以機器學習分析轉動不平衡之現象

Dynamic Unbalance Analysis by Using Machine Learning

易則希(Tse-Hsi Yi)

指導教授 ： 王怡仁

淡江大學/工學院/航空太空工程學系碩士在職專班/碩士(2022年)

https://doi.org/10.6846/TKU.2022.00746

摘要

轉子不平衡在所有旋轉機械中無處不在，對機器的壽命和運行構成嚴重威脅。轉子不平衡和質心不對中是振動的兩個主要來源，大部分的轉動不平衡可透過處理這兩項因子來解決。
本研究動機是希望透過機器學習的方式，解決業界機械元件轉動不平衡的預測提防機制，透過轉動不平衡儀器取得振動數據，接續使用微軟發行之整合機器學習軟體Microsoft ML Azure Studio，進而達到低成本又可獲得實驗結果之目的。
本論文首先利用轉動測量儀器測得轉動平衡儀是否平衡，若無平衡，將於不同的位置增加不同質量，來觀察不同變因後之響應。本研究實驗之數據是使用TECOM PRO 9000多通道頻譜分析儀及壓電式振動規，進行測量並蒐集加速規之轉動情形，在不同位置的負重所得到不同的振動數據來進行分析。
機器學習的部分，吾人採用監督式學習法之相關演算法，透過Microsoft ML Azure Studio 之軟體，輸入旋轉振動儀器所取得之數據，進而分析並推測後續時序之振動數據。

關鍵字

轉動不平衡 ； 振動 ； 機器學習

並列摘要

Rotor imbalance is ubiquitous in all rotating machinery and poses a serious threat to the life and operation of the machine. Rotor imbalance and centroid misalignment are the two main sources of vibration, and most rotational imbalances can be resolved by addressing these two factors.
The motivation of this research is to use machine learning to solve the prediction and prevention mechanism of unbalanced rotation of mechanical components in the industry, obtain vibration data through rotating unbalanced instruments, and then use ML Azure Studio, an integrated machine learning software promoted by Microsoft. to obtain the experimental results and achieve low-cost and high-quality products.
In this paper, we first use a rotating measuring instrument to measure whether the rotating balancer is balanced. If there is no balance, it will be in different positions and add different masses to observe the response after changing different variables. The data in the experiment is to use TECOM PRO 9000 multi-channel spectrum analyzer and piezoelectric vibration gauge to measure and collect the rotation of the accelerometer and the different vibration data generated at different positions and loads for analysis.
In the machine learning part, we use the relevant algorithms of the supervised learning method, and through the software of Microsoft ML Azure Studio, we bring in the data obtained by the rotating vibration instrument, and then analyze the vibration data, and make subsequent judgments and prediction of the vibration data in the selected time zone.

並列關鍵字

Rotation unbalace ； Vibration ； Machine learning

參考文獻

[1] MD. Abdul Saleem, G. Diwakar, M.R.S. Satyanarayana, “Detection of Unbalance in Rotating Machines Using Shaft Deflection Measurement during Its Operation,” IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE), ISSN 2278-1684 Volume 3, Issue 3, Sep.-Oct. 2012, pp. 08-20.

[2] B. K. Kumar, G. Diwakar, M. R. S. Satynarayana, “Determination of Unbalance in Rotating Machine Using Vibration Signature Analysis,” International Journal of Modern Engineering Research (IJMER), ISSN: 2249-6645 Vol. 2, Issue. 5, Sep.-Oct. 2012, pp. 3415-3421.

[3] N. Ahobal and S.L. Ajit Prasad, “Study of vibration characteristics of unbalanced overhanging rotor,” ICONAMMA2018, IOP Conf. Sci. Eng. 577 012140, 16-18 Aug 2018, Bangalore, India.

[4] M. Gohari and A. Kord, “Unbalance Rotor Parameters Detection Based on Artificial Neural Network,” Article in International Journal of Acoustics and Vibration, Vol. 24, No. 1, Mar. 2019, pp. 113-118.

[5]A. Khan, H. Hwang, H.S. Kim, “Synthetic Data Augmentation and Deep Learning for the Fault Diagnosis of Rotating Machines,” Mathematics, Sep. 2021, pp. 1-26.