臺灣博碩士論文加值系統

為了避免設備的無預警停機、故障及提昇工作環境品質，在設備上建立一套故障監診系統是必要的。閥門損壞為往複式壓縮機常見之故障，本研究將針對一部運轉中的往復式壓縮機閥門，建立一線上監診系統。因往復式壓縮機振動訊號為時變、非穩態訊號，傳統的傅立葉轉換將無法提供足夠訊息以供判別，因此需借助時頻分析技術。而短時距傅立葉轉換屬於線性時頻分析，受不確定準則影響，有解析度不佳的困擾。若以二次式時頻分佈函數作振動訊號分析工具，必須考慮干擾項之影響，因此必須選擇一適當核心函數作為去除干擾項之濾波器。本研究將以再分佈之平滑式假性韋格納-威利分佈函數作為分析工具，最後以機率類神經網路作為訊號的分類及比對工作，完成設備故障之監診系統。

To avoid unexpected down time or fault of equipment and to improve the quality of working environment, it is necessary to establish a fault detection and diagnosis system. Valve damage is a common fault in reciprocating compressors. An on line fault detection and diagnosis system will be developed in this study for valves of a reciprocating compressor which is in service. The traditional Fourier transform can not provide sufficient information for decision making due to the time-varying and non-stationary characteristics of the valve vibration signals of the reciprocating compressors. Time-frequency analysis technique is then crucial for successful treatment. The short time Fourier transform belongs to the class of linear time-frequency analysis and is subject to the constraint of uncertainty principle. Therefore, its time-frequency analysis resolution is quite limited. On the other hand, when applying the quadratic time-frequency distribution function for analysis of vibration signals, it is necessary to consider the effects of disturbances. The selection of an appropriate kernel to serve as a filter is crucial for removing the disturbances. This study applies the reassigned smooth pseudo Wigner-Ville distribution function as an analysis tool. The probabilistic neural network is applied to classify and signal class, which completes the establishment of the fault detection and diagnosis system.

第一章 緒論 1
1.1 簡介與研究動機 …………………………………………………… 1
1.2 相關研究與重點 …………………………………………………… 3
1.3 章節內容概要 ……………………………………………………… 5
第二章 閥門分析 7
2.1 閥片模態分析……………………………………………………… 7
2.1.1 閥門構造……………………………………………………… 7
2.1.2 模態測試……………………………………………………… 8
2.2 有限元素模型建立與驗證 ……………………………………… 11
2.2.1 有限元素模型參數 ………………………………………… 11
2.2.2 ANSYS分析結果……………………………………………… 12
2.2.3 模態測試與ANSYS分析比對………………………………… 13
2.3 受拘束閥門之動態分析 ………………………………………… 14
2.3.1 正常閥片之分析 …………………………………………… 14
2.3.2 彈簧變異與閥片破裂之分析 ……………………………… 16
第三章 訊號的線性處理與轉換 36
3.1 時域平均技術 …………………………………………………… 36
3.2 傅立葉轉換 ……………………………………………………… 37
3.3 線性時頻分析 …………………………………………………… 38
3.4 壓縮機訊號處理 ……………………………………………… 39
第四章 二次式時頻分佈函數 54
4.1 時頻分佈函數 …………………………………………………… 54
4.1.1 韋格納分佈函數 …………………………………………… 54
4.1.2 時頻分佈函數之通式 ……………………………………… 55
4.1.3 核心函數 …………………………………………………… 58
4.2 再分佈原理 ……………………………………………………… 59
4.3 壓縮機訊號時頻分析 …………………………………………… 61
第五章 訊號識別 72
5.1 機率類神經網路 ………………………………………………… 72
5.2 PNN比對因子……………………………………………………… 75
5.3 壓縮機振動訊號比對 …………………………………………… 76
第六章 結論 100
參考文獻 ………………………………………………………………… 103

