With ComoNeoPREDICT, machine settings can now be improved to the point where quality, cycle time and process reliability are all optimised. Based on the cavity pressure curve, it generates reliable statements about each manufactured part in advance.
The quality predictions are based on DoE models that calculate the characteristics of each injection-moulded part. This cutting-edge technology makes use of supervised machine learning algorithms based on neural networks. The first step is to generate a statistical test plan (DoE) which is worked through with ComoNeo on the machine. ComoNeoPREDICT then uses the parameters obtained (such as pressure and temperature curves) and the measured quality characteristics of the part to create a prediction model that serves as a reference for subsequent production runs. If a part is outside of the defined tolerance band limits, it can be separated out automatically. Custom PC software is provided to generate the DoE and the prediction model.
ComoNeoPREDICT adds a range of practical benefits to the ComoNeo system including:
• Accurate scrap separation method from complete quality records • Users acquire know-how from direct monitoring of predicted quality characteristics • Pseudoscrap is minimized • Tolerances and scrap criteria are easily defined • Relevant curve points for calculation are selected automatically • No special knowledge of maths or statistics is needed to operate the software • Ongoing training of models results in continuous improvement • Significant reduction of manual measuring operations in production. |
#PAuto #Kistler
No comments:
Post a Comment