2026-08-24T05:19:02Zhttps://docta.ucm.es/rest/oai/request

oai:docta.ucm.es:20.500.14352/1081372025-03-10T18:34:25Zcom_20.500.14352_14col_20.500.14352_15

00925njm 22002777a 4500 dc Muñoz-Palomeque, Eduardo author Sierra-García, Jesús Enrique author Santos Peñas, Matilde author 2023 The main control goal of a wind turbine (WT) is to produce the maximum energy in any operating region. When the wind speed is under its rated value, the control must aim at tracking the maximum power point of the best power curve for a specific WT. This is challenging due to the non-linear characteristics of the system and the environmental disturbances it is subjected to. Direct speed control (DSC) is one of the main techniques applied to address this problem. In this strategy, it is necessary to design a speed controller to adjust the generator torque so to follow the optimum generator speed. In this work, we improve the DSC by implementing this speed controller with a radial basis function neural network (NN). An unsupervised learning algorithm is designed to tune the weights of the NN so it learns the control law that minimizes the generator speed error. With this proposed unsupervised neural control methodology, the electromagnetic torque that allows the optimal power extraction is obtained, and thus the best power coefficient (⁠ ⁠) values. The proposal is tested on the OPENFAST non-linear model of the National Renewable Energy Laboratory 1.5 MW WT. Simulation results prove the good performance of this neuro-control approach as it maintains the WT variables into the appropriate range and tracks the rated operation values. It has been compared with the controller included in OPENFAST giving up to 7.87% more power. Muñoz-Palomeque E, Sierra-García JE, Santos M. Wind turbine maximum power point tracking control based on unsupervised neural networks. Journal of Computational Design and Engineering. 2023 Feb;10(1):108-21. https://10.1093/jcde/qwac132 https://hdl.handle.net/20.500.14352/108137 Wind turbine maximum power point tracking control based on unsupervised neural networks