Temperature Regulation of Machine Tools Using Neural Network

• Main Authors: Min-Cian Chang, 張銘謙
• Other Authors: Yau-Zen Chang
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/79327906724843007115

碩士 === 長庚大學 === 機械工程研究所 === 92 === The main purpose of this paper is to study the neuro-control strategy in the temperature regulation of machine tools to reduce thermal effects on machining accuracy. The nonlinear and time-varying relationship between heat generated in the shaft motors and cooling system renders the temperature regulation problem difficult to model and analyze. An experimental system is built with a heater driven by PWM signals to simulate the heat source. The cooling subsystem is constructed with helix pipe driven by a 370 W pump. Temperature is collected from a thermocouple for feedback. Data sets of pump control signals, temperature of target position, and on/off signals of heat sources are collected to train two artificial neural networks of feedforward configuration using the standard back-propagation algorithm. One of the neural networks is used for determination of system order. The other neural network is used for closed-loop control, which combines the feedforward signal of the heat sources. Experimental results show that the neuro-control approach is both effective and relatively easy to apply.