earticle

영어

Preview Control is a field well suited for application to systems that have reference signals known a priori. The use of advance knowledge of reference signal can improve the tracking quality of the concerned control system. The classical solution to the Preview Control problem is obtained using the Algebraic Riccati Equation. The solution obtained is good but it is not optimal and has a scope of improvement, as the Preview Control problem has many parameters to be defined and optimized. The Evolutionary Algorithms, inspired by real-time natural systems, are a solution to this multi-dimensional problem.
This paper studies the performance of Preview Control optimized by three Evolutionary Algorithms (Genetic Algorithm, Particle Swarm Optimization and Marriage in Honey Bees Optimization) for two bench-mark control systems: Industrial Servo and Inverted Pendulum System in terms of processing time, convergence characteristics and the quality of solution obtained.
The results of this paper illustrate the benefits and weaknesses of the Evolutionary Algorithms for solving the Preview Control problem. The PSO algorithm proves to be the best for Preview Control based problems. It performs well in both small and large search spaces. The study reveals that the MBO algorithm requires more computation time while the performance of GA degrades with the increase in number of parameters to be tuned.