Considering the fuzzy boundary problem and the differences between individual transformers, a early warning rule strategy of power transformer health status based on big data mining is proposed in this paper. The fuzzy C-means method is used to identify the optimal characteristics of the transformer. The probability diagram is used to verify that the proposed method can reflect the personalized characteristics of the transformer to the maximum extent, and the full set of dissolution data under the selected characteristics conforme to the Weibull model. The relationship between the distribution characteristics of dissolved gas and the defect/failure rate is analyzed, and the corresponding alarm threshold is calculated. The operation status of transformer can be identified by associating gas concentration and gas increase rate with established warning. On this basis, early warning rules based on different thresholds are proposed and applied to the field operation of transformers. The experimental results show that the accuracy rate of the proposed method is as high as 98.21%, which proves that the proposed method has good status monitoring performance.