基于建设项目动态联盟候选伙伴绩效的内涵分析,确定了决定伙伴绩效的因素,构建了伙伴绩效评价的指标体系。通过主成分分析将众多指标进行综合,消除样本间的信息重叠,降低RBF网络的输入雏数。针对伙伴绩效系统的非线性特征,采用RBF网络高度非线性映射能力,对桌建设项目动态联盟的候选伙伴绩效进行了评价。评价结果表明PCA与RBF网络相结合的方法比单纯的RBF网络方法具有较高的精确度和较好的拟合效果。
Based on the analysis of the meaning of candidate partners' performance of dynamic alliance of construction project, we have ascertained the determinants of partners' performance, and established the evaluation criteria system for partners' performance. Through principal component analysis, we have synthesized numerous criteria, eliminated information overlapping of the ,sample, and reduced the input dimension of RBF network. According to the nonlinear feature of partners' performance system, we have used RBF network altitudinal nonlinear map to evaluate candidate partners' performance for a dynamic alliance of construction project. The results show that the conjoint method — PCA and RBF network is more precise and fits better than the single RBF network method.