How to generate and adjust the membership function and fuzzy rules are difficult problems in the design of fuzzy controller. To solve this problem, adaptive neural fuzzy inference system (ANFIS) is used to design the fuzzy control system, then the fuzzy rules and membership function can be obtained by back propagation or hybrid algorithm of the neural network. In order to verify the validity of the method, the distillation column temperature control system and CFBB temperature control system are simulated respectively. The results of simulation show that the membership functions and the fuzzy rules of the fuzzy controller can be obtained by the training of the PIn control's input/output data, and it can convert PID control experiences into fuzzy control rules efficiently.