【摘要】 依据刀具磨损主轴振动加速度信号所建立的自回归模型,提取了12个反映刀具磨损状态的特征参数,提出用模糊聚类模式识别铣削刀具磨损状态。试验结果表明:建立的刀具振动加速度时间序列信号自回归模型所提取的模型参数估计特征量,含有丰富的刀具磨损运行状态信息。在此基础上,利用模糊聚类分析相似尺度来衡量事物之间的亲疏程度,并以此来实现分类。研究结果表明:基于时间序列分析与模糊聚类分析相结合的刀具磨损状态识别方法,能有效识别铣削刀具磨损运行状态,选取置信水平λ=(0.65~0.79)时所得到的刀具磨损运行状态的识别结果更为准确;采用不同的相似系数计算方法所求得的模糊相似矩阵、自回归系数的阶数、置信水平λ的取值范围不同,识别正确率有所不同。更多还原

【Abstract】 An indirect identification method is developed for monitoring tool wear conditions,where the fuzzy clustering pattern is used to recognize wear conditions based on the auto-regressive model and 12 characteristic parameters caused by tool wear axis vibration acceleration signals.The features of the estimated model parameters contain either plentiful tool wear running information or wear conditions which are a effectively adopted to measure the similarity relations by fuzzy clustering analysis.While selecting the confidence level λ as 0.65~0.79,the idendification gets more accurate,and the correctness rate depends upon the different fuzzy similarity matrix,order of regression coefficients and confidence level range.更多还原