Hierarchical Robust Adaptive Control for Wind Turbines with Actuator Fault

TL;DR

This paper presents a hierarchical robust-adaptive control strategy for wind turbines that effectively manages rotor speed regulation amidst actuator faults, wind disturbances, and model nonlinearities, ensuring stability and improved performance.

Contribution

It introduces a two-stage control design combining an $ ext{L}_2$ controller and an adaptive low-level controller to handle uncertainties and faults in wind turbine systems.

Findings

Significantly reduces rotor speed error RMS compared to existing methods.

Ensures finite-gain $ ext{L}_2$ stability at the high level.

Achieves global asymptotic stability in the low-level error dynamics.

Abstract

This paper solves the problem of regulating the rotor speed tracking error for wind turbines in the full-load region by an effective robust-adaptive control strategy. The developed controller compensates for the uncertainty in the control input effectiveness caused by a pitch actuator fault, unmeasurable wind disturbance, and nonlinearity in the model. Wind turbines have multi-layer structures such that the high-level structure is nonlinearly coupled through an aggregation of the low-level control authorities. Hence, the control design is divided into two stages. First, an $\mathcal{L}_2$ controller is designed to attenuate the influence of wind disturbance fluctuations on the rotor speed. Then, in the low-level layer, a controller is designed using a proposed adaptation mechanism to compensate for actuator faults. The theoretical results show that the closed-loop equilibrium point of the regulated rotor speed tracking error dynamics in the high level is finite-gain $\mathcal{L}_2$ stable, and the closed-loop error dynamics in the low level is globally asymptotically stable. Simulation results show that the developed controller significantly reduces the root mean square of the rotor speed error compared to some well-known works, despite the largely fluctuating wind disturbance, and the time-varying uncertainty in the control input effectiveness.