Design and reliability analysis of rubble mound breakwaters by using artificial neural networks, fuzzy logic systems and genetic algorithm

m DESIGN AND RELIABILITY ANALYSIS OF RUBBLE MOUND BREAKWATERS BY USING ARTIFICIAL NEURAL NETWORKS, FUZZY LOGIC SYTEMS AND GENETIC ALGORITHM (Ph. D Thesis) M. Levent KOÇ GAZI UNIVERSITY INSTUTE OF SCIENCE AND TECHNOLOGY October 2002 ABSTRACT In this study, artificial intelligence techniques were applied to the preliminary design of rubble mound breakwaters and the results obtained were compared with conventional methods. Wave parameters of Alanya region (significant wave height, period and direction) were simultaneously predicted by using feed forward and recurrent neural networks. Artificial intelligence techniques which include feed forward neural networks, fuzzy systems, fuzzy neural networks and genetic algorithm were applied to the stability and reliability analyses of Mersin yacht harbor main breakwater, as a case study. It was found that, in predicting wave parameters, a better performance can be obtained by neural networks than stochastic models and that recurrent neural networks were appropriate for multi step predictions. A "design artificial neural network" which uses Van der Meer's hydraulic model test data, was developed for the preliminary design of rubble mound coastal structures and it was verified for the design applications. Furthermore it has been showed that, with the application of genetic algorithms in the reliability-based design, more reliable results can be obtained for the optimum global solutions, when compared to the second-order methods. Artificial intelligence techniques can handle more accurately the uncertainties inherent in the design of rubble moundIV breakwaters; hence the need of complex models generally used for the design has been significantly decreased. Science Code : 624.02.03 Key Words : Artificial intelligence, breakwaters, neural networks Page Number : 157 Adviser : Ass. Prof. Dr. Can £. Balas : Prof. Dr. A. Samet Arslan