ABSTRACT
The multi-objective particle swarm optimization algorithm is deployed to tune PID controllers considering desired criteria, which are overshoot, robustness to variation in the gain of the plant, settling time, and rise time. The fixed membership function design of a type 1-based fuzzy logic controller leads to the difficulty of rule-based control design when representing the linguistic nature of knowledge. To improve the performance of the controller through adjusting control rules and membership functions, one has to exploit some effective optimization vehicle, such as particle swarm optimization. As a general purpose optimization tool, genetic algorithms should be applicable to any type of neural network for which an evaluation function can be derived. The existence of genetic algorithms for training could aid in the development of other types of neural networks. Genetic algorithms and their cousins, like particle swarm optimization algorithims, are capable of searching the entire solution space with more likelihood of finding the global optimum than conventional optimization methods.
