Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


29296.pdf (3.98 MB)

ETD Abstract Container

Abstract Header

Enhanced System Health Assessment using Adaptive Self-Learning Techniques

Di, Yuan
http://orcid.org/0000-0002-7783-851X
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1522420412871182

Abstract Details

2018, PhD, University of Cincinnati, Engineering and Applied Science: Mechanical Engineering.
System health assessment, as one of the most critical tasks in industrial data analytics, focuses on determining the current health condition and detecting the incipient fault. Recently, it has been challenging that the conventional strategy, which relies on a static health reference model along with a fixed threshold, is asked to fulfill the assessment requirements in the nonstationary monitoring environment. The dynamic data contexts might bring incorrect health estimation to the system. This dissertation presents an enhanced systematic online health assessment approach with adaptive self-learning techniques. The method enables the identification of novel working condition states, such as new rotating speed or processing recipe, and the recognition of new degradation extent in the arriving monitoring data, and includes them into the prior learning models. Hence, such continuously growing model could achieve the assessment more efficiently and accurately. This research work proposes the methodology of the enhanced health assessment approach, along with detailed technologies utilized in each implementation step, including a self-learning technique, a change detection and recognition strategy, and a clustering algorithm. Through a toy case on a rotor test bed, the dissertation intuitively described the detailed assessment process and demonstrated that the proposed approach, compared with the static model solution, could successfully capture the newly encountered patterns in the testing data. The feasibility of the proposed approach was demonstrated by two industrial use cases. For the semiconductor manufacturing process monitoring case, the proposed approach was able to correctly estimate the health states of the data measured from different experiments while being trained by one experiment observations. Additionally, it surpassed two existed assessment methods with higher overall assessment accuracy. For the power electronics modules monitoring case, the proposed approach also demonstrated its capabilities estimating the components’ health states under the situation of dynamic control modes and various units. It outperformed benchmarked unsupervised assessment methods and even obtained competitive results compared with supervised learning solutions. This study presents that the proposed approach with adaptive self-learning techniques could embrace a wide range of applications when the system health assessment is employed in an inconsistent and dynamic monitoring environment. Looking forward, more investigations would be contributed considering more complicated monitoring situations such as continuous working condition variations and incipient fault detection with multiple failure modes. Furthermore, the proposed approach provides a library of degradation patterns under various working conditions so that it can be potentially used to estimate the system remaining useful life.
Jay Lee, Ph.D. (Committee Chair)
Thomas Richard Huston, Ph.D. (Committee Member)
Jay Kim, Ph.D. (Committee Member)
Manish Kumar, Ph.D. (Committee Member)
139 p.
Mechanical Engineering
Health Assessment; Self Learning; Adaptive Learning; Pattern Recognition; Machine Learning

Recommended Citations

Citations

  • Di, Y. (2018). Enhanced System Health Assessment using Adaptive Self-Learning Techniques [Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1522420412871182

    APA Style (7th edition)

  • Di, Yuan. Enhanced System Health Assessment using Adaptive Self-Learning Techniques. 2018. University of Cincinnati, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1522420412871182.

    MLA Style (8th edition)

  • Di, Yuan. "Enhanced System Health Assessment using Adaptive Self-Learning Techniques." Doctoral dissertation, University of Cincinnati, 2018. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1522420412871182

    Chicago Manual of Style (17th edition)

ucin1522420412871182
460
© 2018, all rights reserved.
This open access ETD is published by University of Cincinnati and OhioLINK.