In order to optimize bio-oil / diesel mixing ultrasound parameters for the emulsion fuel emulsification, an emulsion fuel artificial neural network prediction model was established using the neural network training sample data obtained from experiments. The best fuel ultrasonic emulsification parameters (ultrasonic frequency, power duration of action and incentive wave form) were identified using the genetic algorithm that simulated a natural evolutionary process using randomized adaptive search method. A test was conducted determine the consistence of the parameters between the experimental and calculated numerical data. The results showed that use of the mixed emulsion fuel prediction model and the genetic algorithm optimization model can accurately design ultrasonic emulsification parameters with a good mix stability for fuel preparation, which could provide a new preferred mode for preparing the mixed emulsion fuel.