Detecção de falhas de curto-circuito entre espiras de aerogeradores por meio de sinais de vibração

Abstract

Monitoring wind turbines in order to avoid failure in early stages avoids defects that can cause accidents, especially fires. This paper presents a methodology for monitoring and detecting short circuits in wind turbines coils. The vibration signal from the generator is collected with the MEMS accelerometers. A structural signal analysis technique was used to identify vibration signals patterns and a classifier was used to characterize the different machine conditions. The result was evaluated in a wind turbine simulation system, composed of a group of motor-generator controlled by a frequency converter. Using a recommendation algorithm, the detected patterns are analyzed to provide information about the condition of the equipment. In the performed experiments, we achieved a minimum accuracy of 86% and no false-positives.