关键词: 物流竞争力, 社会网络分析, 深度学习, 深层自编码器, 动量

Abstract: This paper analyzes the situation of urban logistics competitiveness deeply, constructs a Deep Auto Encoder Momentum Update Algorithm based on Widrow Function（DAE-WMA）, combined with the relevant theory of deep learning and the standard model and standard algorithm of Deep Auto Encoder（DAE）. According to the data volume characteristics of urban logistics competitiveness, this paper selects three types of UCI datasets to simulate the Standard Algorithm based on Error Function（DAE-ESA）, the Standard Algorithm based on Cross Entropy Function（DAE-CSA）, as well as the recognition ability of DAE-WMA. The simulation results show that the performance of the latter is better than the first two. Secondly, according to the competitive strength and competitive potential of logistics, based on Analytic Hierarchy Process（AHP）, the paper constructs the index system of city logistics competitiveness with 7 dimensions and 19 indicators, carries on cluster analysis and empirical research on logistics competitiveness of 13 core cities in our northwest country, using DAE-WMA and Social Network Analysis（SNA）. Simulation results show that DEA-WMA is more effective than SNA, the classification results of the core node cities are more reasonable and more conducive to analyzing problems. The research results lay the foundation for determining the developing strategy on logistics of cities along the New Silk Road and promoting the future cooperation and development of the domestic logistics industry.

Key words: logistics competitiveness, social network analysis, deep learning, Deep Auto Encoder（DAE）, momentum