Fault—Tolerant Control of an Offshore Wind Farm via Fuzzy Modelling and Identification. Issue 21 (2015)

• Record Type: Journal Article
• Authors: Simani, S.; Farsoni, S.; Castaldi, P.
• Is Part Of: IFAC-PapersOnLine. Volume 48:Issue 21(2015)
• Page Start: 1345
• Page End: 1350
• Publication Date: 2015
• Subjects: Passive fault tolerant control system -- fuzzy modelling and identification -- data-driven scheme -- fault estimation -- offshore wind farm; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2015.09.712
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: In order to improve the safety, the reliability, and the efficiency of offshore wind park installations, thus avoiding expensive unplanned maintenance, the accommodation of faults in their earlier occurrence is fundamental. Therefore, the main contribution of this paper consists of the development of a fault tolerant control scheme by means of a viable approach. In particular, a data-driven strategy based on fuzzy logic is exploited for deriving the model of the required controller. Fuzzy theory is exploited here since it is able to approximate easily unknown nonlinear models and manage noisy measurements. Moreover, the controller fuzzy prototype is directly identified from the wind farm measurements, and it provides the straightforward achievement of the fault tolerant control scheme. In general, an analytic approach, where the system nonlinearity is explicitly taken into account, could require more complex design methodologies. This aspect of the work, followed by the simpler solution relying on fuzzy rules, represents the key point when on-line implementations are considered of the proposed control scheme. To highlight the potential of the proposed fault tolerant control algorithm in real applications, a hardware-in-the-loop test facility representing a realistic offshore wind farm installation is considered to analyse the digital implementation of the designed controller. The results have shown that the developed scheme maintains desired performances, thus validating its reliability also in real-time implementations.