It commonly exists lag, overshoot, large volatility and other characteristics in water bath temperature control, and it is difficult to establish a precise mathematical model. Generally, water bath temperature is controlled by fuzzy controller, but its rules have strong subjectivity, and its control performance often can not reach the objective requirements. In the paper, the genetic algorithm is used to optimize fuzzy control rules and combined with cultural algorithm to form a kind of cultural genetic algorithm with double cooperative evolutionary mechanism. The expert experience knowledge is used to construct belief space, which acts as guidance direction, and then conducts genetic algorithm among the group space. The individual with higher fitness value is selected to update belief space. Finally, through iterative optimization, the individual with high fitness value is obtained and acted as the fuzzy control rules. Simulation results show that the optimized rules have a better performance in dynamic temperature tracking of control process and steady-state error.