Design of coastal structures and estimation of wave height by artificial intelligence and time series methods

Our country surrounded by seas on three sides with 8333 kilometres long coastline can be classified as a sea country. Therefore, necessary structures should be constructed to provide the trade, tourism and fishing industries' operations efficiently and reliably along the coasts. In the design of coastal and marine structures, Hs (significant wave height) is the most important wave parameter having a direct effect on the stability of the structures throughout their economic lives. In this study, Hs (significant wave height), one of the most important parameters used in the design of coastal structures, was obtained in two ways. These are the Turkish Coasts' Wind and Deep Sea Wave Atlas and the ECMWF (European Centre For Medium-Range Weather Forecasts). ECMWF was determined to be appropriate for the deep sea significant wave height source by putting forward the reasons for the preference. The significant wave height in front of the structure was obtained by HYROTAM-3D software using the deep sea significant wave height wave transformation module. A rubble-mound breakwater design with three different single layer protection layer was made using the obtained design wave. For this purpose, a single layer of protection with Accropode II or Xbloc can be used. However, for the use of these layers, our country has to pay high amount of patent fees abroad. For this reason, the design with newly developed Piblok artificial protection layer was applied in the thesis. The wave height having great importance with its dynamic effect in the design of coastal structures has been mostly determined by numerical and stochastic methods in the past. In recent years, it has been frequently used in studies such as wave parameter estimations and missing wave data with the development of artificial intelligence techniques. In this study, it was concluded that the modeling with ANFIS methods could be used for incomplete, future-oriented or incorrect wave height estimation. In addition, by using the same data, the model was estimated by using autocorrelation and partial autocorrelation graphs for the appropriate model in time series analysis modeling. The validity of the AR(2) model, which was found suitable, was investigated, and a method was obtained to obtain a prospective estimation. In order to compare the models, the scatter diagrams of the predicted values of each model and the measurement values were drawn. In addition, it is considered that it is appropriate to use both methods in the prospective estimation of the wave height parameter used in the design of coastal structures, since R2 (regression coefficient) and NSE (Nash-Sutcliffe efficiency) for their statistical performance gave results close to % 90 with both methods. It has been concluded that the significant wave height can be estimated reliably using these two models proposed in this study.