FUZZY CLUSTERING METHODS AS TECHNIQUES FOR IDENTIFYING CONTROL STRATEGIES FOR AN UNDERWATER VEHICLE

摘要

In this paper an algorithmic methodology is proposed to construct fuzzy control strategies based on the choice of different fuzzy clustering algorithms to define the number and actual position of the prototypes and the variance of the clusters. Then the generation of the fuzzy sets is obtained by using projecting methods and their number is reduced by applied merging methods. The proposed methodical steps are applied to identify an underwater robot's control strategy to avoid objects. The objective, therefore, is to identify the fuzzy rules that the operator uses to construct the control surface. The applied inputs are the distance and the angle between the vehicle and the objects. The resulting control surface is very close to the original one and thus the algorithm works very satisfactorily.