Abstract: For the real problem of a Chengdu factory's scheduling for semiconductor assembling and testing line, a math model for the single station is built on minimizing the maximum operation time of the equipment in which both the switching time and initial state are considered. Adaptive parallel genetic algorithm is proposed to solve this problem, combined with the genetic algorithm, which has high solution efficiency and good convergence. Finally, an example based on real data is given in order to illustrate the proposed method, and the results show its efficiency.