Contents I. Introduction
Wind turbines are often installed in clusters called wind farms, providing large-scale energy supply. An individual offshore wind farm can produce enough power to supply the annual electricity demand of thousands of households. Managing such wind power (particularly offshore) means higher complexity, limited accessibility and harsher climate conditions, resulting in higher rates of faults and failures. This issue highlights the essential need for applying advanced fault detection and diagnosis (FDD) and fault-tolerant control (FTC) strategies to wind turbines for improved reliability and availability. Depending on the type and nature of faults, such strategies can be developed at both individual wind turbine and entire wind farm levels. Most of the recent works have rather focused on the application of FDD and FTC at individual wind turbine level (for example see [1]–[5]). At a wind farm level, the available literature is relatively scarce. The majority of works are limited to condition monitoring and fault detection only. For instance, various data mining methods are used to improve models for fault detection/prediction in wind farms [6]–[8]. Different machine learning models have been developed to estimate the relationship between the generated power of a wind farm and wind speed [9]. However, the models in [9] are incapable of isolating and identifying faults in wind farm. A data-driven set based on Takagi-Sugeno's model is proposed by [10] for fault detection and isolation. Authors in [11] present a fault diagnosis approach by interval parameter-varying parity equations considering some noises and modelling errors. In [12], an active FTC scheme is performed based on a model-based FDD method. All of the aforementioned works are unable to handle multiple, simultaneous faults in a wind farm. Moreover, they depend on the wind speed or its estimation which is normally related to the wind direction and the wind farm layout. More recently, the authors in [13] and [14] present different schemes of FTC in a cooperative framework referred to as fault-tolerant cooperative control (FTCC). The FTCC schemes do not depend on speed and direction of wind and they can accommodate multiple simultaneous faults in more than one wind turbine in a farm with arbitrary layouts. This paper aims to extend authors’ works in [13] and [14] by introducing a novel FTCC scheme based on an efficient model reference adaptive proportional-integral (PI) control reconfiguration approach that is augmented with an innovative control reallocation mechanism in a cooperative framework. Contrarily to the works cited in [13] and [14], the proposed scheme in this paper not only acts against mild power loss faults due to mild icing or debris build-up on rotor blades, but also handles the effects of severe power loss faults due to heavy icing. Also, the proposed FTCC scheme is independent of any explicit unit for FDD, and the information of the fault effects is implicitly conceived within the cooperative control loops. In addition to the abovementioned benefits, the fact that this scheme employs a type of PI control approach makes it more industry-friendly, as opposed to other cited approaches in the literature.
