本文研究了最小化拖期任务数并行机调度问题，提出了一种基于知识的遗传算法，并在问题的描述、遗传算法编码、初始种群的产生办法、交叉方法、变异方法等方面作了研究．不同规模问题的数值计算结果表明了本文提出的基于知识的遗传算法优于目前为止最好的启发式算法和普通的遗传算法，并且能适用于较大规模的并行机调度问题 In this paper, the problem of minimizing the number of tasks in parallel machines with tardiness tasks is studied. A genetic algorithm based on knowledge is proposed, and the problem description, genetic algorithm coding, initial population generation method, crossover method and mutation method are studied . The numerical results of different scale problems show that the knowledge-based genetic algorithm proposed in this paper is superior to the best heuristic algorithms and common genetic algorithms so far, and can be applied to large-scale parallel machine scheduling problem