A Study on Deep Learning Application of Vibration Data and Visualization of Defects for Predictive Maintenance of Gravity Acceleration Equipment

被引：16

作者：

Lee, SeonWoo ^{[1
]}

Yu, HyeonTak ^{[2
]}

Yang, HoJun ^{[1
]}

Song, InSeo ^{[1
]}

Choi, JungMu ^{[1
]}

Yang, JaeHeung ^{[3
]}

Lim, GangMin ^{[3
]}

Kim, Kyu-Sung ^{[4
]}

Choi, ByeongKeun ^{[2
]}

Kwon, JangWoo ^{[1
]}

机构：

[1] Inha Univ, Deparment Elect Comp Engn, 100 Inha Ro, Incheon 22201, South Korea

[2] Gyeongsang Natl Univ, Dept Energy & Mech Engn, 38 Cheondaegukchi Gil, Tongyeong Si 53064, South Korea

[3] ATG, R&D Ctr, Seongnam Si 13558, South Korea

[4] Inha Univ, Coll Med, Inha Res Inst Aerosp Med, Dept Otolaryngol Head & Neck Surg, 3-Ga Shinheungdong, Incheon 400711, South Korea

来源：

APPLIED SCIENCES-BASEL | 2021年 / 11卷 / 04期

基金：

新加坡国家研究基金会;

关键词：

artificial intelligence; deep learning; fault detection; hyper-gravity machine; vibration monitoring; FAULT-DIAGNOSIS; ROTATING MACHINERY; AUTOENCODER; NETWORK; MODEL;

D O I：

10.3390/app11041564

中图分类号：

O6 [化学];

学科分类号：

0703 ;

摘要：

Hypergravity accelerators are a type of large machinery used for gravity training or medical research. A failure of such large equipment can be a serious problem in terms of safety or costs. This paper proposes a prediction model that can proactively prevent failures that may occur in a hypergravity accelerator. An experiment was conducted to evaluate the performance of the method proposed in this paper. A 4-channel accelerometer was attached to the bearing housing, which is a rotor, and time-amplitude data were obtained from the measured values by sampling. The method proposed in this paper was trained with transfer learning, a deep learning model that replaced the VGG19 model with a Fully Connected Layer (FCL) and Global Average Pooling (GAP) by converting the vibration signal into a short-time Fourier transform (STFT) or Mel-Frequency Cepstral Coefficients (MFCC) spectrogram and converting the input into a 2D image. As a result, the model proposed in this paper has seven times decreased trainable parameters of VGG19, and it is possible to quantify the severity while looking at the defect areas that cannot be seen with 1D.

引用

页码：1 / 15

页数：15

共 49 条

[1] Fault Detection and Classification Based on Co-training of Semisupervised Machine Learning [J].