Process optimization using Artificial Intelligence in glass industry

Purpose and goal:

The purpose of the project was to use AI to predict and optimize the quality of the glass pulp. The goal was to develop a system for data collection, analysis and feedback. We have created an efficient data collection system and implemented automatic analysis of glass pucks, which should improve the production process.

The prototype for data collection has been tested at three different glassworks: OKB, Reijmyre and Målerås. In addition, an AI based tool to evaluate the quality of glass samples has also been developed. “Acoustic Agree” tested his NAW® technology to detect bubbles in glass. Expected results and effects:

The machine learning tool should be further developed and can improve the quality and reduce defects in glass mass production, with integrated ultrasound detection for the early identification of bubbles in hot glass mass.

Det AI-baserade the tool for automatic analysis of glass bodies can be further developed for other industrial applications.

The project creates a basis for implementation within the art glass industry, with flexibility for different types of furnaces and types of glass and help the glassworks to facilitate work, improve quality and reduce energy consumption. Approach and implementation:

The project was carried out according to the deliverables in the work plan. The project work was highly dependent on the cooperation between glassworks and other project members due to the need to install equipment, conduct experiments and collect data. Due to several problems, it was very challenging to first make an accurate planning and then complete all the activities on time. The main reasons were the pandemic and the personal involvement of project partners.

It is clear that a steering group with representatives from RISE, Zenta, OKB, Reijmyre and Målerås would have been necessary.