|
||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||
Faulty Product Recognition and Seperation by PC Based Machine Vision System
Company:
Turkish Henkel / Izmir
Goal :
There was
a need to a solution which inspects the quality of packing of the product
and prevents the transportation of the faulty products before the packaging
process. Also the locations of the products in the boxes have to be
checked too.
 |
 |
Design and Technology:
During our the system analysis it is found that the front side and backside of the products have to be tested on a line which has a speed of 175 products per second. The following inspections had been made on Real Time:
A) Tap Inspections:
1- Is there
a tap or not?
2- Color, dimension, deformation inspection
3- Is the tap well closed? ( low pressed?, -diş atma-? )
B) Double Side Bottle Body Inspections:
1-
Right body color to the right product ?
2- Is there any deformation or bruise?
3- Inspection of date and production information ( one side only )
C) Double Side Label Inspections:
1-
Is there a label or not? Is it the right label?
2- Any torn? Any separation from the body?
3- Is the label slid or twisted while sticking?
4- Is more than one label sticked?
D) Package ( Box) Inspections:
1-
Number of the products
2- Product locations and their positions.
3- The existence or inexistence of the taps.
In addition the operation requirements must be provided too.
E) Operating Requirements:
1-
0.3 percent measuring sensitivity and a inspection speed of 175 products
per second
2- Easy and quick configuration for the new products
3- Rapid adaptation to the changes in the products
4- Rapid and easy adaptation when the inspected product changed
5- Minimum upkeep and configuration requirements.
6- High reliability and easy usage
7- Hierarchy in user organization
8- Reports for operation and faults
9- Statistical analysis (SPC) structure
10- All inspections are adaptable to be controlled remotely or over
the Web
11- Adaptation to the process line and operating circumstances
 |
||
| 1.
System control enclosure 6. Lighting Systems 7 Product Entry 9. Label Units |
10.Faulty
Product Conveyor 13. Package control unit |
|
With their knowledge and experience, E3TAM R&D group decided that the solution to this type of complex inspections with their suitability to the operation requirements can be developed as a PC Based Machine Vision technology. And they have succeed in developed the first Machine Vision project in Turkey and this project is a significant one of its kind in the World.
In our system the product images which were taken by the help of industrial illustration , industrial cameras and help of optical system and transmitted to the computer by the frame grabber boards are used for all inspections in our system.
All of three cameras are in the system are colored and for the product inspections which has the need of high speed and high sensitivity 2 digital Basler cameras were used. And for the package inspections one analog camera were used.
Schneider Optics lenses were chosen because of their optical and mechanical perfection.
The frame grabbers and the I/O cards and the software development enviroment LabVIEW are the products of National Instruments.
Operator Interface , 3 parallel real time Machine Vision algorithm and the management of the hardware need a high system source. For this reason in a machine vision project of this size its common to use more than one computer and share the duties between them. The optimization in the design made it possible to be confirmed by one computer. But as the system will always be working under hard operating circumstances it was decided to use a double CPU server.
All the components of the system ( lighting, camera, computer etc.)
are rugged in appropriate industrial enclosures.
 |
 |
 |
solution :
On a process with a high variety of products and changeable faults on the product an algorithms which find different faults can not be successful even though they need long computing periods. For these reason the introduction of the ideal products to the system instead of the faults and comparison of the products on the line with the ideal product were chosen for the system. The product image were divided into 25 different areas and analyzed. And for every area different fault tolerance is used. So the operator can be notified and the operation with the aimed quality level is obtained.
The application of higher mathematics and complex algorithms in a Machine Vision system and the hard and complex configuration of the system is familiar to the experts. But for the operators which have many responsibilities in the field. The user interface and the transition between the products were designed to be easy.
For the system, there is two defined different user modes, the operator and the manager. The organization of the users, and introduction of a new product to the system or a change in an introduced product are being handled by the managers. It is enough for the operators to select the product which will be inspected and to pushes the " Run" button.
The introduction of a new product while the ideal products were passing on the line can be completed in 5 to 10 minutes. This introduction is,
1)
The adjustment of the area that camera sees and the recording of the
images of the passing products to the computers.
2) The determination of a ideal products within the recorded images
3) The determination of the areas on the product which will be inspected
4) The automatic determination of the fault tolerance with a test operation.
5) The introduction of the package and the products in the package.
With automatically generated fault and operation reports the information
for the supervising and observing of the process.
Users can easily access to the simple but efficient help file when it
is needed.
As the appropriate cost and effortless and rapid adaptation to the new
product is aimed system is designed and developed integrated to the
production line.
How the system works:
The grabbing of the images are being triggered by digital sensors.
The system software in the server which was developed by the engineers of E3TAM project group has the algorithm to compare the products passing the line with the ideal product according to the tolerances and to find the fault types even they were not foreseen. Front and backside inspections of the products and the inspections on the package can be executed real time with a sensitivity of 0.3 and a speed of 175 products per second.
In a case of fault operator is notified and product situation information is included to the reports and the statistics. In a case of continuous or a periodic fault. The line is stopped and the operator is informed about what to do. And the faulty products are transported to the separation system and taken out of the line.
Result :
Our system " Computer Based Machine Vision Technologies " which is completely a product of Turkish engineering and the products software/hardware products of the world wide known companies are used.
We are proud of our expandable, capable, total and high technology solution and its assistance on the success of our customer.
Application Areas :
Our
system which is designed suitable for different sectors like electronic,
automotive, (metalurji), medicine and for all kind of (talaşlı) production.
With its recent form it can be used for " Final Quality Control
" in chemistry, cosmetic, medicine, food, glass, packaging industries.
This and our familiar systems can be updated and expanded with their
flexible architecture.
General Specifications of The Project:
For all inspections the real time product images which were taken with the help digital cameras and optical systems which are specially designed for this type of applications and transmitted to the computer by the help of frame grabbers are used. For these image grabbing process a special industrial illustration and the triggering system was designed by the E3TAM R&D engineers to take the images in the right time and without any fault.
The system software is completely designed for the requests of the customer.
The software compares the products passing the line with the ideal products
according to the tolerances and finds the faults which were not foreseen.
The system software which can execute the front and backside inspections
of the products and the inspections on the package real time with a
sensitivity of 0.3 and a speed of 250 products per minute. This ratio
is a very high performance according to the functions of the software.
In a case of fault operator is notified and product situation information
is included to the reports and the statistics. In a case of continuous
or a periodic fault. The line is stopped and the operator is informed
about what to do. And the faulty products are transported to the separation
system and taken out of the line. The faults can be reported and is
suitable for any statistical analysis. System is equipped with the ability
to learn and teach. System is ready to log on to Internet to be observed
and controlled remotely.
