Predictive Modeling for Quality Assurance of Extrusion Blow Molding

In Extrusion Blow molding process, the manufacturing of hollow products isdone by using semi-molten tube (parison). During processing of the product, the defects such as underweight of part, warpage, part does not blow and thinning of the weld arises. The deviations in process variables are responsible in causing the defects in products. Predictive models are developed in finding the relationship between process variables and causes of defects using data mining techniques. This paper presents Random Tree, Naive Bayes (Kernel), and Neural Net, Stacking and Vote techniques in Rapid Miner for data mining. These techniques are implemented on Extrusion Blow Molding Process dataset collected during the manufacture of 100 ml made of high-density polyethylene (HDPE) material in building data mining models for the assurance of quality and predictions of product quality.