SNU Research Team Led by Professor Youn Byeng-Dong Develops World’s First Deep Learning Based Power Plant Turbine Diagnosis Technique


▲ SNU Department of Mechanical and Aerospace Engineering Professor Youn Byeng-Dong

 
Korean research team develops a new technique based on deep learning technology that assesses the conditions of the turbine facilities in power plants.
 
SNU College of Engineering (Dean Cha Kook-Heon) announced on 15^th that the research team led by Youn Byeng-Dong, professor of the department of mechanical and aerospace engineering, has developed world’s first integrity evaluation technique that integrates deep learning technology for power plant turbines of various capacities.
 
Recently, due to the aging of power plants over the world, maintenance fees are consistently increasing. In order to reduce unexpected failures and excessive maintenance fees, failure prediction and integrity evaluation technology that analyzes big data drawn from power plants is required.
 
Current evaluation technique in use demands high expertise and long development time. Hence, not only large development cost is ordered, but it is also reported that the technique’s accuracy at field is strikingly low.
 
The new technology that Youn’s research team developed resolves these issues.
 
It is currently being provided to POSCO Hyeong-San Power Plant for turbine diagnosis through the lab’s venture enterprise ‘One Predict’. This application demonstrates that this newly-developed technology has potentials to be applied to variety of turbines at power plants (thermoelectric, gas turbine, combined cycle, etc.) around the world.
 
It is expected that the application of this new technology will be stretched to wind power generator, clean energy development, industrial robots, and any other areas that involve rotary systems.
 
This research is published in the field’s most prestigious journal ‘IEEE Transactions on Industrial Electronics (TIE)’ and is currently on pending for patent application.
 

[Reference]

▲ Mimetic Diagram of the Deep Learning Based Power Plant Turbine Diagnosis Technique