Finding an Optimal Operation Conditions of Plastic Molding by Artificial Neural Network

• Published in: Transactions of the Japan Society of Mechanical Engineers Series C Vol. 63; no. 614; pp. 3538 - 3543
• Main Authors: YONEHARA, Noriyoshi, ITO, Ikuo
• Format: Journal Article
• Language: Japanese
• Published: The Japan Society of Mechanical Engineers 1997
• Subjects: Generalization; Learning; Molding Method; Neural Network; Optimization; Plastic Molding; Random-Tabu Mehod
• ISSN: 0387-5024; 1884-8354
• DOI: 10.1299/kikaic.63.3538

This report concerns a determination of the optimal operational conditions of plastic injection. Normally, it is difficult to apply mathematical programming to this type of problem because the nature of the problem is ill-defined. As an alternative approach, an artificial neural network (ANN) was chosen to solve the optimal conditions in this study. The ANN required input of 14 attributes (x) under 15 environmental conditions (s). In this specific problem domain, a set of s was used as an input vector, and x as an output vector, which is different from the typical placement of variables where (s, x) was input and evaluation of products, h, was an output. This was because a huge data set dominating failure cases had to be dealt in the typical placement of variables. The manufactures' succeeded data were divided into two data sets, one of which was used for teaching data (s as input and x as output), and the other was for the generalization of the network. A significant improvement of learning time was observed by further reconstruction of two ANNs with divided two groups of data, according to the characteristics of the data.