Dear All,
We are pleased to invite you to the IEOR Online Seminar by Maharshi Dhada (faculty candidate).
Talk title: Multi-level Models for Asset Failure Prognosis and Maintenance
Planning
Speaker: Maharshi Dhada, University of Cambridge
Time: Monday, 16th February, 11:35AM
Venue: Online on MSTeams
Link: https://teams.microsoft.com/meet/44665641044357?p=kXdld22Q0xGm7JVUie
Meeting ID: 446 656 410 443 57
Passcode: gS6vK7jn
Abstract: In this research seminar Maharshi will present his work on
multi-level statistical models for prognosis of asset fleets and using them
for the purpose of maintenance planning. The challenge being addressed here is
for the cases where a fleet-wide model inferred by pooling the data from all
comprising assets is associated with a high bias due to diversity in operating
conditions, whereas the independent assets-specific are associated with high
variance due to sparse data. Such cases of data sparsity are commonly
encountered when the assets operate in varying conditions or are newly
commissioned. It prevents the access to sufficient training data and
information for analytics at the asset-specific level. In such cases, the
assets with sparse data can benefit by borrowing information from other
similar assets. Multi-level models enable this borrowing of information by
systematically extending the asset-specific models such that the parameters of
the assets-specific models are sampled from their corresponding higher-level
distributions that are shared across similarly behaving asset clusters. This
multi-task learning approach enables structured inference of model parameters,
quantify the risk for maintenance planning, and enable incorporating expert
knowledge for asset similarities to control the extent of learning. Maharshi
will also briefly discuss his other activities in logistics in this seminar.
Bio: Maharshi is the SF Express Research Fellow in Logistics based at
Darwin College, University of Cambridge. He works at the Cambridge University
Engineering Department for research and teaching in Logistics. Maharshi holds
a BTech in Mechanical Engineering from the Indian Institute of Technology
(IIT) Indore and PhD in Engineering from the University of Cambridge. He has
been working on engineering asset management for over eight years, and in
logistics for over four years. He has published 17 international conference
and 9 peer-reviewed journal articles. Maharshi collaborates with many leading
industrial organisations such as Siemens, Boeing, Scania CV, BT, Network Rail,
Glovo, Port of Felixstowe and SF Express working on problems spanning across
maintenance planning, asset failure prognosis, and logistics.
Host: Ashutosh Mahajan