DETECTION OF POLYPROPYLENE AGGLOMERATION IN HSBR BASED ON THE ANALYSIS OF ACOUSTIC EMISSION SIGNAL BY AUTO-REGRESSION MODEL

A method of agglomeration detection in horizontal stirred bed reactors(HSBR) was obtained according to the qualitative relationship between acoustic emission(AE) signals and agglomeration,by the use of auto-regression(AR) model,in which the frequency of AE signals varied with different sizes of particles stricken on the wall.By this method the AR power spectrum could be calculated and its variance with that of normal state could be obtained to warn agglomeration early and accurately.It was found from the cold-flow experiments that the AR power spectrum became fluctuant after the agglomeration being added in HSBR,meanwhile the low frequency band energy ratio of AE signals and variance rose.Furthermore,this method was also utilized to detect the agglomeration in the industry plant successfully,showing that this method could predict the agglomeration environmental friendly with fairly good accuracy.