Abstract: Partner selection of virtual enterprise is a multi-objective optimization problem. Ideal spot algorithm is used to change the multi-objective problem into several single-objective optimization problems. Double population self-adaptive genetic algorithm is proposed for solving a single-objective optimal problem. This algorithm concerns two populations and self-adaptive probabilities of crossover and mutation,during the course of optimization; each individual of a population can select its probabilities of crossover and mutation automatically according to its fitness. Thus, each individual owns the ability of self-adaptation according to the variation of the environment. After every iteration, the two populations exchange the better chromosome. This can break the balance of inner-population in the local minimization and escape the local minimization. Examples show that this algorithm can solve effectively the multi-objective optimization problem of virtual enterprise’s partner selection.

Key words: Virtual enterprise; Partner selection; Genetic algorithm; Ideal spot algorithm; Multi-objective optimization

摘要： 针对虚拟企业伙伴选择这一多目标优化问题，采用理想点法将其转换为多个单目标问题，并应用双种群自适应遗传算法进行问题求解。该算法涉及两个种群和自适应交叉、变异概率。在遗传过程中，每个种群的个体都根据适应度自动选择其交叉和变异概率，使个体对环境变化具有自适应调节能力；在一代遗传完成后，种群间交换优秀个体携带的遗传信息，以增加种群的多样性，避免陷入局部极值。通过算例，证实了该算法能很好地解决虚拟企业伙伴选择这一多目标优化问题。

关键词: 虚拟企业；伙伴选择；遗传算法；理想点法；多目标优化