In order to analyze the complex fault process of power grid, a Bayesian network is constructed based on the action information of protective devices and circuit breakers, and a new method of complex fault deduction of power grid is proposed. The Bayesian network is constructed by using the actual topological structure information of the power grid, the action information of relay protection devices and circuit breakers, and the fault components are diagnosed based on the Bayesian network. In view of the two special cases of remote backup protection and near backup protection misjudgment, the rules of expert system are formulated to realize the recognition of protection rejection and misoperation. The analysis results of information and malfunction rejection of the system are integrated according to the sequence of main protection layer-near backup protection-far backup protection-circuit breaker layer, and the sequence of action of each layer of protection device and circuit breaker is deduced to realize the deduction of power grid fault. This work is supported by National Key Research and Development Program of China (No. 2016YFB0101900) and Science and Technology Project of Guangdong Power Grid Company (No. GDKJXM20162205) “Technical Research and Advanced Functions Application of Power Network Fault Fast Diagnosis Based on Fault Record Information”.