海洋工程研讨会论文格式范例.docVIP

PAGE PAGE 23 Simulated wave-driven ANN model for typhoon waves Hsien-Kuo Changa Jin-Cheng Lioua,* Shen-Jung Liub Shyne-Ruey Liawb a Department of Civil Engineering, National Chiao Tung University, Hsinchu 300, Taiwan b CECI Engineering Consultants, Inc., Taiwan, Tai ABSTRACT This paper investigates an artificial neural network (ANN) model for typhoon waves used to modify poor calculations of the numerical model in special cases. Two key factors, local winds and simulated waves produced by the numerical model, were used as input parameters of the proposed ANN model. The waves were simulated by the numerical model from a wave action equation indicating the physical processes of energy transfer and wave propagation. Simulated wave input is a very important parameter for the proposed ANN model, allowing for the accurate calculation of water waves in the sea. An acceptable Mike21_SW model was chosen to provide an accurate calculation. Through model verification, the proposed ANN model has a particularly accurate calculation at the peak of each typhoon and at its occurrence time. The performance of the computed waves of each typhoon in comparison to the observed waves was examined and was also found to be good. Keywords: MIKE21_SW; Artificial neural network; Typhoon waves * Corresponding author. Tel.: +886 3 5131487; fax: +886 3 5131487. E-mail Address: jcliou.cv87g@.tw (J.C. Liou).1. Introduction Information on swell and wind waves is very important for marine activities and for safe navigation. Therefore, some engineers and scientists have devoted their time to the study of the accurate calculation of water waves in the sea. Initially simple empirical prediction models, such as SMB (Sverdrup-Munk-Bretschneider) and JONSWAP (Joint North Sea Wave Atmosphere Program), were proposed to provide a quick but defective estimation of ocean waves [1,2]. With the help of fast computer calculations and the knowledge of physical processes responsible for wave behavior, numeri