Using neural networks and computerized tomography to measure wall thickness of industrial plastics

Several methods can be employed to create plastic products such as injection molding, compression and transfer molding, and blow molding. This work is an extension of previous research my colleague and I have conducted in the field of blow molding. Blow molding is a technique employed to create hollow plastic containers for marketing a variety of products such as milk, spring water, soda, anti-freeze, etc. In this process a thermoplastic is heated and pressed through a mandrel creating a hollow tube of semi-liquid material called a parison which is placed in a mold and inflated with air to the desired shape. A major concern is the measurement of parison wall thickness prior to inflation. If the proper wall thickness can be determined, waste can be minimized and hence cost, but this can be a difficult undertaking considering the elasticity of most polymers when heated. This paper extends the results previously published on a non-invasive approach for wall thickness measurements of semi-liquid plastics through the utilization of computerized tomography and neural networks.