Predicción y detección de averías en aerogeneradores a partir de datos scada.

Résumé

This article proposes a method for predicting and detecting potential failures in the main components of a wind turbine, based on data collected by a built-in SCADA (Supervisory Control And Data Acquisition) monitoring system. Artificial intelligence techniques such as neural networks and boosted trees are used to model the behaviour of the system and to select optimal input parameters. Once the method has been defined, it is applied to an actual case study of a malfunction in a multiplier, with data from a wind farm located in la Rioja (Spain), owned by Grupo Eólicas Riojanas (GER). The combination of a detailed study of the optimal parameters for modelling the specific behaviour of the temperature of the multiplier and the development of a model based on neural networks enables the normal behaviour of a multiplier to be modelled effectively with no deterioration in its operation. This means that the potential processes of deterioration in the multiplier can be analysed regularly and consequently actions can be taken accordingly before an irreparable malfunction requiring replacement.