Inverted pendulum is a typical object for intelligent control. Aiming at the blindness in choosing the weight matrices Q and R during the LQR control process, these matrices are optimized by genetic algorithm in this paper,. The fuzzy control rules are optimized by using information fusion function to change the single complex control policy to hierarchical ones based on quotient space, which can greatly decrease the dimension of the fuzzy control rules. The simulation results show that this method is feasible and the system performs favorable control effect and dynamic properties.