NARX Time Series Model for Remaining Useful Life Estimation of Gas Turbine Engines

Oguz Bektas and Jeffrey Jones
Submission Type: 
Full Paper
AttachmentSizeTimestamp
phmec_16_028.pdf1.76 MBJune 23, 2016 - 3:18am

Prognostics is a promising approach used in condition based maintenance due to its ability to forecast complex systems' remaining useful life. In gas turbine maintenance applications, data-driven prognostic methods develop an understanding of system degradation by using regularly stored condition monitoring data, and then can automatically monitor and evaluate the future health index of the system. This paper presents such a technique for fault prognosis for turbofan engines. A prognostic model based on a nonlinear autoregressive neural network design with exogenous input is designed to determine how the future values of wear can be predicted. The research applies the life prediction as a type of dynamic filtering, in which training time series are used to predict the future values of test series. The results demonstrate the relationship between the historical performance deterioration of an engine's prior operating period with the current life prediction.

Publication Year: 
2016
Publication Volume: 
7
Publication Control Number: 
028
Page Count: 
10
Submission Keywords: 
data driven prognostics
Neural Networks
NARX
multi step prediction
Submission Topic Areas: 
Data-driven methods for fault detection, diagnosis, and prognosis
Submitted by: 
  
 
 
 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed