镗削颤振快速预报技术研究  

Research on Fast Prediction of Boring Chatter

作　　者：王民[1] 费仁元[1] 杨建武[1] 吴季茂[1] 关剑[1] 

机构地区：[1]北京工业大学机电学院,北京100022

出　　处：《机械科学与技术》2002年第4期520-523,共4页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金 ( 5 0 0 0 5 0 0 1);北京市科委科技新星计划 ( 95 5 810 0 90 0 )项目资助

摘　　要：切削颤振切导致产品质量、生产效率、刀具和机床使用寿命的降低 ,同时造成噪声污染 ,影响操作者身心健康。随着工厂自动化发展 ,操作者日益远离加工现场 ,对切削颤振进行在线监视预报和控制变得越来越重要。但由于其发生发展的过程极其短暂 ,使颤振的在线预报十分困难。本文在讨论了镗削颤振发展机理的基础上 ,提出了一种基于神经网络进行颤振预报的新方法。用 L O- RBF模型进行传感信号预处理 ,结果输入 Fuzzy ARTMap模型进行颤振识别 ,大大缩短了信号处理时间 ,提高了识别的准确性 ,取得了满意的结果。Chatter is one of the obstacles which influences products quality in boring. On-line chatter prediction and control is an effective method to prevent its occurrence. Because its developing course is transitory, on-line chatter prediction is difficult. This paper proposes a new prediction method with neural networks based on the analysis of the chatter mechanism in boring. LO-RBF network model is used to decontaminate the noise in the original signal. The processed signal is used as input to recognize the occurrence of chatter. This method shortens the time of recognition and improves its accuracy, and the results are satisfactory.

关 键 词：颤振预报 模糊理论 神经元网络 镗削 

分 类 号：TG53[金属学及工艺—金属切削加工及机床] TH161.6[机械工程—机械制造及自动化]