Residual Generator Fuzzy Identification for Wind Farm Fault Diagnosis

In the recent years the wind turbine industry has focused on optimising the cost of energy. One of the important factors in the achievement of this task consists of increasing the reliability of the wind turbines, which can be obtained using advanced fault detection and isolation strategies. Clearly, most faults are managed quite easily at a wind turbine control level. However, some faults are better dealt with at wind farm level, when the wind turbine is located in a wind farm. This paper aims at proposing a fault detection and isolation solution with application to a wind farm benchmark model. The considered benchmark includes a small wind farm of nine wind turbines, based on simple models of wind turbines, as well as the wind and interactions between wind turbines in the wind farm. The fuzzy approach is proposed here since the model under investigation is nonlinear, whilst the measurements are highly noisy. Therefore, the solution relies on a set of piecewise affine Takagi Sugeno models, which are identified from the noisy measurements acquired from the simulated wind park. The design of the fault detection and isolation strategy is also enhanced by the use of the proposed fuzzy approach. Finally, the wind park simulator is exploited for validating the achieved performances of the suggested methodology.