Ph.D. students Su Jii Kim and Yunhan Kim of SNU Laboratory for System Health & Risk Management (Director: Prof. Byeng Dong Youn) received Best Poster Award and Best Presentation Award respectively at PHM Korea 2020, held at the K Hotel Seoul from July 21st to 23rd. 

Mr. Su Jii Kim’s research is titled “Extraction of Acoustic Emission Sensor Characteristics for Fault Diagnosis of Low-speed Rolling Bearings.” It studies the standard for figuring out the characteristics of the acoustic emission sensors for diagnosing faults of low-speed rolling bearings. The research also proposes a new standard for extracting the characteristics based rankings and scores of relationships between variables. Current standards consider the distribution of variance and distance, or use the concentration of data within distribution and high-dimensional cluster separation. These methods end up overfitting the given data, unable to perform diagnostics new data to a satisfactory level. This research suggests a new method that avoids the problem of overfitting in sensor signal data and produces more accurate diagnostics in regards to new data.

Mr. Yunhan Kim’s research is titled “Deep Learning-Based Fault Detection of High Voltage Circuit Breakers Using Coil Current Analysis.” It proposes a deep-learning algorithm that autonomously detects abnormal waveforms of coil current that causes the circuit to shut down when the circuit breaker is faulty. There is a lack of data circuit breakers since they are rarely used to begin with. The research resolves this problem in two ways, using (a) data augmentation and (b) unsupervised learning using deep autoencoders. It also verifies the algorithm with data collected from actual circuit breakers in electrical substation facilities.