基于BiGRU-EWMA的作动系统在线故障预测

Online Fault Prediction of Actuation System Based on BiGRU-EWMA

针对传统故障预测方法在作动系统的部分可观测条件下预测精度较低且在线预测性能较差的问题,提出一种基于BiGRU-EWMA的作动系统在线故障预测方法,首先采用双向门限循环网络(BiGRU)挖掘系统运行数据的双向时序特征,建立作动系统健康状态预测模型。然后采用指数加权移动平均(Exponentially Weighted Moving-Average, EWMA)控制图监测状态预测模型的预测值与建模参量实际值之间的输出残差,建立故障趋势的在线监测指标并设定自适应故障阈值。最后在飞机机电系统仿真测试平台上进行实验,并与其它方法进行比较,验证了上述方法能够更早的预测出作动系统出现的故障趋势,且在准确性和快速性上具有较好的优势。

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 prediction model was established.Secondly,the Exponentially Weighted Moving-Average control chart was used to monitor 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,experiments 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 accuracy and rapidity than the existing methods.