摘要
针对已有伙伴选择问题方法的不足, 以最小化竞标花费, 交通运费和拖期惩罚成本之和为目标, 给出伙伴问题的 0—1整数规划模型, 并提出了求解问题的混合遗传算法。该算法用染色体的编码确定合作伙伴的组合, 结合专家经验, 通过模糊神经网络对人工给定的隶属函数和模糊规则进行修正, 用所得到的结果改进合作伙伴的组合。实验结果表明, 模糊神经网络的使用提高了模糊推理的准确性, 进而也提高了混合算法的有效性, 使其具有比普通遗传算法更好的寻优能力。
Partner selection problem is an important decision problem in agile manufacturing. We present a hybrid genetic algorithm which is based on the integer programming model of minimizing the total bid cost, tardiness penalty and transport costs. The basic idea of the hybrid computing approach is to use the chromosome code which is produced in genetic algorithm can ensure the combination of partner, then integrate the experts' experience, by using the self-learning function of fuzzy neural network achieve the intention of optimizing the fuzzy membership function and fuzzy rules, using the results to improve the combination. The result of experiment show that using the fuzzy neural network can improve the veracity of fuzzy reasoning and the validity of the hybrid computing approach, compared with the common genetic algorithm, this approach has a much better capability of searching solutions.
出处
《吉林大学学报（信息科学版）》
CAS
2005年第2期184-189,共6页
Journal of Jilin University（Information Science Edition）
基金
国家 863计划基金资助项目 (2001AA135120 1)
关键词
动态联盟
伙伴选择
模糊规则
模糊神经网络
dynamic alliance
partner selection
fuzzy rules
fuzzy neural network
