Neuro-adaptive model-reference fault-tolerant control with application to wind turbines

This work investigates the model-following control problem associated with a class of non-linear systems in the presence of modelling uncertainties and actuator failures. The particular interest lies in the development of designer-friendly and cost-effective control scheme. By combining model-reference mechanism with robust adaptive radial basis function (RBF) neural network (NN), several control algorithms are derived without the need for precise system parameters or analytical-bound estimation on actuator failure variables. It is shown that the developed control algorithms are structurally simple and computationally inexpensive. Application of the proposed strategies to individual pitch control of wind turbines is also addressed. Formative stability analysis and numerical simulation on severe failure scenarios confirm the effectiveness of the proposed methods.