A Study of WEDM Machine-tool Temperature Variation Measurement and Compensation using Back-propagation Neural Networks

• Main Authors: Yu-Sheng Cheng, 鄭玉生
• Other Authors: Shao-Hsien Chen
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/09001856205647234904

碩士 === 國立勤益科技大學 === 機械工程系 === 101 === In the processing of electro-discharge wire-cutting machine, the impact of the machining accuracy can be divided into geometric precision, positioning accuracy and surface accuracy, in which the geometric precision might be generated due to thermal deformation of the structure caused by the temperature rise of low outrigger impacted by heat energy during the discharge processing. The present paper used high-precision gantry type line horse driven EDM, coupling with PC-based controller as well as using intelligent module to match quartz heating tube, to establish a heating temperature rise and thermal displacement measurement system under the outrigger of discharge wire cutting machine. There were a total of eight temperature sensors and three displacement sensors to measure the resulting temperatures and displacement data. Furthermore, this research also used the "Matlab" software in neural toolbox and applied back propagation neural method for data correlation analysis and prediction compensation model. Finally, the verification was conducted via PC-based input compensation model. The experimental results showed that, the thermal error defamation of Y-axis of EDM low outrigger after compensation reduced from 21μm to 4.14μm.