This paper presents a novel method of pattern identification in complex systems using the tools derived from statistical thermodynamics. Complexity issues arise in natural or human-engineered systems due to behavioral uncertainties and nonlinearities involved in the process dynamics. The paper introduces a novel concept of behavioral pattern identification and anomaly detection in mechanical systems from macroscopically observed time series of the available sensor data. The theme is built upon the principles of Statistical Thermodynamics and Information Theory. The efficacy of this method is experimentally validated on a laboratory apparatus where the behavioral changes accrue from the evolving fatigue damage in polycrystalline alloy structures.
- Manufacturing Engineering Division
Pattern Identification in Complex Systems: A Statistical Thermodynamic Approach
- Views Icon Views
- Share Icon Share
- Search Site
Gupta, S, Srivastav, A, & Ray, A. "Pattern Identification in Complex Systems: A Statistical Thermodynamic Approach." Proceedings of the ASME 2006 International Manufacturing Science and Engineering Conference. Manufacturing Science and Engineering, Parts A and B. Ypsilanti, Michigan, USA. October 8–11, 2006. pp. 771-778. ASME. https://doi.org/10.1115/MSEC2006-21090
Download citation file: