Search

Technological advances, especially in the industrial field, have led to the development and optimization of the activities that takes place on it. To achieve this goal, an early detection of any kind of anomaly is very important. This can contribute to energy and economic savings and an environmental impact reduction. In a context where the reduction of pollution gasses emission is promoted, the use of alternative energies, specially the wind energy, plays a key role. The wind generator blades are usually manufactured from bicomponent material, obtained from the mixture of two dierent primary components. The present research assesses dierent one-class intelligent techniques to perform anomaly detection on a bicomponent mixing system used on the wind generator manufacturing. To perform the anomaly detection, the intelligent models were obtained from real dataset recorded during the right operation of a bicomponent mixing plant. The classifiers for each technique were validated using artificial outliers, achieving very good results.