Date of Award
Summer 2017
Document Type
Thesis
Degree Name
Master of Science (MS)
First Advisor
Kaczmarek, Thomas
Second Advisor
Ahamed, Sheikh I.
Third Advisor
Brylow, Dennis
Abstract
Most manufacturing plants contain some amount of time series sensor data – streams of values and time stamps. This data, however, isn’t useful with most types of analytics or machine learning for the purpose of process optimization. This thesis presents a novel and innovative solution to the problem using a software stack leveraging the Predix Complex Event Processing Engine (Edge Analytics) to condition the data, combined with RFID for serialization. Each step in the formation of the solution is documented, from connecting equipment to analyzing and ingesting data produced by the edge analytic. This solution was developed and piloted at the GE Grid Solutions plant in Clearwater, FL.