Extreme disasters may cause large-scale energy supply problems in the electricity-gas integrated energy system

resulting in power outages and gas outages.Existing fault scenario screening methods mostly focus on single fault scenarios and lack modeling of the cascading fault propagation mechanism in the electricity-gas coupled system

resulting in insufficient robustness prediction of the evaluation results for extreme disasters.This paper constructs a cascading failure model for the electricity-gas integrated energy system

considers the fault propagation mechanism of power transmission lines and natural gas pipelines

constructs a dynamic propagation model of cascading failures based on Markov decision process

and uses the proximal policy optimization algorithm to efficiently search and optimize disaster scenarios.Finally

a cascading failure model for the electricity-gas integrated energy system is built based on the IEEE 30-node power grid and 20-node gas grid

and the PPO algorithm training process is tested.The example shows that this method can identify serious fault scenarios and fault propagation paths and ensure good convergence

thereby improving the emergency response capability of the electricity-gas integrated energy system when it is impacted.