detecting system of heterogeneous solid material according to claim 2, which is characterized in that the control Molding block includes host computer unit, the next controller unit and communication unit;
The host computer unit is used for, and testing result is fed back to the next control by the interaction with the next controller unit in time Device cell processing processed;
The bottom controller unit controls the camera shooting, acquisition is non-solid for receiving the host computer unit instruction Positions of materials information, the action for instructing the mechanical mechanism to execute instruction;
The communication unit, the interaction channel for establishing the host computer unit and the next controller unit.
6. the presentation quality on-line detecting system of heterogeneous solid material according to claim 1, which is characterized in that the system System further includes:
Data processing module, for all the image collected data in system to be handled and are classified, including new image Essential information;Tabulation is established for various features;
Data memory module, for all the image collected data of system to be stored and exported according to format, to various The tabulation of feature is preserved;With
Data management module, the data for system are presented, and include data displaying, data statistic analysis, the inquiry of data and right Outer data output.
7. the presentation quality on-line detecting system of heterogeneous solid material according to claim 6, which is characterized in that the system System further includes internet information management module, including:
Operation instruction unit receives the information and status information of equipment of the data management module output by enterprise's Ethernet, Basic data is provided, to establish ERP enterprise resource plan management platforms;With
Information process unit collects the life of the data management module by workshop Ethernet for carrying out information data exchange Data are produced, data informations at different levels and equipment state are uploaded to the ERP platforms of operation instruction unit by enterprise's Ethernet, as With external server interactive interface.
8. a kind of presentation quality on-line checking for the heterogeneous solid material that the system based on described in one of claim 1-8 is realized Detection method, including:
Step 1) control module sends out send instructions to roller-way (1), and the roller-way (1) transmits heterogeneous solid material according to instruction To image acquisition areas;
Step 2) the heterogeneous solid material reaches image acquisition areas, and location information is sent to described by displacement sensor (3) Control module,
After the step 3) control module receives location information, photographing instruction is sent out;
Step 4) the camera (3) executes photographing instruction, obtains image information and is sent to defect processing module;
Defect processing module described in step 5) reads the docking of defect sample image information in real time from the sample image database The image information that receives carries out defects detection, by judge detection image include pixel grey scale pixel set jumpy from And realize the defects detection on measured object surface, form testing result;
If step 6) testing result is zero defect, the next controller unit sends instruction to the revolving platform, and revolving platform is by carbon Block is delivered to normal direction;Otherwise, the next controller unit sends instruction to the revolving platform, and revolving platform is defeated by defective carbon block It send to waste product area.
9. the presentation quality online test method of heterogeneous solid material according to claim 9, the step 5) include:
Step 5-1) from the data memory module defect sample image information is read in real time;
Step 5-2) mark sample image information in brightness change significantly point be defect point;
Step 5-3) from the appearance delection device sampled image information is read in real time;
Step 5-4) according to step 5-2) the apparent point of brightness change in sample digital picture, to step 5-3) sampled images In the significant changes of pixel grey scale compare;The discontinuous, surface direction carried out in depth is discontinuous, material property variation Comparison;
Step 5-5) according to step 5-4) comparison, the surrounding pixel gray scale brought by measured object surface defect is jumpy Pixel set judged, it is consistent significantly to put thresholding with brightness change in sample digital picture, judges that measured object surface is that have Defect;Otherwise zero defect, to realize the defects detection on measured object surface.
10. the presentation quality online test method of heterogeneous solid material, the method further include according to claim 9: The adaptive learning that step 7) carries out system optimizes defects detection module, specifically includes:
Step 7-1) basic defects detection is completed, image information is stored in the sample image database;
Step 7-2) according to step 7-1) the obtained image information of defects detection, establish the neural network structure of sample different scale Frame obtains the image feature information pel of different scale by the decomposition of the different scale characteristic information to input sample image;
Step 7-3) according to step 7-2) the obtained image feature information pel of different scale, divide the spy of imaging surface figure Sign, and then identify different defect states, according to the different respective features of defect state and the general character, build lacking for surface image Fall into characteristic model;
Step 7-4) according to step 7-3) establish defect characteristic model, each detection device on every production line being collected into The surface image information of generation is classified, and data memory module storage to the sample image database is passed through;
Step 7-5) according to step 7-4) information of sample image database, extract surface image defect model feature and database The characteristics of establish corresponding detecting and alarm;
Step 7-6) according to step 7-5) establish detecting and alarm, big data retrieval service is provided, image mould is obtained based on retrieval Type data acquisition system;
Step 7-7) according to step 7-6) data acquisition system in constantly inherit and be added new characteristic information, carry out depth It practises to make system continue to optimize, the result of optimization forms new sample pattern, is detected for real-time online.
CN201810513734.8A 2018-05-25 2018-05-25 A kind of on-line detecting system and method for heterogeneous solid material presentation quality Pending CN108627520A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810513734.8A CN108627520A (en) 2018-05-25 2018-05-25 A kind of on-line detecting system and method for heterogeneous solid material presentation quality

