Adaptive Particle Swarm Optimization based System Identification and Internal Model Sliding Mode controller for Mass Flow System

Abstract

An integrated mass flow meter and control valve forms the Mass Flow Controller (MFC) and is extensively used in semiconductor industries. In this work an Internal Model Sliding Mode Controller based on Adaptive Particle Swarm Optimization (IMSMC-APSO) is proposed to control the valve position in MFC. First, step is to identify an open loop transfer function model for MFC through system identification via the data acquired from a real time mass flow controller. Optimal transfer function parameters are estimated through APSO and compared with PSO for better estimation accuracy. In second step IMSMC is designed for the identified transfer function model and an APSO optimally tunes the control parameters off-line. The control objective is to track a multi step flow trajectory instantaneously with high precision, under model uncertainties and pressure disturbances accounting the limitation on long sensor time constant. For the purpose of comparison a classical PI, a simple IMSMC are also designed. Simulation results show that IMSMC- APSO automated the controller tuning and improved the MFC performance in terms of settling time (average)  by 11.45% and accuracy (ITAE) by 16.33%, 20.24% and 18.26% in normal condition, amid pressure disturbance and model uncertainty respectively  compared to simple IMSMC.