Department of Computer Engineering
CS 590 SEMINAR
Machine Remaining Useful Life Prediction with Sequential Networks
Computer Engineering Department
Practically the industrial components are required to be maintained or replaced before the end of life in performing their assigned task. Predicting the life condition of a component is crucial for industries that have intent to grow in a fast paced technological environment. With the aid of well-designed prediction system for understanding current situation of a component that could be taken out of active service before malfunction occurs. Markov Chain models, optimization algorithms, several machine learning and deep learning approaches are the usual techniques in order to model maintenance tasks. In this study, we investigate Long Short Term Memory and Gated Recurrent Unit networks with high performance on open source data that is about an engine degradation simulation and we aim to predict remaining useful life as a regression problem. Interestingly, we implement our solution on different domain which is about Trauma Care Outcomes with the aim of predicting remaining life of patients. Therefore, we propose reproducibility of our solution in different domains. We observe Aalen Additive Fitter on our predictive model to have survival function which enables us to understand life behaviours of components. Differently, we provide an insight on RIMARC algorithm by maximizing area under curve and increasing model success.
DATE: 26 March, 2018, Monday, CS590 & CS690 presentations begin at @ 15:40