The Department
of Systems Engineering and Engineering Management (SEEM) of City University of
Hong Kong (CityU) organized a Seminar entitled “Physical Model-based
Engineering Products and Systems Design with Assured Reliability and
Resilience: Methodologies and Application”. This seminar aimed to introduced
model-based reliability analysis, significant design errors and proactive
maintenance such as Prognostics and Health Management (PHM). I was honor invited by Prof. Min XIE to
attend Dr. Zhimin XI seminar and we took a photo for memory.
In the
beginning, Prof. Min XIE gave welcome speech and introduced our guest speaker.
Dr. Zhimin XI
(奚志敏) (Assistant Professor, Industrial and Systems Engineering, School of
Engineering, Rutgers, The State University of New Jersey) introduced his talk
content. Firstly, he explained why we need reliability and resilience because
of many failures/errors happened such as in Aerospace System, Nuclear System,
Automotive System and Structure System, etc.
Then Dr. XI
introduced his research objectives that the first one is to build theoretical
foundation to design highly reliable and resilient engineering systems.
The second
objective is to create tangible industry impacts to design highly reliable and
resilient engineering systems.
He said design error happened because of the model error. Then he introduced the Uncertainties in Model Validation (δ). For FE analysis, we needed to considered test uncertainty, model uncertainty and physical uncertainty. Then the framework for model-based prediction and design was discussed and he said effective control design errors resulting from model errors in the design of engineering products and systems.
And then Dr.
XI used some study case to discuss how to determine and improve model accuracy
in the intended uses of the model. Those
challenges questions showed below:
-
How
to calibrate model parameters?
-
How
to characterize model bias?
-
How
to quantify model accuracy?
-
How
to improve model accuracy?
The following
diagram is one of examples to analysis model bias using response surface
approach. 
One of
interested example was Lithium-ion Battery’s model uncertainty in Dr. XI’s
study in 2014. He observed that battery
State of Change (SOC) estimation using a typical extended Kalman Filter (KF) in
fact is not very accurate where the error could range from 5% to 10% or even
more depending on the battery characteristics.
After that Dr.
XI overviewed of PHM including Health Sensing Function, Health Reasoning
Function, Health Prognostics Function and Health Management Function. For Health Prognostics Function, it separated
to Model-based and Data-Driven Approaches. He said we always lacked of data and
proposed to use Copula-based sampling approach (Copula modeling). But
inaccurate copula modeling could occur if insufficient training units.
Lastly, he
explained aircraft engine prognostics to explain reliability and resilience for
design and maintenance. Some case studies were briefed such as Battery Capacity
Prognosis, Resilience-Based Design of a Microgrid System, Simulation of
Layer-by-Layer Selective Laser Melting Process, and Uncertainties in Autonomous
Vehicles.
After the
seminar, we had dinner in CityU together.
(Left: Dr. Zhimin
XI, Prof. Min XIE, Dr. Zijun ZHANG and I)
Reference:
沒有留言:
發佈留言