Abstract: Aiming at the multi-order Job Shop scheduling problem under mixed production patterns, a new genetic algorithm was proposed. A double-coding method was proposed by new algorithm which could decouple assembly constraints and record order information weight effectively to guide subsequent genetic operation. First gene rule was used to improve the population diversity at population initialization step. A multi-parent crossover operator based on orders was designed in the algorithm's cross time, which not only ensured that children inherit their parents more excellent generation information, but also just appeared feasible solution.The mutation operators based on order weight could prevent the non-mature convergence for the algorithm, and ensure the higher weight order complete on time. Through numerical simulation, the effectiveness of proposed algorithm on solving more order Job Shop scheduling problem under mixed production patterns was verified.

Key words: assembly, processing, multi-order scheduling, genetic algorithms

摘要： 针对混合生产形态下（既有加工也有装配）的多订单调度问题，提出一种新的遗传算法。该算法首先提出一种双层编码方法，可以有效解耦装配约束及记录订单权重信息，以指导后继遗传操作；新算法在种群初始化采用"首基因"规则以提高种群多样性，在交叉操作时设计基于订单的多父辈交叉算子,不仅能够保证子辈染色体更多地继承父辈的优秀信息，还不会出现不可行解；基于订单权重的变异算子可以在防止算法非成熟收敛的同时，尽量保证权重高的订单按时完成。通过数据仿真结果证明，该算法可有效求解混合生产形态下的多订单调度问题。

关键词: 装配, 加工, 多订单调度