There are some critical nodes in power system operation, and the failure of these nodes can easily aggravate the propagation of faults and lead to serious power outage accidents. In order to identify critical nodes in power systems and improve system security, this paper proposes a critical node identification algorithm based on PageRank algorithm. A model of the impact of node failure on system line power is proposed based on tide calculation, and this model is combined with the PageRank algorithm to construct point-to-point links to achieve effective identification of critical nodes. Finally, simulation analysis is performed on IEEE-39 system, and the topology and capacity and load are verified and compared with existing methods by simulating deliberate attacks on critical nodes, which proves that the proposed method is useful for identifying nodes in critical positions in the system and can provide guidance for safe and stable operation of the power system.