Title: Residual Life Prediction and Degradation-Based Control of
Multi-Component Systems

*Advisors*: Dr. Nagi Gebraeel and Dr. Jianjun Shi

*Committee members*: Dr. Kamran Paynabar, Dr. Chuck Zhang, and Dr. Jian Liu
(University of Arizona)

*Date and time*:  Thursday, March 05 2015, 10:30AM

*Location*:  Academic Office - Groseclose 204

*Abstract*:
The condition monitoring of multi-component systems utilizes multiple
sensors to capture the functional condition of the systems, and allows the
sensor information to be used to reason about the health information of the
systems or components. This thesis focuses on modeling the relationship
between multi-sensor information and component-level degradation, so as to
prediction both system-level and component-level lifetimes. In addition,
this thesis also investigates the dynamic control of component-level
degradation so as to control the failure times of individual components
based on real-time degradation monitoring.

The research topic that Chapter 3 focuses on is identifying component
degradation signals from mixed sensor signals in order to predict
component-level residual lives. Specifically, we are interested in modeling
the degradation of systems that consist of two or more identical components
operating under similar conditions. The key challenge here is that a defect
in any of the components will excite the same defective frequency, which
prevents an effective separation of the degradation signals of defective
and non-defective components. To the best of our knowledge, no existing
methodologies have investigated this research topic. In Chapter 3, we
propose a two-stage vibration-based prognostic methodology for modeling the
degradation processes of components with identical defective frequencies.
The first stage incorporates the independent component analysis (ICA) to
identify component vibration signals and reverse their original amplitude.
The second stage consists of an adaptive prognostics method to predict
component residual lives. In the simulated case study, we investigate the
performance of the signal separation stage and that of the final
residual-life prediction under different conditions. The simulation results
show reasonable robustness of the methodology.

In Chapter 4, we focus on characterizing the interactive relationship
between product quality degradation and tool wear in multistage
manufacturing processes (MMPs), in which machine tools are considered as
components and the product quality measurements are considered as condition
monitoring information. Due to the sequential structure of MMPs, the
degradation status of a tool affects the product quality current stage,
which, on the other hand, may affect the degradation of tools at subsequent
stages. To the best of our knowledge, although existing literature has
modeled the impact of product quality on the tooling catastrophic failure,
no published work has targeted on the impact of product quality on the
actual process of tool wear. To address this research topic, we propose an
high-dimensional stochastic differential equation model to capture the
interaction relationship between the process of tool wear and product
quality. We then leverage real-time quality measurements to on-line predict
the residual life of the MMP as a system. In the simulation study, we
conclude that our methodology consistently performs better than a benchmark
methodology that does not consider the impact of product quality on the
process of tool wear or utilize real-time quality measurements.

Chapter 5 explores a new research direction, which is the dynamic control
of component-level degradation in the parallel multi-component system, in
which each component operates simultaneously to achieve an engineering
objective. This parallel configuration is usually designed with some level
of redundancy, which means when a small portion of components fails to
operate, the remaining components can still achieve the engineering
objective by increasing their workloads up to the designed capacities.
Consequently, if the component degradation can be controlled, we can
achieve better utilization of the redundancy to ensure consistent system
performance. To do this, Chapter 5 assumes that the degradation rate of a
component is directly related to its workload and develops a strategy of
dynamic workload adjustment in order to on-line control the degradation
processes of individual components, and thus to control their failure
times. The criterion of selecting the optimal workloads is to prevent the
overlap of component failures. We conduct a simulated case study to
evaluate the performance of our proposed methodology under different
conditions.