麻豆传媒

Probabilistic Machine Learning Inference to Model Unshaped Failure Rate Function

Postdoctoral Researcher
Marine and Arctic Technology, Aalto University

Date:聽November 30, 2022
Time: 1:00-2:00 PM
Venue: "I" Building, Room 310, Sexton Campus
Online MS Team:聽

Abstract:
Lacking, limited, or ambiguous knowledge of the remaining useful operation lifetime lead to unnecessary early maintenance actions. This implies that there are opportunities for value optimization of the process, regardless of how the system is artificially intelligent and the operation's autonomy level. The justification for decisions on maintenance optimization is more solid when the system under study is a critical part of the asset, and when its failure can easily cause interruption of the process. To avoid such implications, the uncertainty associated with the state of the system health should be incorporated in the analysis and decision making. In maritme shipping, most present maintenance planning resulting in early replacement is however based on a constant failure rate, not accounting for uncertainties.

In this presentation, a probabilistic machine learning (ML) application using Gaussian process (GP) and Bayesian inference (BI) is given to present an unshaped failure rate to establish the maintenance planning. The recent progress of the Maritime Autonomous Surface Ship (MASS) concerning the maintenance planning of machinery plants is briefly outlined. The talk also touches upon two ML approaches (GP and BI). While ML can reduce the time expense of the estimations, GP and BI can guarantee the precision of the final solution. As a case study application, a critical component of an engine, the spark, is considered, to illustrate the proposed method. Ignition voltage is a process variable by which the reliability of the spark is measured.

Speaker Biography:
Since September 2020, Dr. Ahmad BahooToroody has been working as a postdoctoral researcher in the Marine and Arctic Technology Group of Aalto University. With a Ph.D. in Reliability Engineering from the University of Florence (Italy), he was invited to liaise with researchers at TU Delft (The Netherlands), the Norwegian University of Science and Technology (Norway) and the University of Strathclyde (United Kingdom). His PhD research focused on risk-based asset integrity modeling of the automotive process, and he completed both the comprehensive exam and oral defense with distinction. Ahmad鈥檚 research interests include uncertainty quantification, resilience and trustworthiness of engineering operations, condition-based maintenance optimization, and application of probabilistic methods and machine learning tools in dynamic modeling. He has been keen to pass on his knowledge and experiences through supervising undergraduate and graduate students, and advising Ph.D. candidates.

Industrial Engineering seminar series contact person:
Dr. Floris Goerlandt
email: floris.goerlandt@dal.ca

General Enquiry:
Ms. Tara Parker
Tel: 902.494.3281
email: tara.parker@dal.ca