Online Fault Prediction of Actuation System Based on BiGRU-EWMA
Aiming at the problems of low prediction accuracy and poor online prediction performance of traditional fault prediction methods for actuation systems under partially observable conditions,an online fault prediction method for actuation systems based on BiGRU-EWMA was proposed.Firstly,the Bi-directional Gated Recurrent Unit network was used to mine the bidirectional time series characteristics of the operating data,and the health state pre-diction model was established.Secondly,the Exponentially Weighted Moving-Average control chart was used to moni-tor the output residual between the predicted value of the state prediction model and the actual value of the modeling parameters to establish the monitoring index of the online fault trend and set the adaptive fault threshold.Finally,ex-periments were carried out on the aircraft electromechanical system simulation test platform.And compared with other methods,it is verified that this method can earlier predict the fault trend of the actuating system,and has higher accu-racy and rapidity than the existing methods.
Actuation systemOnline fault predictionBiGRU networkEWMA control chart
NETL
NSTL
万方数据
