Abstract: In order to improve the accuracy of metro network performance analysis, topological structures of Chengdu metro network were studied based on complex network theory, an evaluation index system of node importance was constructed, 136 nodes were evaluated by using grey correlation and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) method to identify key nodes. Four parameters such as network efficiency were used as network performance indicators. Trends in network performance were observed after 30 key nodes were attacked maliciously and 106 common nodes randomly. The results show that the average degree of the Chengdu metro network is 2.147, the average path length is 13.146 5 under the Space L model, the network betweenness, connectivity and efficiency are low, and the East Passenger Station (56) ranks first in 30 key nodes, and that both malicious and random attacks will make the network efficiency and the natural connectivity decrease slowly, but will make both the network connectivity and the maximal connected subgraph decrease faster.

Key words: urban rail transit, complex network, topology structure, key node, network performance