Health‐aware model predictive control of wind turbines using fatigue prognosis

Abstract

Wind turbines components are subject to considerable fatigue due to extreme environmental conditions to which are exposed, especially those located offshore. Interest in the integration of control with fatigue load minimization has increased in recent years. The integration of a system health management module with the control provides a mechanism for the wind turbine to operate safely and optimize the trade-off between components life and energy production. The research presented in this paper explores the integration of model predictive control (MPC) with fatigue-based prognosis approach to minimize the damage of wind turbine components (the blades). The controller objective is modified by adding an extra criterion that takes into account the accumulated damage. The scheme is implemented and tested using a high fidelity simulator of a utility scale wind turbine.

A review of failure prognostics for predictive maintenance of offshore wind turbines

“…Stress data is either empirical or derived from simulation models. Except for the two methods, there are several self-developed nonlinear dynamic roughness models for wear process [31], stiffness model for blade bending [32], and strain model for monopiles [33]. These fatigue models emphasize physical features such as torsion, strain, stress, and stiffness and can describe the structural failure in depth.…”

“…The frequently used model is a 5MW offshore wind turbine FAST (aero-hydro-servo-elastic simulator) model developed by the National Renewable Energy Lab (NREL) [36]. For example, Sivalingam et al employ the codes of an induction machine [24] while Sanchez et al use the baseline controllers in the FAST model [32]. Other simulation models include multi-body simulation, finite element modeling (FEM), torsional model, and bond graph [25,27,30].…”

A Simple Nonlinear Observer for State and Unknown Input Estimation: DC Motor Applications

“…modeling [1], control design [2], state/parameter estimation [3]) or at the operation stage (i.e. fault diagnosis [4], fault tolerant control [5], prognosis [6]). When a DC motor operates, it is subject to different operating conditions (predictable or not) caused by exogenous variables, disturbances, noise or faults, which can be induced by many causes such as vibrations, friction, overload, and voltage variations.…”

Remaining Lifetime Prediction for Reliability-based Dry Clutch Control

“…For example in (Salazar, Nejjari, & Sarrate, 2014) an MPC with integrated wear model was used for reliabilitybased control of a twin rotor system. In (Sanchez, Escobet, Puig, & Odgaard, 2017) a reliability-based control was realized using a MPC for a wind turbine system with linearisation of calculated load. Finally, the literature review reveals two open research issues: a model-based approach for lifetime prediction with identified model parameters and reliability-based control for clutches.…”