【作者】 夏添；

【导师】 李曦；

【作者基本信息】 华中科技大学 ， 机械工程， 2012， 硕士

【摘要】 在金属切削中，铣削加工是一种先进的金属切削方式，广泛应用于模具制造业、航空工业和汽车的粗精加工中。铣削过程中的失稳会降低加工工件表面质量、缩短刀具寿命、严重时甚至会危及生命和财产的安全，因此对铣削加工过程进行稳定性监测研究具有十分重要的工程意义。本文研发了一种基于主轴电机电流信号的铣削加工过程稳定性监测方法，主要研究内容如下：（1）通过对常见的时频分析法进行对比分析，确定采用希尔伯特黄变换对铣削加工过程主轴电机电流信号进行辨识，从中更加准确地获取相应的加工状态信息。（2）根据主轴电机电流信号采集的功能需求，开发了多通道数据采集平台，包括硬件选型、传感器接口电路的设计以及主轴电机电流信号采集、处理和分析的软件程序设计等。（3）根据用于稳定监测的主轴电机电流信号实用性，确定稳定性监测的总体方案，在不同加工条件下进行多组实验研究。（4）通过实验所采集的现场数据，用希尔伯特黄变换对平稳铣削和失稳征兆的电流信号进行对比分析，提取对应的特征向量，然后用支持向量机对特征向量进行训练并建立模型，最后用训练模型去对测试集进行模式识别，从实验分析结果上看，分类识别准确率达到93.33%，能够满足稳定性监测应用的要求。更多还原

【Abstract】 Milling is an advanced way of metal cutting, which is widely used in the rough andfinish machining of mold manufacturing，aviation industry and automobile industry.Instability in milling process leads to workpieces’ poor surface quality，shorter toollife,and even endangers the lives of operators and loses the property of plants. Therefore,study of stability monitoring in milling has a very important engineering significance. Thisthesis develops a method of stability monitoring based on spindle motor current in milling.The main research contents are as follows:(1) According to the comparisons of some common time-frequency analysis, this thesisidentifies the spindle motor current signal during machining by Hilbert-HuangTransform. The corresponding processing status information is accurately obtained.(2) According to functional requirements of acquisition system for spindle motor currentsignal, the platform used for acquiring multi-channel signals is developed, includingthe selection of hardware parts, the design of sensor interface circuit and the programsfor signal acquisition, processing and analysis, and so on.(3) According to the practicality of spindle motor current for stability monitoring, theoverall plan of stability monitoring are designed. and the corresponding experimentsare performed under different processing conditions.(4) Through the data collected by the experiments, the current signals of the stability andinstability in milling are analyzed by Hilbert-Huang Transform, the related featurevectors are extracted. And then the feature vectors are trained by Support VectorMachine, and the training mode is established. At last the training mode forrecognizing the test suite gets a classification accuracy rate about93.33%from theexperimental results, so the approach can meet the demand of stability monitoringapplication.更多还原