Automation of skill fuzzy control system is an important research aspect of fuzzy control fields. It's significant for those control instances consisted in production and people's daily life. But, how to control a system not movement or behavior rules but only relied on movement parameters, that problem still had not be resolved. This paper proposes a new method used a genetic algorithm based on immune mechanism to learn the degree of membership, at same time, simplifying the corresponding movement equation; its efficiency will be indicated by an example.