The challenges associated with manufacturing PET containers are varied and complex. Process windows are short, and even the daily changes in ambient temperature and humidity in the production hall require regular manual adjustments to various process parameters to ensure the required level of container quality. Added to this are the steadily increasing production speeds, where even the slightest deviations from ideal conditions can have a major impact on the production result. Meanwhile, demands on personnel are also evolving: Today, one operator is responsible for multiple machines and systems – and has less time for visual quality checks and manual process control on the stretch blow molder.
Since its market launch, Contiloop AI has not only convinced customers around the world, but also this year's jury of the AI Breakthrough Award, consisting of recognised experts from the fields of business, marketing, sales, analysis and science. The Krones solution received the MLOps Innovation Award, which recognises the most innovative AI product in the field of machine learning operations.
Training and optimising Contiloop AI
As part of an automated test run, various process settings are carried out and the resulting measurement results are forwarded to the Krones IIoT platform. There, the data flows into the Krones AI pipeline, which is used to train the future control algorithm, the AI agent. Once the training is complete, this AI agent is transferred to the machine's Contiloop AI and is then available for production operation. The system thus learns to be able to adjust the stretch blow-moulding process precisely to the perfect bottle quality even under the new conditions.
