臺灣博碩士論文加值系統

臺灣四面環海，周遭海域豐富的漁業資源及多樣化海洋資源與得天獨厚的地理位置致使海上交通運輸船舶活動相當熱絡，由於船舶在航行時，常因受限水域、海況、天氣條件、操舵人員的操舵經驗以及船舶自身操縱條件等不同因素之影響而有所變化。此外，船舶在海上航行亦常有外力干擾如：風、浪、海流及潮汐等之影響，致使船舶容易受到干擾而偏離預定之航道，因此需要俱有即時修正船舶航線之反應能力，以確保船舶能以較佳的航行方向與速度避免災難的發生。海上航行存在諸多不可抗拒的危險，當值人員若應變不當或一時疏忽時，將會導致船舶遭遇例如意外碰撞等事故、增加生命財產的損失。因此如何避免船舶意外碰撞的問題即具有其急迫性，而為了降低人為操作因素所產生之疏失，本論文將應用海上船舶回傳之船舶自動識別系統(Automatic Identification System, AIS)資料，進行船舶航行安全之研究，其研究主要內容為透過大數據(Big Data)分析技術與深度學習(Deep Learning, DL)技術對目前既有的AIS資料進行特徵學習，分析臺灣海域船舶航行特性，並設計船舶航行安全預測系統，使系統能針對海上船舶航行行為及碰撞告警進行有效預測分析，促進臺灣海域的航行安全及效率。

Taiwan is surrounded by the sea, the abundant fishery resources and diverse marine resources in the surrounding waters and the unique geographical location make the maritime transport vessels very frequent here. When the ship is sailing, it often affected by different factors such as restricted waters, sea conditions, weather conditions, helmsman's experience and the ship's own operating conditions. Furthermore, ships sailing on the sea often have external incident such as wind, waves, currents and tides, which cause the ship to be easily disturbed and deviate from the scheduled course. Therefore, it is necessary to have the ability of correcting the ship's route to ensure that the ship can avoid disasters with better navigation direction and speed immediately. There are many irresistible dangers in maritime navigation. If the duty officer is negligent or not action properly, it will lead to accidents such as collision, cuase the loss of life and property. In view of this, how to avoid the accidental collision of the ship has its urgency, and in order to reduce the mistake caused by the human operation factors, this paper applies the Automatic Identification System (AIS) data returned from the ship to research the safety of navigation. The main content of the research is to learn the characteristics of the existing AIS data through Big Data analysis technology and Deep Learning (DL) technology, by this way to learn the sailing characteristics of ships around Taiwan, then, designed a ship navigation safety prediction system based on the sailing characteristics of these ships. This system can effectively predict and analyze the navigation behavior and collision warning of sea vessels, and promote the safety and efficiency of navigation in Taiwan.

中文摘要 ----------------------------------------------------------------------- i

ABSTRACT ----------------------------------------------------------------------- ii

誌謝 ----------------------------------------------------------------------- iv

目錄 ----------------------------------------------------------------------- v

表目錄 ----------------------------------------------------------------------- vii

圖目錄 ----------------------------------------------------------------------- viii

專有名詞對照表 ----------------------------------------------------------------------- x

