| 研究生: |
張書榮 Chang, Shu-Jung |
|---|---|
| 論文名稱: |
應用類神經網路於水面船艦目標之識別 Target recognition of Surface ship by use of Artificial Neural Network |
| 指導教授: |
涂季平
Too, Gee-Pinn James |
| 學位類別: |
碩士 Master |
| 系所名稱: |
工學院 - 系統及船舶機電工程學系 Department of Systems and Naval Mechatronic Engineering |
| 論文出版年: | 2012 |
| 畢業學年度: | 100 |
| 語文別: | 中文 |
| 論文頁數: | 71 |
| 中文關鍵詞: | 船艦識別 、特徵參數擷取 、倒傳遞類神經 |
| 外文關鍵詞: | characteristics recognition, ship recognition, neural network |
| 相關次數: | 點閱:77 下載:6 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
在國防科技方面更是倚賴各種水面船艦進行巡防之任務,所以在港口監控船艦已成為重要課題。本文主要研究水面船艦的聲場建模與船艦識別,而船艦輻射聲場有三種不同頻域所組成且在三個部位,即中部輔機部分、中後部主機部分和尾部螺旋槳部分,這可稱為船艦近場的三亮點。本文參考[1]所建立船艦輻射聲場近場的數學模型,模擬船艦聲場時頻域訊號,並逐一修正各項不同參數(如螺槳轉速、螺槳葉片數等),作為船艦特徵辨識系統之測試。並運用數位訊號處理技術,將取得船艦訊號以傅立葉轉換來描述,將所建立之船艦訊號頻譜圖作特徵參數擷取。再將擷取之特徵參數導入倒傳遞類神經網路進行船艦特徵值基本訓練,經過倒遞傳類神經辨識系統的實驗,已能辨識船艦特徵,其誤差率在1%之內。最後利用三個不同頻域求出三個不同部位通過水聽器時間去推估船長,並且利用聲場頻譜求出臨界頻率去推估船速,其誤差率皆在10%之內。
In defense science and technology, the Harbor surveillance of ships is an important topic. The main studies of this thesis is characteristic recognition of ships by use of underwater acoustic signal. Basically, there are three kinds of signal were produced from different regions which are the auxiliary in the middle part , the main engine between stern and middle part and the propeller in the stern to construct the ship radiated noise. These three kinds signal model is called the three highlight of a ship. This paper simulates the mathematics model of the ship-radiated noise established by reference[1]. The principal sources that are the major power contributors of underwater sound field are necessary to be classified and simulated, and different parameters (such as Propeller RPM and Blade number, etc.) of the ship are discussed. Furthermore, the digital signal processing (such as Fourier transform) is used for characteristics recognition. The extracted characteristic is used for the training process of neural network algorithm. Then, a characteristics recognition system can be established using this process, and the error rate of recognition process is within 1%. Finally, the length of ship is estimated from the characteristic of the three highlight model, and the critical frequency of the ship-radiated noise is evaluated to estimate the speed of the ship. The error rates of estimated ship length and speed are within 10%.
[1] 湛雅倩,考慮艦船噪聲多源信號的仿真與輸出實現,西北工業大學,碩士學位論文,2007年3月。
[2] R. Eberhart and J. Kenndy, “A New Optimizer Using Particle Swarm Theory”,Proceeding of 6th International Symposium on Micro Machine and Human Science, 1995 , pp.39-43.
[3] R. Rajagopal, B. Sankaranarayanan, P. Ramakrishna, “Target Classification in a passive sonar An expert system approach”, IEEE ICASSP-90, 1990, 5:2911-2914.
[4] 程錦房、何希盈、王水文,“基於艦船聲磁相關的水雷抗掃性能研究”,水雷戰與艦船防護,第16卷第3期,2008年8月,9-12頁。
[5] 何正耀、張翼鵬,“艦船輻射噪聲建模及仿真研究”,噪聲控制,2005年12期:52-55。
[6] 吳國清等,“艦船噪聲識別(I)-總體框架、線譜分析和提取”,聲學學報,Vol. 23,No. 5,Mar. 1998。
[7] 吳國清等,“艦船噪聲識別(II)-線譜穩定性和唯一性”,聲學學報,Vol. 24,No. 1,Jan. 1999。
[8] 吳國清等,“艦船噪聲識別(III)-雙重譜和平均功率譜的特徵提取和模板圖”,聲學學報,Vol. 24,No. 2,Jan. 1999。
[9] 吳國清等,“艦船噪聲識別(IV)-模糊神經網路識別”,聲學學報,Vol. 24,No. 3,Jan. 1999。
[10] 熊紫英、朱錫清,“基于LOFAR譜和DEMON譜特徵的艦船輻射噪聲研究”,船舶力學,2007年02期。
[11] R. J. Urick, “Principles of Underwater Sound”, McGraw-Hill New York, c1983.
[12] D. Ross, “Mechanics of Underwater Noise”, Pergamon, Jan 1976.
[13] 李侃遠,“在時頻域使用聚類式粒子群最佳化演算法之隱蔽稀疏訊號分離研究”,東華電機碩士論文,2007年7月。
[14] Shilling, “Introduction to Digital Signal Processing Using MATLAB”,全華圖書,2011年。
[15] 向輝平、傅瑞錦等,“船艦寬帶連續譜噪聲的模擬”聲學技術第24卷第3期,2005年9月。
[16] S. Robin and C. Doursat, “Simulating Random Variables”, INA-PG, Biometry , July, 1999.
[17] Wales SC, Heitmeyer RM, “An ensemble source spectra model for merchant ship-radiated noise”, J. Acoust. Soc. Am., 111, 1211-1231, 2002.
[18] 羅華強,“類神經網路-MATLAB的應用”,高立圖書有限公司,2005年。
[19] 葉怡成,“類神經網路模式應用與實作”,儒林,1993。
[20] 黃瀚生,“機械臂之 -類神經滑動綜合控制器設計”,成大系統所碩士論文,2011。
[21] 彭成麒,“貫入試驗之倒傳遞類神經與頻散曲線之有限差分法頻估地盤剪力波速”,台大土木所碩士論文,2001。
[22] 梁淑芳,“導入LM法之平行倒傳遞演算法”,東吳資訊科學系碩士論文,2006年7月。
[23] 羅華強,“訊號處理-MATLAB的應用”,全華科技圖書股份有限公司,2003。
[24] 黃淑枝,“應用類神經網路預測材料聲阻抗之研究”,成大造船所碩士論文,2003。
[25] 林俊宏、蕭子健,“圖型系統設計之硬體界面專題製作LabVIEW 8X”,高立圖書有限公司,2007年9月。