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810513734.8A CN108627520A (en) 2018-05-25 2018-05-25 A kind of on-line detecting system and method for heterogeneous solid material presentation quality

Publications (1)

Publication Number Publication Date
CN108627520A true CN108627520A (en) 2018-10-09

Family

ID=63690539

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810513734.8A Pending CN108627520A (en) 2018-05-25 2018-05-25 A kind of on-line detecting system and method for heterogeneous solid material presentation quality

Country Status (1)

Country Link
CN (1) CN108627520A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110646064A (en) * 2019-09-24 2020-01-03 红云红河烟草(集团)有限责任公司 Production line material blocking diagnosis method based on k-wave radar level imaging technology
CN111102922A (en) * 2019-12-30 2020-05-05 陕西泰诺特检测技术有限公司 Pipeline geometric dimension detection device and method

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110646064A (en) * 2019-09-24 2020-01-03 红云红河烟草(集团)有限责任公司 Production line material blocking diagnosis method based on k-wave radar level imaging technology
CN111102922A (en) * 2019-12-30 2020-05-05 陕西泰诺特检测技术有限公司 Pipeline geometric dimension detection device and method

Similar Documents

Publication Publication Date Title
CN100468246C (en) Real time monitoring system for production processes and monitoring method
CN102616613B (en) Elevator monitoring system
CN108627520A (en) A kind of on-line detecting system and method for heterogeneous solid material presentation quality
CN205750504U (en) A kind of management system of backlight module automatic production line
CN105867337B (en) A kind of management system and its method of backlight module automatic production line
CN107169546A (en) A kind of power construction managing security tools and instruments management method and system based on Quick Response Code
CN107402221A (en) A kind of defects of display panel recognition methods and system based on machine vision
CN102601131A (en) Steel billet surface quality online detection device
CN105334224B (en) Automate quality testing cloud platform
CN110400387A (en) A kind of joint method for inspecting, system and storage medium based on substation
RU2007103326A (en) Method of repair and maintenance and applicable in the method of hardware-software complex for diagnostics and system for control of quality of repair and maintenance
CN106991731A (en) Intelligent polling method and system that integrated equipment is safeguarded
CN102081808A (en) Field apparatus inspection method and system
CN103778472A (en) Mobility building product traceability system and production line operation controlling method
CN110366031A (en) A kind of monitoring abnormal state and method for diagnosing faults digitizing workshop MES system view-based access control model
KR101608371B1 (en) Management system and method for production line in manufacturing plant
CN107705023A (en) Realize the control method and system of production and qualitative control informationization and standardization
CN110244665A (en) A kind of paint line Remote Intelligent Management System for Storage
CN205253592U (en) Automatic change product identification and screening production line
CN109460805A (en) A kind of the experimental data collection monitoring system and method for combination two dimensional code and video monitoring
CN107958326A (en) A kind of plant process method for inspecting based on planar bar code technology
CN110865917A (en) AR technology-based electric power machine room inspection operation method, system and application
CN107169627B (en) Electric energy meter quality control method and system based on augmented reality technology
CN107316131A (en) A kind of electric power meter mounting process quality detecting system based on image recognition
CN110796754A (en) Machine room inspection method based on image processing technology

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