PSYLOGIC

South Korea
South Korea
SEARCH FILTERS
Time filter
Source Type

Oh S.-D.,Kyung Hee University | Kim Y.-J.,Kyung Hee University | Lee T.-H.,PSYLOGIC
Transactions of the Korean Society of Mechanical Engineers, A | Year: 2013

In this study, the authors develop a methodology for a diagnostic system with a vibration parameter that is influenced by environmental factors. The data tends to have a varying average over time. Often, these features are found in statistical data retrieved from a production line. If we utilize existing statistical techniques for these features, we could derive an incorrect diagnostic conclusion based on the different average values. To overcome the limitations of previous methods, the authors apply a function analyzed through regression analysis to predict the mean value and corresponding upper and lower limits at each stage. This technique also provides corresponding statistical parameters in varying dynamic means. To validate the proposed methods, we retrieve data from the engine assembly line of H Motors and verify the results. © 2013 The Korean Society of Mechanical Engineers.


Oh S.-D.,Kyung Hee University | Kim Y.-J.,Kyung Hee University | Seo H.-Y.,Kyung Hee University | Lee T.-H.,Psylogic | Lee J.-W.,Psylogic
Transactions of the Korean Society of Mechanical Engineers, A | Year: 2011

We develop a diagnostic system to monitor failures in an engine-assembly line. Existing techniques such as sensory analysis, time domain analysis, frequency analysis, and statistical analysis have limitations in the diagnosis of engine-assembly failure when there are abnormal vibration waveforms (crashing and damping signals) during the assembly. We use a wavelet technique to deal with crashing and damping signals. We also implement a new technique for developing diagnostic rules from sensor data, and we demonstrate its validity.© 2011 The Korean Society of Mechanical Engineers.


Oh S.-D.,Kyung Hee University | Kim Y.-J.,Kyung Hee University | Lee T.-H.,Psylogic
Journal of Mechanical Science and Technology | Year: 2014

It is difficult to analyze the raw vibration signals of complex vibrating machines because these signals have complicated patterns. An appropriate preprocessing method has to be applied to enhance the signal resolution. In most cases, these preprocessed data are also difficult to inspect, however, because distributions of these data may have non-parametric and multi-modal distributions. If we apply the currently available methodologies to these data, we will encounter problems such as low accuracy, long delay times, and so on. To overcome these limitations, we developed the FPRIS (fast pattern recognition inspection system). FPRIS guarantees high diagnosis accuracy with fast running time, and the usefulness of FPRIS is demonstrated through the learning of sampled data. © 2014, The Korean Society of Mechanical Engineers and Springer-Verlag Berlin Heidelberg.

Loading PSYLOGIC collaborators
Loading PSYLOGIC collaborators