![formulae demonstrating machine learning](http://www.uwindsor.ca/dailynews/foodservices/humankinetics/sites/uwindsor.ca.dailynews/files/styles/full/public/900_machine_learning_2.jpg?itok=UfQ-3ivT)
A research partnership with deep connections to the University of Windsor aims to improve the aerodynamic performance of road vehicles.
Math and statistics professor Mohamed Belalia is teaming up with a nascent company founded by retired UWindsor professor Ron Barron on a two-year project that will first generate vast amounts of data related to the flow of air over a generic vehicle, then use artificial intelligence to analyze that data to come up with the best solutions for problems related to aerodynamics.
While designed to benefit the automotive industry, the goal of the project is to develop artificial intelligence tools that could also apply to other industrial sectors, including aerospace, petroleum and gas, water resource engineering, and the cooling of electronic components and systems.
“Such data-driven analyses have become an integral component in many application areas,” said Dr. Belalia. “For this project, we are bringing together a team of computational fluid dynamics engineers and applied statistics specialists.”
Computational fluid dynamics involves analyzing data to solve problems involving the flow of liquids and gases. The analysis will involve about 200 numerical simulations accounting for a wide variation of relevant parameters such as vehicle speed, wind speed and direction, and distance between vehicles.
It would take humans months to conduct the kind of analysis AI could do in mere seconds or minutes.
Dr. Barron founded his company, State-of-the-Art Engineering Simulations (SOTAES) Inc., to take advantage of such AI.
He likened the work to an example he once heard at a conference.
“Imagine you are the operator of a train going through a tunnel, and there is a fire on the train,” Barron began. “You need to get people to safety, but there’s a lot of parameters you need to consider before you can tell people what to do.”
Where on the train is the fire? How many cars are on the train? How many exits are there and where are they located? Where in the tunnel is the train? Are there fans in the tunnel? How fast is the train going?
“There are a lot of parameters that determine which decision you should make,” Barron said. “But you can’t wait hours to analyze all that information to tell people how to get to safety.”
AI can provide an answer in seconds.
“That’s sort of what we’re trying to do,” he says. “We need to generate these huge data sets, use machine learning algorithms to analyze them, and then predict the outcome of previously unknown scenarios based on new combinations of parameters.”
The project is funded through a $120,000 grant from Mitacs, a national not-for-profit organization that brings together Canadian academia, private industry, and government to provide research and training opportunities.
The grant will pay stipends to a Master’s student, a doctoral student, and two post-doctoral fellows involved in the project.
—Sarah Sacheli