基于主成分分析与C-支持向量机的刀具磨损状态监测

Tool Wear Condition Monitoring based on Principal Component Analysis and C-Support Vector Machine

为了监测刀具磨损状态,建立了一个基于功率传感器的刀具磨损状态监测系统.提出了一种基于主成分分析(PCA)与C-支持向量机(C-SVM)相结合的刀具磨损状态监测模型.通过功率传感器采集切削过程中的电流和功率信号,采用PCA对采集的参数进行特征提取,选择对刀具磨损状态影响最大的主成分作为C-SVM的输入样本,实现对刀具磨损状态的准确识别.通过数控车床切削实验表明,即使在较少的样本条件下,该方法仍然有效,并与反向传播(BP)神经网络进行了性能比较.

In order to monitor tool wear condition（TWC）,the power-sensor-based monitoring system on the state of machining tool wear was designed.The monitoring model of TWC was proposed based on principal component analysis（PCA）and C-support vector machine（C-SVM）.Current and power signals were obtained from power sensor during cutting process.After that,the features of these signals were extracted using PCA.The principal components,mainly affecting TWC,were chosen as the input samples of C-SVM to carry out monitoring the tool condition with accuracy.The results of computerized numerical control（CNC）turning machine tool show that the model is effective even in the case of a small samples.Moreover,a comparison about the monitoring and prognostics capability between the presented method and back propagation（BP）neural network has been made.