Big Data Analytics Framework for System Health Monitoring

Presenter: 
Sathis Kumar, Assistant Professor of Computer Sciences (CCU)
Event Location: 
BCCMWS 100
Event Date: 
Thursday, March 31, 2016 - 3:00pm
Event Type: 
SCMSS Seminar Series

Abstract

In this talk, Machine Learning (ML) based big data analytics framework that was developed and tested to improve the quality and performance of Auxiliary Power Units (APU) health monitoring services will be presented. In addition, the text mining classification framework that was implemented and evaluated using big datasets and the Cloud Enabled Brain Computer Interface (CEB) architecture for intuitive decision making to augment the current data processing techniques will be presented as well.

Key contributions in this work include the development and use of customized ML algorithms that have been tested and used to analyze multiple data sources and to provide useful insights and increase the ability to predict (1) APU wear from 39%to 56% and (2) APU shutdown events from 19% to 60%. In addition, the proposed framework and the algorithms create predictive models to predict the category labels for a given text input with high classification accuracy based on the experiments and by improving our algorithms and framework integrating with Natural Language Process (NLP) tools. Preliminary experimental results indicate that the classification and prediction accuracy of the predictive models created from the framework improved from 40% to 90%.

Such system health monitoring can be integrated with the widely used condition based maintenance (CBM) services. Users can use this cloud based analytic toolset and access the big data through any devices (PCs, Tablets, smart phones) anytime and anywhere. Future work will be focused on expanding the big data analytics framework towards data-driven predictive analytics for related applications in other domains.

Speaker Information

Dr. Sathis Kumar is an Assistant Professor at Coastal Carolina University in the Department of Computer Sciences. He received his Ph.D. in Computer Science from the University of Louisville in 2007.

More information about him and his work can be found here.