Abstract: The construction of safety monitoring models using dam deformation monitoring data is a crucial method for the quantitative analysis of deformation patterns. However, existing deformation monitoring models often exhibit significant shortcomings in the selection of influencing factors and resistance to outlier interference. To address these issues, this study proposes an adaptive identification method for deformation outliers in concrete dams based on optimized statistical models. This method not only identifies outliers during the regression modeling, but also prevents distortion of the monitoring model due to erroneous data cleansing. Initially, Bayesian model selection techniques are employed to reduce redundancy among the influencing factors affecting concrete dam deformation, facilitating the identification of significant explanatory variables during the statistical modeling phase. Subsequently, robust regression analysis using least trimmed squares estimation is applied to the deformation monitoring data, establishing a monitoring model capable of adaptively identifying various types of anomalies in concrete dam deformations. Finally, we design and implement a visualization strategy for different types of anomalies within the data series, providing an intuitive representation of their locations and potential impacts. Case study demonstrates that the proposed method effectively identifies key influencing factors in concrete dam deformation, adaptively mitigating the interference of different anomaly types on regression analysis. This enhancement leads to improved significance of regression results, as well as increased goodness of fit and predictive accuracy. The approach exhibits strong applicability in anomaly detection within monitoring data and quantitative analysis of dam safety status.