Norway Grants

PSIS PROJECT

Express on-line PET preform defects inspection system based on machine vision and AI technology

learn more

About project

Some of the most common defects in PET preforms with increased rPET concentration are different types of inclusions, haziness, and yellowing. Hence, for an upcoming R&D project funded by Norway Grants Sprana intends to develop a new inspection system that enables the detection of black specks and other contaminations at an early stage of production.

In this project, the company will carry out research and experimental activities in order to improve the quality control of PET preforms in on-line monitoring and analysis and to increase recycled materials concentration in production processes. This is realised by using machine vision and AI technology. Sprana will also develop the system’s prototype, which will be tested and installed on-site. The company states that by optimising and improving quality control of PET preform production, manufacturers will be able to reduce the routine human labour part of the quality assurance process, which is seen as inefficient, time-consuming, and increases the possibility of defective products. Released from routine tasks, staff will instead be able to carry out higher value-added processes and focus on higher quality of the products.

Overall, the project not only aims to promote the usage of secondary PET resin in the industry and to stimulate the market for rPET trading, but the partnership will also take a further step towards the EU goal of creating the circular economy for the PET plastic packaging industry.

...
...

Progress

from 1st of January 2022 to 28th of February 2022
...

The „Alfa“ prototype casing has been developed by SPRANA team, the optimal interior of the casing has been selected to minimize the negative geometric effect of light reflection and scatering, and an IR light source has been designed and manufactured - an LED panel that has been integrated into the original prototype design.


...

Experiments with power supply units for LED panels have been performed, negative effects of flickering have been observed, which will be eliminated by improving the design of electronics.
Software analysis of different cameras resolution was performed. The optimal relationship between photo resolution and computational capabilities was sought.
A list of required calculation techniques has been compiled.


...

Optimal machine vision IR / VIS camera is selected from world leading manufacturers.
Trace simulations with IR machine cameras and IR filters were performed.


...

Primary defect detection algorithms were developed, and primary initial results were obtained.


Contact


Ricardas Razgaitis
+370 640 04521

Address


Nalsios st. 11, LT-14332
Vilnius, Lithuania