Fault prognosis of Engineered Systems: A Deep Learning Perspective

In modern industry, engineered systems are generally required to work under complex operational conditions to complete specific missions. But most of existing data-driven prognostic methods still lack an effective model that can utilize operational conditions data to predict remaining useful life (RUL) of engineered systems. To fill these practical gaps, this paper develops a novel prognostic method based on Bi-Directional Long Short-Term Memory (BLSTM) network. The proposed method can effectively integrate multiple sensors data with operational conditions data for RUL prediction of engineered systems. In the proposed prognostic framework, multiple raw sensors data, operational conditions data, and inspection time epoch are preprocessed to form the desired sequences data with fixed length, and then are taken as main inputs and auxiliary inputs to fed into proposed BLSTM based network. And the labeled RUL are used as actual output for training the proposed model. Different from other deep learning (DL) based prognostic methods, the developed method can simultaneously model both sensors data and operational conditions data in a consolidated framework. The proposed approach is validated through a case study on aircraft turbofan engine, and comparisons with other existing state-of-the-art methods are also included.

[1]  L. Peel,et al.  Data driven prognostics using a Kalman filter ensemble of neural network models , 2008, 2008 International Conference on Prognostics and Health Management.

[2]  Jie Liu,et al.  A multi-step predictor with a variable input pattern for system state forecasting , 2009 .

[3]  Zhigang Tian,et al.  An artificial neural network method for remaining useful life prediction of equipment subject to condition monitoring , 2012, J. Intell. Manuf..

[4]  Kay Chen Tan,et al.  Multimodal Degradation Prognostics Based on Switching Kalman Filter Ensemble , 2017, IEEE Transactions on Neural Networks and Learning Systems.

[5]  Xiang Li,et al.  Remaining useful life estimation in prognostics using deep convolution neural networks , 2018, Reliab. Eng. Syst. Saf..

[6]  H.A. Toliyat,et al.  Condition Monitoring and Fault Diagnosis of Electrical Motors—A Review , 2005, IEEE Transactions on Energy Conversion.

[7]  Xi Zhang,et al.  Multiple Sensor Data Fusion for Degradation Modeling and Prognostics Under Multiple Operational Conditions , 2016, IEEE Transactions on Reliability.

[8]  Kay Chen Tan,et al.  Multiobjective Deep Belief Networks Ensemble for Remaining Useful Life Estimation in Prognostics , 2017, IEEE Transactions on Neural Networks and Learning Systems.

[9]  Jürgen Schmidhuber,et al.  LSTM: A Search Space Odyssey , 2015, IEEE Transactions on Neural Networks and Learning Systems.

[10]  Li Lin,et al.  Remaining useful life estimation of engineered systems using vanilla LSTM neural networks , 2018, Neurocomputing.

[11]  Abhinav Saxena,et al.  Damage propagation modeling for aircraft engine run-to-failure simulation , 2008, 2008 International Conference on Prognostics and Health Management.

[12]  Ruqiang Yan,et al.  Learning to Monitor Machine Health with Convolutional Bi-Directional LSTM Networks , 2017, Sensors.

[13]  Mayorkinos Papaelias,et al.  Condition monitoring of wind turbines: Techniques and methods , 2012 .

[14]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[15]  Xiaoli Li,et al.  Deep Convolutional Neural Network Based Regression Approach for Estimation of Remaining Useful Life , 2016, DASFAA.

[16]  Fatih Camci,et al.  Failure prediction on railway turnouts using time delay neural networks , 2010, 2010 IEEE International Conference on Computational Intelligence for Measurement Systems and Applications.