Combining BPN and TM to build a prediction model of process parameters: a case study. (3rd May 2016)

• Record Type: Journal Article
• Authors: Fu, Hsin-Pin; Chen, Cheng-Yuan; Yeh, Hsiao-Ping; Yeh, Wen-Chin; Lu, Jia-Shiang
• Is Part Of: Total quality management & business excellence. Volume 27:Number 5/6(2016)
• Page Start: 666
• Page End: 680
• Publication Date: 2016-05-03
• Subjects: BPN -- TM -- parameter prediction -- PVC -- extrusion molding; Total quality management -- Periodicals; Quality control -- Periodicals; Quality assurance -- Periodicals; Production management -- Quality control -- Periodicals; 658.401305
• DOI: 10.1080/14783363.2015.1040754
• Languages: English
• ISSNs: 1478-3363

Abstract

Abstract : In general, thermal expansion and contraction of PVC (polyvinyl chloride) can occur when the environmental temperature changes. Therefore, the process parameters of PVC plastic spiral pipes have to be changed often to prevent the production of defective products. Traditionally, experienced workers have been relied upon to adjust the process parameters to suit the changing external temperature. To replace the manual adjustments of the experienced workers, this study empirically collected data from a manufacturer of extruded PVC plastic spiral pipes, and used a combination of the back-propagation network and the Taguchi method to establish a parameter prediction model that can predict the extrusion process parameters necessary for the changing external temperature. With a higher than 99% prediction accuracy, the proposed model can assist manufacturers in automatically and accurately adjusting the PVC pulling (or rotating) speed of the haul unit during extrusion.