Early career researcher wins grants to push boundaries of data analysis
Machining an airplane part, 3D-printing an architectural model, giving drugs to a patient intravenously—each of those processes depends on precise operations over time to produce good results.
Two projects at SUNY Polytechnic Institute could lead to new, automated tools to monitor these kinds of precise activities and make adjustments as needed. Dr. Firas Khasawneh, assistant professor of mechanical engineering, received a $196,000 grant from the National Science Foundation (NSF) for one project in April 2016 and a second grant, for $93,111, in October 2016.
Khasawneh earned his PhD in 2010 and has taught at SUNY Poly since 2013. These significant awards to an early-career researcher mark a special success for the Utica campus as it grows beyond its foundation as an undergraduate teaching center, says Andrew Wolfe, interim dean of SUNY Poly’s College of Engineering.
“We started the engineering program with the idea that it would be the catalyst for getting grants and becoming a research university,” Wolfe says. “Dr. Khasawneh is leading the way.”
Khasawneh’s first grant supports a project called “Collaborative Research: A Unified Framework for the Investigation of Time Series Using Topological Data Analysis.” This work involves two kinds of activities—controlled experiments with a pendulum, and workshop studies with a metal lathe.
Using microphones to detect sound or sensors to detect vibrations, Khasawneh and his undergraduate researchers gather data on how the “signal” emerging from a process varies over time. For example, a sensor might measure how the vibrations in a lathe change as the tool cuts into a metal cylinder.
The researchers use a technique called topological data analysis (TDA) to describe the “shape” of the data coming from the process over time, and then to identify which shapes correspond to which behaviors. “For example, how do we see chaos using TDA?” Khasawneh says. “How do we see periodic, or almost periodic, behavior?”
Right now, it takes special expertise to analyze performance in a system such as a machining tool, Khasawneh says. “You have to know how to vary the parameters to get the desired behavior.” The use of TDA could change that. “We can take the signal and apply mathematical tools that will minimize the amount of expertise needed to study the systems from which the signal came.”
Knowledge gained in this study could have applications in areas beyond mechanical engineering. “One of them is controlling human postural sway,” Khasawneh says. TDA has been used, for example, to analyze a person’s breathing and detect wheezing.
The second grant is for Khasawneh’s part in a collaboration with researchers at the University at Albany and Michigan State University. In this project, Khasawneh will use TDA in conjunction with machine learning, training a computer algorithm to read and evaluate the signal emerging from a process over time.
As in the first project, researchers will capture data about vibration from a machining system—in this case, a milling tool—and use TDA to describe the shape that data makes. “While collecting signals, I will vary the parameters in the system to force it to go through different transitions,” Khasawneh says. “Maybe I will cut in the range where it’s cutting smoothly, and then I’ll push the tool harder into the workpiece, so I’m cutting deeper.” When the tool operates improperly, it starts oscillating, making signals known as “chatter.”
Using TDA, researchers can distinguish between patterns of data that indicate smooth operation and patterns that indicate problems. They will train an algorithm to make those distinctions and apply them in the future to new situations.
This work could lead to new, intelligent sensors to monitor machining systems, Khasawneh says. “If I have a sensor on the machine, based on these descriptors and the training sets, the machine can detect when chatter is happening or about to happen and adjust the cutting parameters to steer away from that undesirable behavior.”
Because the technique is flexible, one might also use it in a range of other operations, Khasawneh says. “The same pipeline can be applied to improve accuracy in additive manufacturing, for example, and in drug delivery systems.”
Within SUNY Poly’s young engineering program, Khasawneh can serve as an example and mentor to others, Wolfe says. “He can provide assistance to even more junior faculty in engineering who are looking at how to get funding.”
Khasawneh is a rising star not only because he has drawn support from the NSF, but because of his collaborative work, Wolfe says. “He’s becoming a known person in the research universe, which is a major accomplishment as well.”
comments powered by Disqus