In view of the problem of inaccuracy of the fault separation caused by inaccurate modeling of the wind turbine variable pitch system

a fault separation method combining J divergence and multi-model is proposed.Firstly

the state of the system is estimated by particle filtering（PF） based on the multi-fault model of the wind turbine pitch system. Then a bandwidth coefficient is introduced to optimize the adaptive threshold to improve the real-time and accuracy of the fault detection. Finally

the J divergence of the output of the actual system and the output of each fault model are calculated

and determines the fault type and increase accuracy of fault separation.Research results show that the method significantly improve the accuracy of the fault diagnosis and reduce the calculation time.