Case Study: Krones improves bottle quality and reduces scrap with MathWorks AI-based process control

Krones AG Builds Reinforcement Learning–Based Process Control in the Blow Molder Contiloop AI for PET and rPET Bottles

Krones AG, a German manufacturer of production lines for the process, filling, and packaging industry, needed a better way to control stretch blow molding for PET and rPET bottles. As lighter recycled bottles became more sensitive to changes in raw materials and environmental conditions, Krones turned to MathWorks tools such as MATLAB, Simulink, and Reinforcement Learning Toolbox to help address the challenge.

MathWorks helped Krones develop Contiloop AI, a reinforcement learning–based control system that automatically adapts process settings in the blow molder. The solution improved bottle quality, reduced scrap, lowered operator interventions and manual errors, and enabled continuous quality measurement with early drift detection. Krones also built an end-to-end workflow for data analysis, model training, deployment, and code generation using Simulink and related MathWorks products.

Krones

Benedikt Böttcher

Research and Development Engineer