Abstract: Aiming at the defects which are generated when the traditional interactive genetic algorithm deals with the contradictions between maintaining population diversity and accelerating the convergence, an improved niche genetic algorithm is proposed for tacit objective optimization problems, with the purpose of avoiding the local converge, enhancing algorithm efficiency and reducing user fatigue.Combining niche ideas based on sharing m echanism, the concept of niche entropy and adaptive niche radius is introduced to ensure population diversity.The simulation experiment in automobile modeling sketch design shows the validity and effectiveness of the proposed algorithm.

Key words: sharing mechanism, Niche Entropy, population diversity, user fatigue, automobile modeling design based on concept sketches