Degradation modeling and RUL prediction using Wiener process subject to multiple change points and unit heterogeneity

Significance 

Rapid developments in areas relating to information and sensing technologies will offer great opportunities in real-time degradation monitoring and prediction in various modern systems. Degradation signals, such as the vibrational signal of rotating machinery and the luminosity of LED lamps – among others, are commonly used for system reliability assessment due to their direct relation with underlying physical degradation processes.

So far, the developed degradation and remaining useful life of models are either physics-based or data-driven based. The latter has received more attention owing to the unprecedented data availability in recent times, but its models has inherent drawbacks relating to the degradation path. As for the former, they possess nice mathematical processes and physical interpretations, and thus their application is preferred. In general, the applicability of the physics- or data-driven based models is impeded by a set of challenges. Particularly, the degradation signals show multiple phases, whereas the conventional degradation models are often inadequate. Therefore, it is imperative that these shortcomings be resolved.

Recently, Ms Yuxin Wen (PhD candidate at University of Texas at El Paso) and Dr. Jianguo Wu at Peking University in China in collaboration with Dr. Devashish Das at University of South Florida with Professor Tzu-Liang (Bill) Tseng at University of Texas at El Paso developed a new model where they incorporated change points to the stochastic process for prognostic improvement. To be precise, they proposed a multiple change-point Wiener process as a degradation model to better characterize the degradation signals of multiple-phase characteristics. Their work is currently published in the research journal, Reliability Engineering and System Safety.

In brief, the research method they employed commenced with the presentation of a Wiener process degradation model with multiple change points with prior parameters specification and estimation provided. Next, the research team addressed and presented the technical details on how to sequentially update the posterior distributions of the current phase, latest change point, and Wiener process parameters of the current phase, and how to predict the remaining useful life (RUL). Lastly, they demonstrated the effectiveness and accuracy of their proposed technique through comprehensive simulation and real case study.

The authors observed that, in the simulation stage, using the leave-one-out cross-validation approach, the identified optimal change-point number for each dataset was equivalent to the true value. This effectively demonstrated the efficacy of the proposed approach for model selection. In addition, they deduced that specifying two change points only improved the prediction within the sudden-jump phase and had a direct association to the systems they used.

In summary, Yuxin Wen and her colleagues successfully presented the development of a Bayesian multiple change-point Wiener process for degradation modeling and online remaining useful life prediction. Generally, all the model parameters except the number of change points were modeled with random distributions so as to take into account the unit heterogeneity. Altogether, simulation and real case studies demonstrated that the proposed prognostic framework could effectively improve the prediction accuracy.

About the author

Yuxin Wen received the B.S. degree in Medical Informatics Engineering from Sichuan University, Sichuan, China in 2011, the M.S. degree in Biomedical Engineering from Zhejiang University, Zhejiang, China in 2014. Currently, she is pursuing the Ph.D. degree in Electrical and Computer Engineering at the University of Texas at El Paso (UTEP), TX, USA. Her research interests are focused on statistical modeling, prognostics and reliability analysis.

About the author

Jianguo Wu is an Assistant Professor in the Dept. of Industrial Engineering and Management at Peking University, Beijing, China. He was an Assistant Professor at the Dept. of IMSE at UTEP, TX, USA from 2015 to 2017. He received the B.S. degree in Mechanical Engineering from Tsinghua University, Beijing, China in 2009, the M.S. degree in Mechanical Engineering from Purdue University in 2011, and M.S. degree in Statistics in 2014 and Ph.D. degree in Industrial and Systems Engineering in 2015, both from University of Wisconsin-Madison.

His research interests are focused on statistical modeling, monitoring and analysis of complex processes/systems for quality control and productivity improvement through integrated application of metrology.

About the author

Devashish Das received a B.Tech. (Honors) degree in manufacturing science and engineering from the Indian Institute of Technology – Kharagpur, India, in 2010, and a Ph.D. degree in industrial engineering from the University of Wisconsin – Madison, WI, USA, in 2015. Currently, he is an Assistant Professor with the Department of Industrial and Management Systems Engineering, University of South Florida, Tampa, FL, USA.

His research interest involves integration of statistics, operations research and information theory to develop statistical monitoring methods, with a focus on advancing the science of health care delivery.

About the author

Tzu-Liang (Bill) Tseng is a Professor and Chair of Department of IMSE at UTEP. He received his M.S. degree in Industrial Engineering from the University of Wisconsin-Madison in 1993 and 1995 respectively and Ph.D. in Industrial Engineering from the University of Iowa in 1999.

His research area cover quality assurance in additive manufacturing, industrial data analytics and cyber based decision support systems.

Dr. Tseng is currently serving as an editor of Journal of CSI and editor boards of JDMMM and AJIBM. He is currently a Senior Member of IISE, SME and the Program Chair of Manufacturing Division of ASEE.

Reference

Yuxin Wen, Jianguo Wu, Devashish Das, Tzu-Liang (Bill) Tseng. Degradation modeling and RUL prediction using Wiener process subject to multiple change points and unit heterogeneity. Reliability Engineering and System Safety, volume 176 (2018) page 113–124.

Go To Reliability Engineering and System Safety

Check Also

Advancing Fusion Energy: High-Field REBCO Superconducting Magnets in the SPARC TFMC Program - Advances in Engineering

Advancing Fusion Energy: High-Field REBCO Superconducting Magnets in the SPARC TFMC Program