基于改进EEMD和香农熵的铣刀磨损状态监测研究

Research of Tool Wear Monitoring Method Based on Modified EEMD and Shannon

为了分析铣削加工中复杂的声发射信号和克服EMD方法的模态混叠缺陷,在原始集合经验模态分解(EEMD)的基础上,提出一种改进的EEMD方法应用于铣刀磨损的状态监测。通过引入白噪声准则和中值滤波优化EEMD的算法,并基于香农熵从分解得到的IMF分量中提取有效分量,剔除虚假分量。最后将有效IMF分量的能量作为特征向量输入支持向量机(SVM)分类器来识别铣刀的磨损状态。经过在立式铣削加工中心上进行实验,结果表明此方法在识别铣刀磨损状态方面具有更高的准确性。

For the complex acoustic emission signals in milling process and modal aliasing defects of EMD method, a new approach based on original ensemble empirical mode decomposition （EEMD） was proposed to achieve the detection and identification of tool wear in milling process. Through introducing white noise criteria and median filter to optimize the algorithm of EEMD. Then extracting effective intrinsic mode functions （IMFs） and excluding false functions according to Shannon. At last, the energy of effective IMF functions are taken as inputs of support vector machine （SVM） classifier to identify the state of cutter. After experiments on vertical milling centers, the results showed that this method could more accurately identify the tool wear.