针对带有交货期窗口硬约束，并对提前/拖期零件进行惩罚的一类作业车间调度问题，设计了一种改进型遗传算法EGA。交货期窗口硬约束是指每个零件均有交货期和deadline，零件可以在交货期后完工，但绝不能超出deadline。EGA采用“逆向后推”和“顺向前拉”相结合的两阶段求解策略。针对部分染色体在解码过程中违反交货期窗口硬约束而产生非法解的问题，采用基于关键路径的染色体修复方法来调整染色体基因序列，以期实现在满足交货期窗口硬约束的同时降低零件拖期成本；在保持第一调度阶段拖期成本不变的基础上，采用基于逆向重调度的目标值修订方法来延迟零件开工时间，以减小成品提前库存成本。-This paper deals with the job shop scheduling problem with due window to minimize the sum of weighted earliness and weighted tardiness of jobs. This due window is a time duration between the due date and the latest acceptable completion date, namely dealine, of a job. Penalties are incurred if jobs are completed within the due window and the illegal schedule will occur if one job is finished after its deadline. A new Enhanced Genetic Algorithm (EGA), based on a two phase schedule mechanism was introduced. To avoid the violations of due window constraint and simultaneously reduce tardiness costs, a chromosome repair method based on critical paths was designed to adjust chromogene sequence. Then, on keeping tardiness costs unchanged, an objective value revision method based on reverse rescheduling was put forward to delay job release date for reducing earliness costs introduced by each job.