Dissolved Gas Analysis(DGA) is a widely used technique for diagnosing faults in power transformers, specifically designed to identify potential issues within them. Despite recent advancements, the modeling of relationships between samples remains insufficient. This limitation results in less contextualized feature representations, ultimately impacting the accuracy of diagnosis. To this end, we propose a novel graph deep learning model called GCDGA (Graph Contrastive Learning for Dissolved Gas Analysis) to enhance Dissolved Gas Analysis. In our model, we construct two types of graphs: Similar KNN ( \(G_{S}\) ) and Dissimilar KNN ( \(G_{DS}\) ), where nodes represent individual records of dissolved gases, and edges denote the similarity and dissimilarity relationships between nodes. To better generate node representations, we design a dual-channel graph encoding component, with each channel playing a crucial role in capturing the corresponding edge type view. Furthermore, in order to better distinguish the feature differences among various fault categories, we employ ranking contrastive loss to refine the node representations. Experimental results demonstrate the high accuracy and robustness of our model in transformer fault diagnosis.