As regional integration advances, logistics hub networks have become increasingly complex, making their reliability and robustness critical to efficient system operations. However, in actual operations, logistics hubs are often disrupted due to unforeseen events such as natural disasters and technical failures, leading to significant economic losses. To address this issue, this paper proposes a hub network design problem based on two-stage robust optimization, considering the uncertain risk of hub disruptions. The aim is to minimize the setup cost of the hub network and the total transportation cost of commodities under the worst-case scenario. In this problem, the first stage involves comprehensive decisions on the location of logistics hubs and the opening of hub arcs before uncertainty is revealed. In the second stage, after the occurrence of disruption scenarios within the budget uncertainty set, the optimal transportation paths for commodities are adaptively determined. To enhance the computational efficiency of the model, this paper designs an improved Benders decomposition algorithm, where a two-step solution strategy is proposed to efficiently solve the subproblems and strengthen the generated Benders cuts. Extensive computational experiments are conducted on classical datasets. The results demonstrate that the proposed improved Benders decomposition algorithm significantly outperforms the classical approach in computational performance. Moreover, the proposed two-stage robust hub network design model is effective in mitigating hub disruption risks.