Power Curve Modeling for Wind Turbine using Artificial Intelligence Tools and Pre-Established Inference Criteria

This paper proposes a new way of developing non-parametric models of power curves, using artificial intelligence tools. Here, nine models are developed, one being parametric and eight non-parametric, to emulate the behavior dictated by the power curve of the wind farms under study. Also, a comparison between the power curve models based on ANNs and those based on Fuzzy logic is proposed. For this, some of the power curve models based on artificial neural networks (ANNs) are used and based on fuzzy inference systems (FISs) created and evaluated in [1] and later published in [2], as well as two new FISs with the proposed new heuristic. In order to do so, an initial “pre-training” is proposed, resulting from the characteristics derived from the expert's inference followed by a transformation of a Fuzzy Mamdani system into a Fuzzy Sugeno system. The results showed the new models of pre-trained FIS have a better precision when compared relation to ANN models and FIS models, although the presented values by the error indicators are comparable. The comparative study is carried out in two wind farms located in northeastern Brazil, and the proposed approach becomes a very relevant alternative for the improvement of power curve approximation based on a Fuzzy Inference System.