Cascading failures of interdependent multilayer system for urban critical infrastructures

With the continuous improvement of urban critical infrastructure systems, the interdependent effect between different urban infrastructure systems becomes more obvious, that is, when a critical infrastructure system fails, the system coupled with it will also be affected by it. Therefore, how to measure the interdependent relationship between systems, determine their influence scope and influence path, and reduce the failure probability of critical infrastructure systems are particularly important for improving urban stability. In this paper, the interdependent mechanism between multi-layer critical infrastructure systems is analyzed and combined with the topology diagram, the interdependent topology mechanism model of multi-layer critical infrastructure systems is proposed. On this basis, combined with graph neural network and expert experience, a dynamic multi-layer critical infrastructure interdependent system prediction model is proposed. This model introduces time sequence information to update the relevant formulas of edge weights and nodes in the static network, so that the operating state of critical infrastructure can change with the change of time and space, so as to build a two-layer dynamic coupling network of critical infrastructure systems, which makes the model more close to the actual physical failure process and provides a basis for the subsequent prediction of the failure probability of critical infrastructure.