Researchers at Purdue University have developed a system for efficient, low-cost monitoring for machine health, including overall quality, condition, and operation status.
The Purdue team’s innovation uses audio-based artificial intelligence technology to monitor the overall conditions of machines in factories, hospitals, and other locations. The Purdue system uses a stethoscope-like system as a sensor and analyzes the data with a neural network-based framework.
“Our solution is to use the concept of doctors listening to a body to assess the initial condition or experts listening to the machine sounds to know what is going on,” said Martin Jun, a Purdue innovator and associate professor of mechanical engineering. “We are using artificial intelligence to train a wide range of sounds from the machine and determine many things about the machine or process autonomously.”
Jun said this system can detect anomalies without being fed a training set and is easier and more cost-effective than accelerometers or acoustic emission sensors.
The Purdue technology is designed to use internal sounds from a machine to determine the machine status, assess process conditions, diagnose machine conditions, and predict machine failures.
“Since only sound is used, it can be used for a number of different applications,” Jun said. “Having one low-cost sensor for many different purposes can address the current challenges in the area where most of the solutions are quite customized to specific problems.”
Maijker Corp., a startup created to commercialize technologies involving sensors and artificial intelligence.
Jun and the team have worked with the Purdue Research Foundation Office of Technology Commercialization to patent the technology. They are looking to license it and are seeking collaborators for further development. For more information, contact D.H.R. Sarma of OTC at [email protected] and reference track code 2020-JUN-68816.