Study On Fault Prediction For Wind Turbine

Wind energy is recognized worldwide as one of the sustainable energies and its development plays a key role in both the reduction of the emission of carbon dioxide and the sustainable development of the economics and society. However, with running time goes on, wind turbines’growing costs of maintenance and probability of fault and errors force us to improve the theories and technologies on data monitoring and fault prediction. Due to terrible working condition, the occurrence of wind turbines’fault, especially on the generator, gearbox, will not only decrease its occupating coefficient, but also bring down the active power, moreover, the prohibitive cost of maintenance causes unnecessary economic loss. Hence, it is necessary to develop a wind turbine monitoring and fault predicting system to protect faults and errors from happening. Currently, there are no one general data monitoring system that can integrate all the wind turbines since different wind turbines’data monitoring systems are separated from different wind turbine manufacturers, also, they can only monitor faults and errors in real-time instead of predict them before faults happen.This paper introduces and chooses the fault diagnosis and prediction algorithm of wind turbines at first, then summarizes the frequencies and kinds of each faults that happened within6months, and analyzes the characteristics and reasons of them. Also, this paper establishes the fault predicting model of generator bear temperature bases on history data which obtained from SCADA, and proves that GA-BP artificial neural network is feasible and effectual in predicting. At last, we introduce the design and implementation of integrated wind turbine remote monitoring system.