NRD: A node importance evaluation algorithm based on neighborhood reliance degree for power networks

Abstract

Identifying critical nodes is significant for reinforcing system stability and optimizing system performance. Due to the enormous scale and high real-time requirements of the power network, the algorithms based on global properties are not applicable. Meanwhile, the existing local algorithms are based on static connections, ignoring the hidden reliance and dynamic interactions between nodes in real-world networks. To address this problem, we propose an innovative concept of node reliance and introduce the asymmetric reliance matrix to describe the unequal relationship between nodes. Furthermore, the neighborhood reliance degree is modeled to evaluate the values of the above relationships accurately. More importantly, the proposed algorithm focuses on the dynamic interaction between the reliance matrix and reliance degree so as to realize the mining of node importance information. We compared the proposed algorithm with eight benchmark algorithms on nine power networks. The experimental results show that the proposed algorithm performs better in the dimensions of network efficiency, connectivity, vulnerability, ranking distribution, and complexity, indicating that it achieves more effective identification and accurate evaluation of important nodes.