ABSTRACT
At present, the proportion of wind power in the power system is constantly increasing, which has become the third largest energy after thermal power and hydropower (Shen and Li 2013). However, the cost of maintenance is higher and higher with the growing of wind turbine monomer generating capacity. Statistically, the maintenance costs accounted for 25% ~ 30% of the total cost (Chen et al. 2011). Although the most wind field is equipped with a complete Supervisory Control And Data Acquisition (SCADA) monitoring system, most wind field does not contain the vibration test project, because the cost of conventional vibration diagnosis sensor is high and the problems of installation and maintenance are exist. In view of this situation, a variety of indirect methods which are used for fault diagnosis of wind turbine research are gradually increasing in recent years. For example, methods of use existing SCADA data for indirect diagnostic, modeling methods and electrical parameters, etc. (Chen et al. 2011) Among them, diagnostic method which fully uses a wind turbine electrical parameters (such as voltage, current, etc.) stands out, because it has low cost, easy to maintain, and the electrical parameters transducer can be centrally installed at the substation. This method does not need install vibration and other physical sensors in the equipment, only use voltage or current converter, so it is also known as non-sensor diagnostic method.
