Aiming at the problem that key transmission sections of the power grid are frequently transferred and difficult to identify quickly and accurately in large-scale renewable energy grid connection scenarios, a key transmission section identification method based on improved normalized cut and security risk index is proposed. A weighted graph of network topology is constructed based on spectral graph theory, considering the influences of node voltage and branch power flow factors when assigning weights to branches. This aims to comprehensively reflect the electrical connectivity between nodes. The network is partitioned using an improved spectral clustering algorithm based on normalized cut, and for the transmission section formed between partitions, key transmission sections are selected using security risk index. The results from case studies on the IEEE 39-bus system and Guangdong Power Grid indicate that the proposed method can accurately and effectively identify key transmission sections within the system and assess their risk of exceeding limits. Moreover, it can identify some unrecognized key transmission sections in multi-scenario applications in the actual power grid.