This paper integrates Knowledge Graph, Scale-free Network, and Bayesian Network for fault diagnosis of the power grid. The proposed model combines Knowledge Graph based on typical accident reports with the Scale-free Network model to establish and simplify the causal network for better representing accident scenarios in the power grid. Subsequently, it constructs a Bayesian Network for quantitative analysis of accidents, from causes to consequences. Then, sensitivity analysis is conducted to provide valuable support for decision-making in power grid safety management. Based on the results of sensitivity analysis, the critical causation path of “Circuit Breaker Trip” is “Animal Intrusion”→“Improper Safety Distance”→“Two-phase Short Circuit”→“Switch Trip”→“Circuit Breaker Trip”.