Injection Molding: Artificial Intelligence Predicts Part Quality
The K 2019 show made clear that artificial intelligence (AI) is the leading edge of innovations to create smarter machines and production systems (see show report in January issue). Kistler Instrument revealed it is using AI to predict part quality based on process data. Kistler’s newly commercial ComoNeo Predict software (pictured) is powered by STASA QC, developed by Steinbeis Angewandte Systemanalyse GmbH (Steinbeis Applied System Analysis) in Germany. STASA QC uses machine learning to predict quality data such as part dimensions (or strength) based on the cavity-pressure curve derived from in-mold sensors. The result is automated QC without time-consuming metrology tests.
Kistler’s ComoNeo Predict incorporates STASA QC software that uses artificial intelligence to predict part quality from cavity-pressure data.
The STASA QC software is “trained” by means of DOE configured by the software and executed automatically on an injection machine. The DOE involves five to seven repetitions of variations of three parameters—hold pressure, hold time, and temperature—in three steps each. The trial parts are measured and the data entered into STASA QC. The software then determines the influence of each parameter of part quality. (Cycle time can also be included in the optimization.) The result is a “systematic process setup” in which the user enters the upper and lower quality limits for the part, and the software determines how to optimize settings to achieve the desired quality. In production, ComoNeo Predict can be used to make accept/reject quality decisions on each cycle based on the cavity-pressure data.
At the K show, Sumitomo (SHI) Demag said it is experimenting with STASA QC, which it interfaced with its NC5 machine controls.
Producing plastics parts with undercuts presents distinct challenges for molders.
One of the important process parameters to establish and record for any injection molded part is its injection or fill time. Research research reveals the limitations of popularly taught methods of establishing this critical parameter.
One of the more prominent trends in processing is the need for higher plastic pressures to mold parts.