Multiple fault minimal hitting generation is the first step in the development of multiple fault diagnosis strategy, a discrete differential evolution algorithm (DDE) based on a hybrid strategy is proposed in order to generate the minimal hitting set firstly. Redefine the meaning of the individuals of population according to the characteristics of Multiple fault minimal hitting generation; Hybrid differential strategies and random changes in the scaling factor are introduced，thus the contradictory of global optimization and convergence rate is further balanced. By the addition of two selection mechanisms, timeliness of algorithm and diversity of populations are improved. Through a example discrete particle swarm algorithm is compared to verify the effectiveness of the algorithm.