The hydraulic variable-pitch system is one of the components of wind turbines frequently prone to faults. Early fault detection of the hydraulic variable-pitch system can effectively improve operation reliability and reduce unnecessary economic loss. Wind turbines suffer much environmental interference

while data-based fault detection is apt to lead to false alarms influ-enced by these factors. In addition

fault isolation is considerably hard. For this reason

a state estimation approach is proposed to remedy the situation. First

a state space model and state observer of hydraulic variable-pitch system are developed to obtain the residual. Secondly

the residual root mean square(RMS)is used as the residual evaluation function

and an adaptive threshold is used to make decisions. In the end

by simulating the hydraulic variable-pitch actuator fault and pitch-angle sensor fault

the ef-fectiveness of the proposed fault detection approach is verified. This work is supported by National Natural Science Foundation of China(No. 51277074).