RUL Prediction of Device Based on Multi-scale Working Condition Enhancement Network and Informer
It is of great significance for the remaining useful life(RUL)prediction of devices to improve their reliability safety,and reducing maintenance costs.By discovering the health status and potential faults of devices in advance,RUL prediction helps to reduce the risk of sudden failure,extend device life,improve work efficiency,and ensure the normal operation of tasks.However,with the increasing complexity of devices,and the collected sensor data have increasingly higher dimensions,traditional methods and some deep learning methods have limitations in processing the feature relationships,long time series data and mining important sensor data.Based on the multi-scale work condition enhancement network(MWCEN)and Informer model,this paper proposes a hybrid model of MWCEN-Informer to improve the prediction accuracy.The MWCEN encodes the device time series data by using the dynam-ic work condition coding algorithm,fully extracts the feature information by performing one-dimensional multi-scale hybrid convolu-tion on the device sensor information,enhances the effective features by using the multi-branch channel attention mechanism,inputs the enhanced sensor data into the Informer model to analyze the correlation of the device sensor timing data,and achieves more accu-rate RUL prediction of the device.Validation is carried out on a generic turbofan engine data set based on C-MAPSS,the results show that the model reduces the RMSE by an average of 5.5％and the S-Score by an average of 4.7％on the four subsets,which effective-ly improves the RUL prediction accuracy of the device under complex operating conditions and complex faults.
NETL
NSTL
万方数据
