Title: Research on prediction method on RUL of motor of CNC machine based on deep learning

Authors: Chu-chu Rao; Ren-wang Li

Addresses: Institute of Mechanical and Electrical Engineering, Quzhou College of Technical, 324000, Quzhou, China ' Institute of Mechanical and Manufacturing Automation, Zhejiang University of Sci-Tech, Zhejiang, 324203, China

Abstract: To solve the problem of high fault frequency and sudden occurrence of the motor of computer numerical control (CNC) machine tool, the paper proposes a deep learning remaining useful life (RUL) prediction model based on DFS-LSTM. Through collecting the motor life cycle data by sensors, constructing the dataset, then extracting the depth feature set from the original data by DFS (feature depth synthesis), and the depth feature will be inputting into the LSTM(long-short term memory) model for training, then the prediction model is obtained. In order to realise the function of predicting RUL, Deadline time function is designed in data processing, and residual life is calculated by data before Deadline time. The model is applied to the RUL prediction of the motor of computer numerical control (CNC) machine tool, and obtained a good prediction result.

Keywords: motor; DFS; depth feature synthesis; LSTM; long-short term memory; deadline time; RUL; remaining useful life; CNC machine tool; predict; feature.

DOI: 10.1504/IJCSM.2021.120689

International Journal of Computing Science and Mathematics, 2021 Vol.14 No.4, pp.338 - 346

Received: 08 Sep 2020
Accepted: 07 Dec 2020

Published online: 03 Feb 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article