Adaptive nonlinear filters for joint fault estimation and accommodation of a wind farm benchmark

In order to improve the availability of offshore wind farms, thus avoiding unplanned operation and maintenance costs, which can be high for offshore installations, the accommodation of faults in their earlier occurrence is fundamental. Therefore, this paper addresses the design of an active fault tolerant control scheme that is applied to a small wind park benchmark of nine wind turbines, based on their nonlinear models, as well as the wind and interactions between the wind turbines in the wind farm. The controller accommodation scheme exploits the on-line estimate of the fault signals generated by nonlinear filters using the nonlinear geometric approach to obtain estimates decoupled from both model uncertainty and the interactions among the turbines. This paper proposes also a data-driven approach to provide these disturbance terms in analytical forms, which are subsequently used for designing the nonlinear filters for fault estimation. The wind farm benchmark is considered to validate the performances of the suggested scheme in the presence of different fault conditions, modelling and measurement errors.