Abstract: Accurately assessing the structural reliability of water landing belly structure of wing-in-ground aircraft with low failure probability, which is influenced by multiple sources of uncertainty, typically requires tremendous finite element analysis. In this paper, based on the study of deep ensemble-active learning structural reliability analysis method, deep neural networks are employed as the surrogate model, and the local predictive variances are obtained by deep ensemble method. Combined with active learning strategy, adaptive sampling is performed in regions with high predictive variances and close to the failure boundary, thus reducing the number of required samples. In order to ensure the fitting accuracy of the failure boundary, weighted mean square error loss function is constructed based on the limit state function, and the neural network model with locally accurate surrogate capability is obtained. The structural reliability analysis of wing-in-ground aircraft belly water landing is carried out, and the results show that under the premise of accurately evaluating structural reliability, the number of structural response solutions is effectively reduced.