Process Parameters Optimization in Plastic Manufacturing Industry using Machine Learning

Abstract

This research is aimed at optimizing the process parameters in plastic manufacturing industry using machine learning. Taguchi L27 orthogonal array was employed in the design of experiments and small-the-better signal to noise ratio was used in the analysis of volumetric shrinkage variation (VSV) of the plastic tray. The analysis of variance (ANOVA) was used to determine the significant parameters affecting VSV of the product. Artificial neural network (ANN) was employed in optimization of the machine process parameters settings and prediction of the quality characteristic.  The results showed that the optimal VSV occurred at mould temperature of 44⁰C, melt temperature of 210⁰C, holding pressure of 41MPa, injection time of 6 seconds, and packing time of 8 seconds. The ANOVA results showed that the mould temperature was the dominant factor influencing VSV of the plastic product. The optimization of the machine process parameters and quality characteristic using ANN model reduced the shrinkage value of the product by 29.9%. This study concludes that the optimal process parameters settings determined using ANN are better than the settings being used in the industry.