Abstract：Aiming at the vibration problem of wind turbine towers caused by natural factors or actions of wind

turbine components，a wind turbine tower vibration warning model based on fuzzy inference optimization was

proposed. Firstly，based on the data from the wind turbine supervisory control and data acquisition（SCADA）

system，variables with a high degree of correlation with the vibration of the wind turbine tower were used to

calculate the velocity class Mahalanobis distance and the non-velocity class Mahalanobis distance of the normal

vibration state，and three different vibration state thresholds were determined. Then，the fuzzy sets were divided

according to the thresholds of different vibration states，the affiliation functions were established. Finally，the

particle swarm optimization algorithm was used to optimize the parameters of the affiliation functions to obtain the

fuzzy optimization inference model，and the output numerical value line graphs were used to judge whether the

vibration of the wind turbine tower is alarmed. The results show that by applying the optimization model proposed

for offline training，the vibration warning for wind turbine towers is carried out 2.47 h in advance on average，with

a vibration anomaly detection rate of 90% and a false alarm rate of 40%. The online application of the proposed

model can provide effective vibration warnings for wind turbine towers，however，there is a certain false alarm rate.

      Key words：condition monitoring；tower vibration；Mahalanobis distance；fuzzy inference；particle swarm

optimization（PSO）