Approach for inline monitoring and optimization of a thermoplastic injection molding process with Bayesian networks taking the example of the quality feature weight

Injection molding is an important and widely used process for the production of thermoplastic parts. The efforts of industry, society and politics to increase the use of recycled plastics in the future pose new challenges. These include ensuring consistently high product quality despite fluctuations in material properties. To reduce waste in the interests of resource efficiency, it is necessary to closely monitor the process and continuously adapt the machine settings to the changing conditions. Various machine learning algorithms have already been tested in this context. This paper presents another approach based on Bayesian networks (BN) for making quality predictions, diagnosing defects and recommending actions to machine operators. Using the quality characteristic of part weight as an example, the entire process from data acquisition to model validation is considered.