本研究將採用機器學習中的倒傳遞神經網路(Back-propagation Network, BPN)以及支持向量機(Support Vector Machine, SVM)兩種不同的方法對兩顆不同的螺槳產生之噪聲訊號進行螺槳聲場的辨識與分類，預期將會使用BPN進行螺槳聲場的轉速辨識以及分類、使用SVM進行螺槳聲場的辨識，並加以比較兩種不同的方法對於螺槳分類之結果有何差異之處。
船艦的水下聲紋資料的取得十分困難，故本研究使用成功大學拖航水槽與單獨螺槳試驗機進行螺槳噪聲資料之量測。由於成功大學拖航水槽的槽壁並無加裝防止迴響之吸音材，所以本研究使用時間反轉法(Time Reversal Method, TRM)將量測到的噪聲訊號進行還原，並加以比較還原前後對於機器學習預測的成效差異。
根據本研究之成果，當螺槳噪聲經過時間反轉法的還原後，其預測成果皆有顯著之提升，而透過減少訓練資料比對之成果，我們可了解到傳遞神經網路訓練資料量與預測成果之相對關係。

This research applies the neural network and support vector machine to identify and classify the noise signals generated by two different propellers. It is expected that neural network will be used to predict the propeller rotating speed and the classification of the number of blades, and the SVM algorithm will be used to classify the number of propeller blades, and to compare the two different methods is for the propeller classification.
In addition, Time Reversal Method is used to restore the signal to improve the prediction results of neural network and SVM. The results show that the improvement effects of TRM on the two machine learning methods.
According to the results of this study, it can be found that after the propeller noise is restored by time reversal method, the prediction results are significantly improved. Reducing training data will give poor prediction via the neural network. Therefore, it is important to make sure the sufficient data is used for the databases.

[1] H. Seol, B. Jung, J.-C. Suh, and S. Lee, "Prediction of non-cavitating underwater propeller noise," Journal of Sound and Vibration, vol. 257, no. 1, pp. 131-156, 2002.
[2] J. Bosschers, F. Lafeber, J. de Boer, R. Bosman, and A. Bouvy, "Underwater radiated noise measurements with a silent towing carriage in the Depressurized Wave Basin," in AMT Conference, Gdansk, 2013.
[3] ITTC, "Model-Scale Propeller Cavitaion Noise Measurements," ITTC Recommended Procedures and Guidelines 2017.
[4] J. Bosschers et al., "The Specialist Committee on Hydrodynamic Noise," 2017.
[5] W. Soares-Filho, J. M. de Seixas, and L. P. Caloba, "Averaging spectra to improve the classification of the noise radiated by ships using neural networks," in Proceedings. Vol. 1. Sixth Brazilian Symposium on Neural Networks, IEEE, pp. 156-161,2000.
[6] M. Zhu, A. Hahn, Y.-Q. Wen, and A. Bolles, "Identification-based simplified model of large container ships using support vector machines and artificial bee colony algorithm," Applied Ocean Research, vol. 68, pp. 249-261, 2017.
[7] M. Khishe and M. R. Mosavi, "Classification of underwater acoustical dataset using neural network trained by Chimp Optimization Algorithm," Applied Acoustics, vol. 157, 2020.
[8] W. B. Qiao, M. Khishe, and S. Ravakhah, "Underwater targets classification using local wavelet acoustic pattern and Multi-Layer Perceptron neural network optimized by modified Whale Optimization Algorithm," Ocean Engineering, vol. 219, 2021.
[9] D. Borkowski, "Identification of the optimal control characteristics of a small hydropower plant using artificial neural networks and the support vector machines method," Journal of Hydraulic Research, vol. 57, no. 5, pp. 715-723, 2019.
[10] 王思皓, "拖航水槽之螺槳噪聲標準量測分析技術建立," 碩士, 系統及船舶機電工程學系碩博士班, 國立成功大學, 台南市, 2021.
[11] 羅華強, 類神經網路-MATLAB的應用. 新竹市: 清蔚科技股份有限公司, 2001.
[12] 張書榮, "應用類神經網路於水面船艦目標之識別," 碩士, 系統及船舶機電工程學系碩博士班, 國立成功大學, 台南市, 2012.
[13] H. Yu and B. M. Wilamowski, "Levenberg-marquardt training," Industrial electronics handbook, vol. 5, no. 12, p. 1, 2011.
[14] C. M. Bishop, Pattern recognition and machine learning. springer, 2006.
[15] (2020). 區塊鏈分析中的過度擬合挑戰[Online]. Available: https://tw.icointime.com/post/164508423953.html.
[16] T. Fletcher, "Support vector machines explained," Tutorial paper., Mar, p. 28, 2009.
[17] 盧奕中, "運用機器學習之支援向量機進行船艦分類及辨識" 碩士, 系統及船舶機電工程學系碩博士班, 國立成功大學, 台南市, 2020.
[18] 許莅樺, "水下通訊應用於海洋聲場" 碩士, 系統及船舶機電工程學系碩博士班, 國立成功大學, 台南市, 2018..
[19] 涂季平, 工程聲學量測理論與應用講義. 2016.