一、 緒論----------------------------------------------------------------- 1

1.1 研究背景----------------------------------------------------------------- 1

1.2 研究目的----------------------------------------------------------------- 3

1.3 文獻探討----------------------------------------------------------------- 6

1.4 論文架構----------------------------------------------------------------- 9

二、 相關研究軟體與技術-------------------------------------------- 11

2.1 自動識別系統---------------------------------------------------------- 11

2.2 深度學習----------------------------------------------------------------- 12

2.3 Python與Keras-------------------------------------------------------- 14

2.4 Anaconda---------------------------------------------------------------- 15

三、 AIS資料處理------------------------------------------------------ 16

3.1 錯誤數據整理---------------------------------------------------------- 16

3.2 AIS資料壓縮演算法------------------------------------------------- 25

四、 船舶航行安全預測系統----------------------------------------- 28

4.1 船舶行為預測系統--------------------------------------------------- 29

4.1.1 特徵及標籤分類------------------------------------------------------- 30

4.1.2 人工神經元------------------------------------------------------------- 32

4.1.3 循環神經網路---------------------------------------------------------- 33

4.1.4 長短期記憶網路------------------------------------------------------- 34

4.1.5 雙向循環神經網路--------------------------------------------------- 38

4.1.6 研究成果----------------------------------------------------------------- 39

4.2 船舶航行安全評估系統--------------------------------------------- 49

4.2.1 船舶航行異常---------------------------------------------------------- 50

4.2.2 船舶碰撞危險度分級------------------------------------------------ 58

五、 結論與未來展望-------------------------------------------------- 64

5.1 結論----------------------------------------------------------------------- 64

5.2 未來展望----------------------------------------------------------------- 65

參考文獻 ----------------------------------------------------------------------- 66

[1]Lloyd’s Register, Group Review 2016, Global Marine Trends Publications, 2016.
[2]Organization for Economic Co-operation and Development, 2016, “Shipping intelligence weekly”, CLarksons research.
[3]Statista商業數據平台，https://www.statista.com/statistics/254266/global-big-data-market-forecast/.
[4]科技新報，https://technews.tw/2018/09/06/appier-report-taiwanese-company-layout-ai/.
[5]Sheng-Long Kao, et al., 2007, “A Fuzzy Logic Method for Collision Avoidance in Vessel Traffic Service”, The Journal of Navigation.
[6]Chih-Hao Wen, et al., 2012, “Identifying Smuggling Vessels with Artificial Neural Network and Logistics Regression in Criminal Intelligence Using Vessels Smuggling Case Data”, Asian Conference on Intelligent Information and Database Systems (ACIIDS 2012), pp. 539-548.
[7]Ahmad C.Bukharia, et al., 2013, “An intelligent real-time multi-vessel collision risk assessment system from VTS view point based on fuzzy inference system”, Expert Systems with Applications, pp. 1220-1230, March.
[8]Shangbo Mao, et al., 2016, “An Automatic Identification System (AIS) Database for Maritime Trajectory Prediction and Data Mining”, Extreme Learning Machines 2016 (ELM-2016), pp. 241-257.
[9]ITU-R M.1371-5, 2014, “Technical characteristics for an automatic identification system using time division multiple access in the VHF maritime mobile frequency band”, Recommendation.
[10]Geoffrey E.Hinton, Alex Krizhevsky, Ilya Sutskever, 2012, “ImageNet Classification with Deep Convolutional Neural Networks”, NIPS'12 Proceedings of the 25th International Conference on Neural Information Processing Systems, pp. 1097-1105.
[11]Statista商業數據平台，https://www.statista.com/statistics/808190/worldwide-large-scale-visual-recognition-challenge-error-rates/.
[12]Keras官方網站，https://keras.io/why-use-keras/
[13]Gao, M.; Shi, G.Y.; Li, W.F., 2018, “Online compression algorithm of AIS trajectory data based on improved sliding window”. J. Traffic Transp. Eng., pp. 218–227, June.
[14]維基百科，https://en.wikipedia.org/wiki/Haversine_formula.
[15]ZENG Jian-Hui, 2012, “Research Of Method of Calculating Ship Collision Risk Index”, College of Automation of Harbin Engineering University, Master of Telematics.
[16]Michael Galetzka, Patrick O. Glauner, 2017, “A Simple and Correct Even-Odd Algorithm for the Point-in-Polygon Problem for Complex Polygons”, 12th International Joint Conference on Computer Vision， Volume 1: GRAPP, Porto, Feb 27 - Mar 1.
[17]羅逢源，吳東明，蘇東濤，2007，”多目標船舶碰撞資訊的演算解析法”，技術學刊，22卷，3期，頁243~252。