應用類神經網路及支持向量機於銑削顫振精準預測
Other Title
High Accuracy Milling Chatter Detection Based on Neural Networks and Support Vector Machine
Part Of
中華民國振動與噪音工程學會論文集
Journal Issue
2019
Start Page
100
End Page
105
Date Issued
2019-06
Author(s)
楊傑程
Abstract
在工業4.0的浪潮下,智慧機械開始蓬勃發展。而智慧機械其中一個非常重要的領域便是自我異常檢測。要達到自我異常檢測則勢必要使用到機器學習，因此本論文先建立具有時變性切削時間延遲的銑削系統穩定性圖後，沿不穩定邊界切削取得穩定及不穩定資料點，接著以切削訊號的相對小波包能量熵作為分類特徵，最後以類神經網路及徑向基函數核的支持向量機進行顫振預測。
Under the wave of Industry 4.0, smart machinery began to flourish. One of the most important areas of smart machinery is self-anomaly detection. In order to achieve self-anomaly detection, machine learning is necessary. Therefore, after creating the stability lobe diagram (SLD) during milling, the paper obtains stable and unstable data points along the unstable boundary of the SLD, and then uses the relative wavelet packet energy entropy (RWPEE) of the cutting signal as a classification features. Finally, the chatter prediction is performed by a neural networks and a support vector machine (SVM) with the RBF kernel.
Subjects
顫振
類神經網路
支持向量機
機器學習
chatter
neural network
support vector machine
machine learning
Type
conference paper