COMPUTATIONAL INTELLIGENCE IN CONTROL APPLICATIONS

Abstract

Sensitivity and Robustness is the primary issue while designing the controller for non-linear systems. One of the performance objectives for controller design is to keep the error between the controlled output and the set-point as small as possible. The control of many non-linear, inherently unstable systems using conventional methods is both difficult to design and marginally satisfactory in implementation. The introduction of optimization techniques in control engineering that makes use of evolutionary computation and an implicit imprecision is successful in counteracting these limitations. The field of computational intelligence has incorporated to such systems with an objective to achieve higher optimality and satisfactory performance The main aim of this work is to design and implement biologically inspired optimization algorithms based control system. The performance of Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Bacterial Foraging (BF) Optimization has been improved using hybridization. The novel algorithms developed are hybrid BF-PSO or adaptive BF, Genetically Bacterial Swarm Optimization (GBS0). The algorithms are first tested on basic mathematical functions and then implemented on various control engineering problems: Inverted Pendulum system, Ball and Beam System and trajectory tracking in robot manipulators. The proposed algorithms also played a vital role in eliminating the curse of dimensionality to an acceptable value.