This paper presents an advanced Delegated Proof of Stake (DPoS) consensus mechanism, enhanced by a Probabilistic Double Hierarchy Linguistic Term Set (PDHLTS). This improvement targets the delegate election process within blockchain systems, which often suffer from oversimplified voting options that compromise expression and selection accuracy. By integrating PDHLTS, our approach enables nodes to articulate a broader range of voting preferences and certainty levels. This richer representation of voter intentions addresses critical issues such as voter sentiment oversimplification and potential centralization risks, thus enhancing the democratic nature and governance efficiency of blockchain systems. The introduction of a double-layered voting structure allows nodes to express their support or opposition with associated confidence levels, leading to a more balanced and representative delegate selection. Our results indicate that this method not only increases the accuracy of delegate selection but also enhances the transparency and fairness of the DPoS voting process. The Voting Algorithm with Probabilistic Double Hierarchy Linguistic Information (VAPDHLI) further refines this process by calculating the score and deviation degree of each node, forming a comprehensive basis for delegate selection. The feasibility and effectiveness of the improved DPoS mechanism are demonstrated through model implementation and practical examples.