[1] 韓捷, 張瑞林, “旋轉機械故障機理及診斷技術,” 機械工業出版社, 大陸北京, 1997.
[2] 劉衛華, 郁永章, “往復壓縮機故障診斷技術研究與展望,” 壓縮機技術, 第3期, 1999.
[3] M. Kato, M. Kurohashi and M. Aoshima, “Dynamic behavior of valves with pneumatic chambers for reciprocating compressors,” Journal of Vibrations and Acoustics, vol. 115, pp. 371-376, October 1993.
[4] O. Bardou and M. Sideahmed, “Early dection of leakages in the exhaust and discharge systems of reciprocating machines by vibration analysis,” Mechanical Systems and Signal Processing, vol.8, no. 5, pp. 551-570, 1994.
[5] S. Gade and Klaus Gram-Hansen, “The analysis of nonstationary signals,” Sound and Vibration, pp. 40-46, January 1997.
[6] M. Chiollaz and B. Faver, “Engine noise characterisation with Wigner-Ville time-frequency analysis,” Mechanical Systems and Signal Processing, vol. 7, no. 5, pp. 375-400, 1993.
[7] Y. S. Shine, J. J. Jeon and S. G. Spooner, “Pseudo Wigner-Ville distribution and its application to machinery condition monitoring,” MFPT Meeting, no. 47, pp.235-250, 1993.
[8] Y. S. Shin and J. E. Harding, “Time-frequency domain representation for application to machine condition monitoring,” MFPT Meeting, no. 49, pp.337-347, 1995.
[9] K. C. Chou, “Multiscale statistical modeling approach to monitoring mechanical systems,” MFPT Meeting, no. 50, pp.431-440, 1996.
[10] D. F. Specht, “Probabilistic neural networks,” Neural Networks, vol. 3, pp. 109-118, 1990.
[11] D. F. Specht, “Probabilistic neural networks and polynomial adaline as complementary techniques for classification,” IEEE Transactions on Neural Networks, vol.1, pp. 111-121, 1990.
[12] L. Cohen, “Time-frequency distribution—A review,” Proc. IEEE, vol.77, no. 7, pp. 941-981, July 1989.
[13] F. Hlawatsch and G. F. Boudreaux-Bartels, “Linear and quadratic time-frequency signal representations,” IEEE Singal Processing Magazine, pp. 21-67, April 1992.
[14] T. J. Cavicchi, Digital Signal Processing, John Wiley & Sons, New York, ch. 5, 2000.
[15] A. V. Oppenheim, A. S. Willsky and S. H. Nawab, Signals and Systems, Upper Saddle River, N.J., 2nd ed., 1997.
[16] L. Cohen, Time-Frequency Analysis, Prentice Hall, Englewood Cliffs, New Jersey, 1994.
[17] F. Gu and A.D. Ball, “Use of the smoothed pseudo-Wigner-Ville distribution in the interpretation of monitored vibration data,” Maintenance, vol.10, no.2 , pp. 16-23, 1995.
[18] T. J. Wahl and J. S. Bolton, “The application of the Wigner distribution to the identification of structure-borne noise components,” Journal of Sound and Vibration, vol. 163, no. 1, pp.101-122, 1993.
[19] J. Jeong and W. J. Williams, “Alias-free generalized discrete-time time-frequency distribution,” IEEE Trans. on Signal Processing, vol.40, no.11, pp.2757-2765, Nov. 1992.
[20] J. C. Moss and J. K. Hammond, “A comparison between the modified spectrogram and the pseudo-Wigner-Ville distribution with and without modification,” Mechanical Systems and Signal Processing, vol. 8, no. 3, pp. 243-258, 1994.
[21] J.-H. Lee and J. Kim, “Development of enhanced Wigner-Ville distribution function,” Mechanical Systems and Signal Processing, vol. 15, no. 2, pp. 367-398, 2001.
[22] H. Garudadri and E. Velez, “A comparison of smoothing kernels for computing bilinear time-frequency representations,” Int. Conf. IEEE, 1991.
[23] E. Chassande-Mottin, I. Daubechies, F. Auger, P. Flandrin, “Differential reassignment,” IEEE Signal Processing Letters, vol.4, no.10, pp.293-294, October 1997.
[24] F. Auger and P. Flandrin, “Generalization of the reassignment method to all bilinear time-frequency and time-scale representations,” Proc. IEEE ICASSP, pp. IV-317-320, 1994.
[25] F. Auger and P. Flandrin, “The why and how of time-frequency reassignment,” IEEE-SP Sympo sium on time-frequency and time-scale analysis, pp. 197-200, October 1994.
[26] B. Liang, F. Gu and A. D. Ball, “Detection and diagnosis of valve `faults in reciprocating comprocessors,” Ninth International Congress and Exhibition on Condition Monitoring and Diagnostic Engineering Management (COMADEM 9), Sheffield, pp. 421-430, 1996.
[27] B. Liang, F. Gu and A. D. Ball, “A preliminary investigation of valve fault diagnosis in reciprocating compressors,” Maintenance, vol.11, no.2, pp. 3-8, 1996.