Breaking News

Fault diagnosis: Innovation in machine learning for manufacturing – Penn State News

UNIVERSITY PARK, Pa. — As a doctoral candidate in the Penn State College of Engineering’s Harold and Inge Marcus Department of Industrial and Manufacturing Engineering (IME), Toyosi Ademujimi’s work with manufacturing and machine learning has been a lifelong fascination. 

Always amazed by how things are made, Ademujimi credits his upbringing in Nigeria with his keen interest in manufacturing. Nigeria has a small and declining manufacturing sector, and Ademujimi hopes his education will help him rebuild manufacturing in his home country. 

“This curiosity led me to Penn State because it is one of the best industrial and manufacturing engineering programs in the world, and I aspire to become a world-class expert in manufacturing,” Ademujimi said. 

Ademujimi has spent time working with the Service Enterprise Engineering Initiative (SEE 360), led by Vittal Prabhu, professor of industrial engineering and Charles and Enid Schneider Faculty Chair in Service Enterprise Engineering. SEE 360 helps engineers tackle issues within the service sector of the economy. 

Specifically, Ademujimi studies the implications of machine learning technologies on machine failures. Machine failures pose a significant threat to a factory’s productiveness and competitiveness. 

“Ademujimi is willing and eager to research issues that cut across disciplines by exploring ways to leverage models to accelerate machine learning while combining data for the improvement of asset management,” Prabhu said. “His work has the potential to open up new classes of data-driven services and analytics.” 

Avoiding breakdown 

The demand for production is higher than ever, according to Ademujimi, placing significant pressure on companies to manufacture products quickly, at low costs and with high quality. He noted that, within manufacturing industries, the maintenance department has the responsibility of keeping production machines running at all times. 

“To improve the fault diagnostics of machines, machine learning techniques can be applied to leverage recent technological developments and the vast amount of historical data available in manufacturing industries to create smarter fault diagnostics,” Ademujimi said. 

By utilizing machine learning, workers can identify machine issues even faster and reduce the potential downtimes. Ademujimi explained that during machine downtimes, most of the time spent is on determining the issue rather than fixing it. Once the root cause is found, though, fixing the machine usually comes easily.  

“If I’m sick, I might have a headache or develop a fever, but this does not necessarily tell you what is wrong with me,” Ademujimi said. “The symptom is what we see but linking that symptom to the root cause is the most difficult part of any diagnosis.”   

With current improvements in sensor technology, data storage and internet speeds, factories are becoming smarter and generating more process data. Researchers are focusing on how to utilize this “big data” to improve manufacturing competitiveness. 

Machine learning technologies use preexisting data to teach a machine learning algorithm how to perform a specific task. In the case of system failures, an engineer uses this technology to analyze possible causes of failure. When the machine learning algorithm is tasked with a root cause analysis, a system for identifying the fault when it occurs, it will theoretically have enough data to perform the analysis successfully.   

“It’s very challenging and complicated,” Ademujimi said. “It’s an area of research where if we can improve fault diagnosis, it will help manufacturing systems, improve production and reduce downtime. This will then funnel back into the economy in a positive way.” 

For Ademujimi, conducting research that goes beyond the lab is essential. Although his current research mainly applies to the manufacturing industry, he said he can see how it could impact all areas of the service sector. 

“Service systems, from banks to grocery stores to restaurants, as well as manufacturing, make up a large area of the United States’ economy,” Ademujimi said. “Improving these systems improves our lives.” 

Almost done 

Ademujimi anticipates graduating in fall 2020 and joining industry to apply his new skills and knowledge to improve manufacturing processes continuously. 

Manufacturing or not, Ademujimi said he foresees the need for the number of service engineers to increase as consumers continue to seek quality products quicker, and sustainability becomes more prevalent. Big manufacturers and local businesses alike are looking for ways to stay competitive in the fast pace climate of today’s society. 

In addition to his work with SEE 360, Ademujimi has worked with Volvo Group as a mechanical and systems engineering intern for data analytics. Before starting his doctoral program, Ademujimi worked full time in Nigeria as a manufacturing engineer for Nigerian Aluminum Extrusions and as a materials and quality engineer for Dantata and Sawoe Construction Company. 

Ademujimi received his bachelor of engineering in metallurgical and materials engineering from the Federal University of Technology Akure and his master’s degree in industrial engineering from North Carolina State University. 

Student spotlight series
The student spotlight series by the Penn State Harold and Inge Marcus Department of Industrial and Manufacturing Engineering (IME) aims to highlight innovators, makers and those that personify engineering excellence in their academic studies. The department currently has 90 doctoral students, 59 master’s students and 436 undergraduate students. In addition, the department hosts 42 full-time and courtesy faculty members. Established in 1908, the department is home to the first industrial engineering program in the world and has made a name for itself in the engineering industry through its storied tradition of unparalleled excellence and innovation in research, education and outreach. To learn more about IME and how you can get involved, visit ime.psu.edu. 

Source: https://news.psu.edu/story/614250/2020/04/06/academics/fault-diagnosis-innovation-machine-learning-manufacturing