Abstract: To realize the analysis of batch step quality cutting based on processes monitor signal's features,a signal feature extraction method based on improved Angle Measure Technique(AMT)algorithm was proposed.In this method,the computer complement operation principle was adopted to improve base points selection of AMT.The groups spindle power signals of batch drilling were taken as research object,these signals were preprocessed by methods of z-score normalization and segmentation,and scale features of the monitor signals were extracted from local to global by the improved AMT.The dimensionality of these scale features was reduced to obtain a distribution of the batch drilling step quality by the method of Principal Component Analysis(PCA).The results showed that the improved AMT could be used to overcome the negative influence of features extraction from the great signal amplitude magnitude differences and the great signal data scale|the conclusion of the step quality classification could meet with the physical test result,and the result could provide theoretical guidance for artificial sampling observation of high precision cutting step quality.

Key words: batch drilling, step quality, angle measure technique algorithm, feature extraction, principal component analysis

摘要： 为实现基于切削过程传感器信号特征的批量工步质量分析,提出一种基于改进算法的传感器信号特征提取方法。该方法基于计算机补码运算原理选择曲率尺度测量技术算法的基本点,从而改进曲率尺度测量技术算法。以批量钻削工步主轴功率信号为例,首先对监测信号进行z-score标准化和分段处理,然后利用改进后的曲率尺度测量技术算法提取批量钻削工步监测信号从局部到全局的尺度特征;最后采用主成分分析法进行特征降维,从而获取批量钻削工步质量的分布情况。计算结果表明,曲率尺度测量技术算法能够降低监测信号幅值数量级差距和数据规模过大对特征提取效果的影响,批量钻削工步质量分类符合实际情况,可解决批量工步质量的人工抽检随意性问题。

关键词: 批量钻削, 工步质量, 曲率尺度测量技术算法, 特征提取, 主成分分析