Ensemble of optimized echo state networks for remaining useful life prediction

NEUROCOMPUTING

卷:281

Article

Rigamonti, Marco¹  Baraldi, Piero¹  Zio, Enrico^1,2  Roychoudhury, Indranil³  Goebel, Kai⁴  Poll, Scott⁴ 

[1] Politecn Milan, Energy Dept, Via Ponzio 34-3, I-20133 Milan, Italy

[2] Univ Paris Saclay, Cent Supelec, Fdn Elect France EDF, Chair Syst Sci & Energy Challenge, F-92290 Chatenay Malabry, France

[3] Stinger Ghaffarian Technol Inc, NASA, Ames Res Ctr, Moffett Field, CA 94035 USA

[4] NASA, Ames Res Ctr, Moffett Field, CA 94035 USA

关键词: Echo state networks;    Recurrent neural networks;    Ensembles;    Prediction uncertainty;    Prediction Intervals;    Differential Evolution;   

DOI  :  10.1016/j.neucom.2017.11.062

来源: Elsevier

【 摘 要 】

The use of Echo State Networks (ESNs) for the prediction of the Remaining Useful Life (RUL) of industrial components, i.e. the time left before the equipment will stop fulfilling its functions, is attractive because of their capability of handling the system dynamic behavior, the measurement noise, and the stochasticity of the degradation process. In particular, in this paper we originally resort to an ensemble of ESNs, for enhancing the performances of individual ESNs and providing also an estimation of the uncertainty affecting the RUL prediction. The main methodological novelties in our use of ESNs for RUL prediction are: i) the use of the individual ESN memory capacity within the dynamic procedure for aggregating of the ESNs outcomes; ii) the use of an additional ESN for estimating the RUL uncertainty, within the Mean Variance Estimation (MVE) approach. With these novelties, the developed approach outperforms a static ensemble and a standard MVE approach for uncertainty estimation in tests performed on a synthetic and two industrial datasets. (C) 2017 Elsevier B.V. All rights reserved.

【 授权许可】

Free