中文摘要：

制造伙伴选择是组建动态联盟的一个关键问题。在对已有制造伙伴选择方法总结和分析的基础上，给出了制造伙伴选择综合评价参考体系，提出了一种基于神经网络的制造伙伴选择模型，采用基于连接结构优化的粒子群优化算法（SPSO）对神经网络进行训练，实验证明，该算法不仅使训练的收敛速度大大提高，且其训练的神经网络的性能也显著增强，为制造伙伴选择问题提供了有效的解决途径。

英文摘要：

One of the critical issues in the formation of Virtual corporations is the manufacture partner selection. Based on the survey of existing research and theoretical analyses of manufacture partner selection, h presents a multiple attributes evaluation system of partner selection for manufacturing enterprises, and proposes choosing the cooperative partner with the artificial neural network trained by a Structure-improving particle swarm optimization algorithm. Simulation experiments show the feasibility of the algorithm and the superiority of the system over other evaluation approaches.