(268f) Multi-Scale Structure Resolution of AE Signals and Its Application to Early Agglomeration Detection in HSBR

This investigation was performed to study the underlying structure characteristics of acoustic emission (AE) signals, which could be helpful not only to understand a relatively complete picture of hydrodynamics in multiphase flow systems, but also to extract the most useful information from the original signals with respect to a particular process measurement requirement. However, due to AE signals are made up of emission from many acoustic sources at different scales, the resolution of AE signals is often very complicated and appears to be relatively poorly researched. In this study, the structure characteristics of AE signals measured in gas-solid fluidized bed, liquid-solid stirred tank and horizontal stirred bed reactor (HSBR) were researched and compared by resorting to wavelet transform and rescaled range analysis. A general criterion was used to resolve AE signals into three physical-related characteristic scales, i.e. micro-, meso- and macro-scale. Multi-scale resolution of AE signals implied that AE signals in micro-scale represented totally the dynamics of solid phase and could be applied to measure particle-related properties. Furthermore, based on the structure characteristics of AE signals, the most useful features related to particles motion were extracted and the possibilistic fuzzy c-means (PFCM) clustering method was adopted to detect novel agglomeration mode in HSBR. The satisfactory results illustrated the effectiveness of proposed method, which provides an effective on-line ?early and accurate? agglomeration detection technique in HSBR.