Modeling of the residence time distribution in a buss kneader with a back‐propagation neural network

Abstract

AbstractWith a back‐propagation neural network, the residence time distribution (RTD) characteristics in a buss kneader were modeled on a series of experimental RTD data measured by a digital image processing method. The operating conditions (screw speed and feed rate) were chosen as the inputs of the network. The four‐layered back‐propagation neural network predicted not only the RTD character indices, including the shortest delay time, mean residence time, and variance of distribution, but also the complete RTD curve. On the basis of the mean residence time, the average degree of fill in the extruder was also calculated. Furthermore, the effects of the operating conditions on the RTD and average degree of fill were analyzed. The method provided herein can also be used to predict RTDs in other kinds of extrusion equipment. © 2008 Wiley Periodicals, Inc. J Appl Polym Sci, 2008