碩士 / 國立臺灣海洋大學 / 資訊工程學系 / 105 / Taiwan is an island country filled with fishing ports. According to official records, there are currently 224 fishing ports which need to be maintained, requiring a high upkeep of over 1 billion NTD per year. Since not every fishing port is highly utilized, a method of lowering the cost is to shut down some fishing ports. However, doing so by considering on utilization may have a major impact on safety issues. For example, fishing ports that serve as a sanctuary can not be dismissed, for it serves as a protection for fishing vessels during extreme weather conditions, i.e., typhoons. Trying to identify sanctuary fishing ports by using human labor will also cost a fortune, thus, seeking help from machine intelligence is inevitable. We have developed an automatic method to identify these sanctuary fishing ports using voyage data recorder data along with external weather data.
This research observes the voyage data recorder (VDR) information of fishing vessels and typhoons from 2007 to 2015. Over 5.2 billion GPS points and 209 typhoons are analyzed. We aggregated the VDR information and weather data using different criterions and created indices to show the purpose of a fishing port. We compute the best threshold value to partition the indices into sanctuary port group and non-sanctuary port group using the training data, which is acquired by experts observations. Experiment results show that the following index, the average number of docked vessels when the typhoon is within 1 nautical mile proximity of a fishing port, is the most significant signal.
The method proposed in this research can be further improved if more criteria can be considered, e.g., torchlight fisheries can be influenced by moonlight, seasonal fishing activities, and changes in ocean currents. Moreover, many small fishing vessels that are smaller than 10 gross tons can not be tracked as they do not have VDR data and so the accuracy can deteriorate because these small vessels contribute to a major population of off shores and near seas fishing activity. These lead to future work which can improve the identification accuracy of the algorithm.
| Identifer | oai:union.ndltd.org:TW/105NTOU5394020 |
| Date | January 2017 |
| Creators | Wang, Hao-Shiun, 王浩勳 |
| Contributors | William W.Y. Hsu, 許為元 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 59 |
Page generated in 0.0101 seconds