To achieve the accurate diagnosis of fault components of power grid, a hierarchical fuzzy petri net (HFPN) evaluation model considering the dynamic weight is proposed based on the idea of comprehensive variable weight. First, a subnet containing suspicious fault line’s main protection, near backup protection and far backup protection is established at both ends, and then a comprehensive diagnostic subnet is established for the three types of protection, forming a double-layer model for line fault diagnosis. In view of the disadvantage of traditional fixed weighting in the case of warning information loss that may cause fault line misjudgment, considering the dynamic effect of the component"s state level on the decision, the model is given the dynamic weight to amplify the truth of the action elements. The fault diagnosis method combining fixed weight and dynamic weight can effectively identify the fault lines in the case of information loss. Through the multi-group fault simulation of typical power grid, the results show that the petri net diagnosis method based on comprehensive variable weight can improve the accuracy of fault diagnosis.