期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2020
卷号：98
期号：23
页码：3770-3782
出版社：Journal of Theoretical and Applied
摘要：In order to obtain quality cable products in the thermoplastic extrusion process, it is important that during the polymer extrusion process, a melt that is homogenous in both temperature and composition is delivered. To achieve this, it is important to control, monitor, identify and select the important parameters during the extrusion process which directly impacts the product output. Some of these parameters include the melt pressure, temperature, line speed, screw speed, amongst others. In developing countries, however, these parameters are often selected on a trial and error basis which often leads to waste of material and the production of poor quality cables. This paper focuses on a technique which can be used to predict realistic extrusion process parameters for medium to high voltage cable insulations using artificial neural network. Real life datasets for the extrusion of Polyethylene (PE) thermoplastic were obtained and a three-layered feed-forward neural network as developed in the MATLAB environment. The neural network model developed can predict the manufacturing extrusion process parameters for different grades of PE thermoplastic which is used for medium to high voltage electrical cable insulation. A regression value of 0.99569 was obtained and a mean square error of 2.98052�〖10〗^(-6) was achieved.
