INTELLIGENT FLOW RATE REGULATION USING HYBRID METAHEURISTIC PID OPTIMIZATION AND ANFIS CONTROL

Abstract

This thesis discusses some of the advanced control techniques applied in flow rate regulation in industry, including optimization based tuning and intelligence based modeling. There are two major parts to the thesis. In the first one, the Proportional Integral-Derivative (PID) controller is optimally tuned using Hybrid Ant Colony Optimization (ACO)-Particle Swarm Optimization (PSO). The combination of the strengths of the two optimization algorithms (global search ability of PSO and local search abilities of ACO) enables optimal tuning of the PID controller. Consequently, there will be improved dynamic performance with low settling time, reduced overshoot and stability. In the other part, an Adaptive Neuro-Fuzzy Inference System (ANFIS) model is used as an intelligent data-driven model for flow rate control. Training of the ANFIS model is done using system's input-output data. ANFIS can adaptively control flow rates and has good performance characteristics under different operating conditions and external disturbances. Comparison of performances of the optimized PID controller and ANFIS controller are made using various performance measures like rise time, settling time, overshoot and IAE. Results from simulation show that even though tuning improves the performance of the PID controller, ANFIS-based approach provides better robustness and disturbance rejection performance.