Adaptive Neuro-Fuzzy Controller for Real-Time Melt Pressure Control in Polymer Extrusion Processes

Melt pressure is a key indicator of melt flow stability and quality in polymer extrusion processes. The melt pressure level affects the degree of mixing and melt pressure stability which in turn influence the melt quality. Meanwhile, short-term melt pressure fluctuations result in dimensional variations in the extruded products in continuous extrusion processes. In industrial polymer extruders, the melt pressure is measured using pressure transducers installed close to the die entry. Process operators ensure the safe operation of the process based on these melt pressure measurements. This study proposes an intelligent control system based on the adaptive neuro-fuzzy inference system to manipulate the screw speed to maintain the melt pressure at a desired level while minimizing melt pressure fluctuations. Data recorded from an actual extrusion process was used to determine the parameters of the membership functions and the rule base of a Sugeno fuzzy inference system using neuro-fuzzy learning. The controller was designed and validated through simulation using Matlab Simulink. The results indicated that the proposed controller was capable of maintaining the desired melt pressure level while minimizing melt pressure fluctuations across different extrusion processing conditions, by manipulating the screw speed. Therefore, this will be an attractive solution to improve the dimensional stability and product quality of extruded products in continuous polymer extrusion processes.