Have a UROP opening you would like to submit?
Please fill out the form.
Machine learning applied to optical fiber production control systems
2: Mechanical Engineering
Theresa Werth: email@example.com
The goal of the project is to develop a data driven model of the fiber draw manufacturing process and to simulate new fiber draw control strategies. You will be supplied with production data from multiple draw towers and specification(s) of the drawing process. You will develop a full simulation of the fiber draw process taking into account and modeling the control system as currently implemented. You won’t develop models from complete first principles; feedback loop structures will serve as constraints on the data analysis learning algorithms. You will learn an aggregate model of the full system. You will develop a realistic virtual model of the full controllers and system as implemented – to use as a simulation tool. Then, in simulation, you will modify the controllers in order to predict how a new control setting or system will perform. The simulation will serve as a design and decision tool to guide experiments and deployment on real draw towers. Impact: This project is of interest to fiber draw processing and other continuous production systems.
Experience and/or interest in Machine learning and data analytics, regression and autoregression, data visualization, process modelling, hybrid modelling.