In order to improve the fault tolerance and solving efficiency of existing analytical models for power grid fault diagnosis, this paper proposes an optimization model for power grid fault diagnosis that takes into account the reliability of protection and circuit breaker action state alarm information. First, the reliability of alarm information is evaluated based on the protection and circuit breaker self-test information. Secondly, a multi-objective optimization model is constructed to minimize the difference between the warning information and the expected action state, and maximize the expected action state credibility of protection and circuit breaker. Thirdly, the optimization model is linearized and the optimal solution is obtained based on the weighted sum method and the fuzzy membership function. Finally, the examples verify the efficiency and versatility of the model. The model in this paper can quickly diagnose power grid faults when there are errors in the alarm information, and has good application prospects. This work is supported by the Natural Science Foundation of Guangdong Province (No. 2017A030313304).