An exact mathematical description of the wave overtopping processes is impossible due to the complex nature of the processes. Therefore the dependency of overtopping from wave parameters and coastal structures was mostly studied by physical model tests. To avoid the uncertainties due to imperfect statistics of wave heights in the irregular wave trains performed in physical models, the mean overtopping rates of irregular waves can be determined by the probability calculation method (PCM) (Goda, 2000) based on the regular wave data. The PCM combined with an artificial neural network (NN) technique is proposed in this paper to determine the mean overtopping rates of irregular waves on coastal structures based on learning from the regular wave data. The NN is used to quantify the overtopping volumes for the individual waves and the PCM is used for calculating the cumulative wave effect of individual waves of random nature. Determination of wave overtopping at a vertical wall with a parapet is presented as an application of the present model. Good agreement with the available experimental data and the empirical formulas shows that the present model offers an alternative to determine the mean overtopping rates of irregular waves on coastal structures. The method itself allows an insight in the reasons and the extent of scatter to be expected in physical model tests.
Tsai, Ching-Piao; Daemrich, Karl-Friedrich; and Ho, Chuen-Lin
"PROBABILITY CALCULATION METHOD WITH NEURAL NETWORK FOR ESTIMATING WAVE OVERTOPPING AT COASTAL STRUCTURES: LEARNING FROM REGULAR WAVE TESTS,"
Journal of Marine Science and Technology: Vol. 24:
3, Article 10.
Available at: https://jmstt.ntou.edu.tw/journal/vol24/iss3/10