APPROACHES ON SOCIAL ATTRIBUTE NETWORKS BASED ASSESSMENT MODEL AND ITS APPLICATION ON SUBSTATION SYSTEM DESIGN

作者：Xu, Gang; Tan, Yuanpeng^*; Zhao, Kunxiang

来源：Journal of the Balkan Tribological Association, 2016, 22(2A): 1843-1862.

• attribute measure
• social attribute network
• attribute spatial projection
• attribute recognition model

In this paper, a novel Social Attribute Networks (SANs) theory is proposed to give a further discussion on the qualitative description of measurement problems in complex systems and relationship mining among all sub-attributes and their associated elements. In this theory, the attribute space is divided into many attribute subspaces by introducing attribute spatial projection, index correlation factor, and spatial correlation factor. By employing these two factors, attribute subspace correlation matrix is established. Based on the attribute subspace correlation matrix, assessments of targeted object in all attribute subspaces are made and integrated into the final synthetical assessment scores. Moreover, according to the proposed SANs theory, a novel Social Attribute Preference Learning (SAPL) based assessment algorithm is proposed. In SAPL based method, according to the historical training samples, the attribute preference weights of indexes are calculated by employing supervised random weighted networks learning. Then, the assessments of test samples are performed by using SANs based synthetical assessment model and well-trained attribute,preference weights. The experimental results demonstrated that the proposed SANs theory and SAPL based assessment method are reasonable, and the proposed SAPL based assessment method holds better performances than other traditional ones.

• 出版日期2016
• 单位华北电力大学（保定）; 华北电力大学