بحث بعنوان Deep Learning Application of Vibration Data for Predictive Maintenance of Gravity Acceleration Equipment

بحث بعنوان Deep Learning Application of Vibration Data for Predictive Maintenance of Gravity Acceleration Equipment
اسم المؤلف
SeonWoo Leea, Yu-Hyeon Takb, Ho-Jun Yanga, Jae-Heung Yangc, Gang-Min Limc, KyuSung Kimd, Byeong-Keun Choib and JangWoo-Kwona
التاريخ
14 نوفمبر 2020
المشاهدات
253
التقييم
(لا توجد تقييمات)
Loading...

بحث بعنوان
Deep Learning Application of Vibration Data for Predictive Maintenance of Gravity Acceleration Equipment
SeonWoo Leea,1, Yu-Hyeon Takb,2, Ho-Jun Yanga, Jae-Heung Yangc, Gang-Min Limc, KyuSung Kimd, Byeong-Keun Choib,+ and JangWoo-Kwona*
aDeparment Electric Computer Engineering, Inha University, 100, Inha-ro, Michuhol-gu, Incheon,
Republic of Korea
bDepartment Mechanical Engineering, Gyeong-Sang National University, 38, Cheondaegukchi-gil,
Tongyeong-si, Gyeong sangnam-do, Republic of Korea, 530-64
cR&D Center, ATG, #1104, KINS Tower, 331-8, Seongnam-daero, Bundang-gu, Seongnam-si,
Gyeonggi-do, Korea
dDepartment of Otolaryngology-Head and Neck Surgery, Inha University College of Medicine,
Incheon, 3-Ga Shinheungdong, Jung-Gu, Incheon 400-711, Korea
Abstract. Hypergravity accelerators are used for gravity training or medical research. They are a kind of large
machinery, and a failure of large equipment can be a serious problem in terms of safety or costs. In this paper, we
propose a predictive maintenance model that can proactively prevent failures that may occur in a hypergravity
accelerator. The method proposed in this paper is to convert vibration signals into spectograms and perform
classification training using a deep learning model. We conducted an experiment to evaluate the performance of
the method proposed in this paper. We attached a 4-channel accelerometer to the bearing housing which is a
rotor, and obtained time-amplitude data from measured values by sampling. Then, the data was converted into a
two-dimensional spectrogram, and classification training was performed using a deep learning model for four
conditions of the equipment: Unbalance, Misalignment, Shaft Rubbing, and Normal. Experimental results
showed that the proposed method has an accuracy of 99.5%, an increase of up to 23% compared to existing
feature-based learning models.
Keywords: Artificial Intelligence, Deep Learning, Preventive maintenance, Hyper-gravity Machine, Vibration Monitoring
Conclusion and Future Work
In this study, in order to prevent accidents that may occur in large equipment such as a gravitational
accelerator, we measured vibration signals with accelerometers, used the measured data to train and
test a deep learning model by using spectrogram visualization based on MFCC and STFT, and
attempted to evaluate the proposed method.
The major methods used in this experiment were to convert vibration signals into images and apply
a modified VGGNetwork to a fault model. The proposed deep learning architecture enables the
diagnosis of a total of 4 conditions, such as Normal, Rubbing, Misalignment and Unbalance, and both
MFCC and STFT models showed the average accuracy rate of 99.5%. In addition, the proposed
models were compared with feature-based machine learning models using existing traditional
methods. Experimental results showed that the proposed models have better performance in all
evaluation parameters of accuracy, recall, precision, and F1-Score, compared to existing feature-based
learning models. These results indicate that the proposed method can be successfully used as a fault
diagnosis and assessment model if a monitoring environment is constructed by attaching sensors in theassessment of the stability of gravity acceleration equipment in the future.
In addition, it was shown that existing vibration data can also be converted into image data such as
spectrograms, one of the methods used in speech recognition, and they can be applied to an imagebased deep learning model. The method proposed in this study has the following limitations. First, the
patterns of fault data should be prepared in advance. These shortcomings should be addressed through
further studies such as research on outlier detection. Second, training takes considerable time and
requires additional hardware such as GPUs. Taking into account these limitations, a method of
reducing computation amounts should be performed so that the proposed method can be used for small
edge devices required for commercialization.
كلمة سر فك الضغط : books-world.net

The Unzip Password : books-world.net

تحميل

يجب عليك التسجيل في الموقع لكي تتمكن من التحميل
تسجيل | تسجيل الدخول

التعليقات

اترك تعليقاً