碩士 / 建國科技大學 / 土木與防災研究所 / 98 / Typhoon-wave forecast for coastal areas for the purposes of disaster prevention and vessel, play an important role, but the Typhoon wave phenomenon than the General wave phenomenon differs in its impact factor is more complex, and it has the characteristics of nonlinear, it is not easy to experience traditional numerical methods or formulas can accurate forecasting. This artificial neural network features include: (1) high-speed computing power, (2) self learning ability, (3) high-capacity memory, (4) fault-tolerant capability and use of local weather data to perform a type of artificial neural network to projections of Typhoon waves.
First off the northeast coast of the Taiwan longdong station for object, classes neural model of test and validation. Network indicator height, average cycle, peak periods, average wind direction, wind gust, three seconds, the wave direction, station pressure, average temperature, surface temperature and nine other input factors. Observatories data to two hours for a sum of money, after testing framework for the introduction of this network of five input values, 1 layer hidden layer, 1 output value, there are five floors, learning factor 0.7 neurons. In addition, this article is 2004 three different Typhoon track, Typhoon Mindulle, Aere and Nanmadol, respectively, using artificial neural network (ANN) and SWAN model projections of typhoons, wave and showed this pattern projection wave value and the actual measurement values are consistent with, and predictable after 6 hours of Typhoon waves.
Keywords - Artificial Neural Network (ANN) ; SWAN Model
| Identifer | oai:union.ndltd.org:TW/098CTU05653007 |
| Date | January 2010 |
| Creators | 楊志群 |
| Contributors | 黃清和 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 80 |
Page generated in 0.011 seconds