Chatter detection in high-speed milling processes based on ON-LSTM and PBT
- PDF / 4,942,368 Bytes
- 18 Pages / 595.224 x 790.955 pts Page_size
- 44 Downloads / 204 Views
ORIGINAL ARTICLE
Chatter detection in high-speed milling processes based on ON-LSTM and PBT Fei Shi1 · Hongrui Cao1 · Yuke Wang2 · Boyuan Feng2 · Yufei Ding2 Received: 5 June 2020 / Accepted: 19 October 2020 / Published online: 10 November 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Chatter is a kind of self-excited vibration which frequently occurs in high-speed milling processes, which induces severe damage to both spindle tools and workpieces. In this paper, we introduce a new chatter detection technique using orderedneurons long short-term memory (ON-LSTM) and population based training (PBT). First, we conduct a large number of milling experiments on a computer numerical control (CNC) milling machine with 4 accelerometers to get the dataset and employ vanilla LSTM for chatter detection. Then, to interpret the performance on time series of recurrent neural networks (RNN), a variation of LSTM named ON-LSTM is applied to chatter detection and a hyperparameter tuning method PBT is used for training. Finally, we compare the trained ON-LSTM with the time-frequency spectrum of the original signals obtained by short-time Fourier transform (STFT), and they show a certain degree of consistency. Keywords Chatter detection · High-speed milling · ON-LSTM
1 Introduction High-speed milling has been widely used in many manufacturing fields due to its high efficiency and low heat generation. However, a kind of fault named chatter appears frequently because of self-excited vibration. Chatter occurrence in a machining process has several severe
Hongrui Cao
[email protected] Fei Shi [email protected] Yuke Wang yuke [email protected] Boyuan Feng [email protected] Yufei Ding [email protected] 1
Mechanical Engineering, Xi’an Jiaotong University, Xi’an, China
2
Computer Science, University of California, Santa Barbara, Santa Barbara, CA, USA
adverse effects, such as poor resultant surface quality, unacceptable inaccuracy, excessive noise, disproportionate tool wear, and machine tool damage [1]. To eliminate the damage chatter causes, scholars all over the world are developing researches from 3 main aspects [2], which are the analytical study of chatter stability [3], chatter detection [4], and online active control [5]. Due to the tight coupling and highly time-varying properties of the spindle system, engineers cannot guarantee the accuracy of analytical studies, and chatter may still occur within the stable zone of the stability lobe diagram [6]. Online active control is an excellent solution for chatter, which can eliminate the relative vibration between the cutting tool and the workpiece by an external force and works only after chatter occurrence, and some scholars also utilize dynamic vibration absorber to control the vibration [7]. However, even if we want to utilize active control techniques, chatter detection techniques should also be applied first to monitor the current condition. Compared with analytical studies, chatter detection techniques work regardless of the componen
Data Loading...
