Fuzzy Regressıon Analysıs and An Applıcatıon

Abstract

Wind energy, one of the renewable energy sources, is immensely popular today due to the increase in environmental awareness, the decrease in the number of fossil fuels, and the increase in the cost of these fuels. Although wind energy is a clean and nature-friendly energy source, the wind is not a continuous energy source. In addition, the establishment of farms that convert wind energy into electrical energy is expensive and requires technical capacity. Determining the locations where wind turbine farms will be established, which will provide long-term profit to its investors and require considerable amounts of financing at the beginning, is significantly vital in terms of the economic use of resources. It is necessary to collect many meteorological data such as wind speed, wind direction, air density, temperature, pressure, and relative humidity from at least one year ago at the stage of determining the wind turbine construction locations. The wind turbine manufacturer creates theoretical information about how much electrical energy the turbine will generate at what wind speed. Following the collection of meteorological data, various numerical and statistical models are made with the help of the theoretical electricity generation data, and the suitability of the construction location is evaluated. However, when similar wind turbines are examined, it will be seen that there are differences between the theoretical production amount given by the manufacturer and the actual amount of electricity produced at the same wind speed. In this condition, it is clear that there is a fuzzy relationship between wind speed and the electrical energy produced. For this thesis, the amount of electrical energy to be produced by a wind turbine is estimated by using only wind speed or wind speed and wind direction data with fuzzy linear regression methods. In addition, the amount of produced electrical energy and the wind speed in the data set are fuzzified. Succeeding, crisp input crisp output, crisp input fuzzy output, and fuzzy input fuzzy output situations were estimated with four different fuzzy regression methods and the results were compared. This application is intended to determine the general framework for the locations where the wind turbine is planned to be installed before the complex calculations and modeling, or when seasonal observations are made rather than annual, or in cases where the observed values are not dependable or there are many site alternatives but there is not enough time to decide on site selection. It has been determined that it will be beneficial in situations. Therefore, it will bring a different approach to the literature. Finally, this study will open a new window to the methods by establishing the basis for the Fuzzy Partial Regression Method and Fuzzy Nonlinear regression methods that are expected to be used in the future in estimating the energy produced by wind turbines.