Fuzzy logic modeling of the dissolved oxygen fluctuations in Golden Horn

Abstract

The modern modeling techniques show a significant progress in recent years. Fuzzy logic approach is one of these methods that can be used for forecasting purposes and identification of complex systems. In this study, it is aimed to model monthly dissolved oxygen (DO) amount variations. Firstly, regression technique is used to remove the trend from the actual dissolved oxygen time series. Detrended time series is then modeled with Takagi–Sugeno Fuzzy logic approach. The monthly historical records of DO as considered in this paper provide a fundamental data exhibiting persistence. It is this dependence that gives opportunity for the serial modeling of the data sequence concerned. These models are the prime and sole means to derive the future likely excedences over the historical values and hence provide a chance for the future planner with decision variables such as the minimum DO amounts and their persistence for some duration. In the scope of this study, DO concentration changes in the two stations located in Golden Horn at the 0.5 m (upper layer) depth were modeled. As a result of the study, it is seen that one can forecast the next month's DO amount from antecedent measurements within an acceptable relative error limits.