Comparison of parametric modeling approaches for wind turbine power curves

As a sustainable and clean energy source, wind energy has a wide range of applications around the world. However, due to wind dynamics, the stochastic nature of wind energy and the operational characteristics of wind turbines, many difficulties are encountered in the stable operation of electrical power systems. One of the important ways to overcome these difficulties is to accurately model the wind turbine power curve. In this thesis, initially, wind turbine power curve models were created using metaheuristic optimization-based parametric methods, and compared in detail in terms of the goodness-of-fit statistics. At this stage, the design coefficients of 3-, 4- and 5- parameter logistic, 5th-, 6th- and 7th-degree polynomial and modified hyperbolic tangent functions were found by African vultures, slime mould, marine predators, Fick's law and geometric mean optimization algorithms. Among the power curve models developed, marine predators optimization algorithm-based modified hyperbolic tangent model showed the most effective performance in terms of the goodness-of-fit statistics. Afterwards, k-means clustering algorithm was integrated into this hybrid power curve model to obtain lower sum of squared errors and root mean squared error results, thus producing a more stable power curve model.