The Application of Neural Network in Self-tuning Adaptive Control of Constant Turuning Force

• Main Authors: Chiou Gwo Ching, 邱國慶
• Other Authors: Hwang Shiuh Jer
• Format: Others
• Language: zh-TW
• Published: 1993
• Online Access: http://ndltd.ncl.edu.tw/handle/09631183642430031670

碩士 === 國立臺灣科技大學 === 工程技術研究所 === 81 === The purpose of constant turning force control is to increase metal removal rate (MRR) and to prevent tool breakage in the turning process. Then the productivity could be increased. This object can be reached by using feedrate manipulation to maintain a specified cutting force. Classical control theory (PID) which has good robustness is applied in constant turning force control system with fixed cutting depth. However, this system may became unstable when the cutting depth is changed significantly. In this conditon, a pole assignment self-tuning adaptive control theory with recursive least square parameters estimator is proposed to solve this problem.Unfortunately, the adaptability and robustness of self- tuning adaptive control system can not be maintained in good condition simultaneously. To address these problems, in this paper a method for self- tuning adaptive control based on neural network is derived which can improve both the adaptability and the robustness. In order to verify this new method, we design a feedrate mechanism to model the turning process. The experimental results of implementing the control theory to this imitative mechanism are satisfactory.