The integrity of dam safety monitoring data is crucial for ensuring the safe operation of dams.However，data loss is inevitable during long-term monitoring，severely compromising the effectiveness of data analysis and early warning capabilities.Existing methods for handling missing data such as omission，linear interpolation，spatial proximity-based imputation，and mathematical modeling still exhibit limitations when dealing with complex types of missing values.To address this issue，this paper proposes a missing data imputation method based on K-Means clustering and LSTM networks.This method overcomes the limitations of traditional statistical modeling factors by automatically identifying groups of monitoring points with similar deformation characteristics through K -Means clustering.The time-series data of highly correlated points within the same cluster are used as input variables for the LSTM model，constructing a spatiotemporal fusion-based prediction model for missing values.This approach fully leverages the spatial correlations among monitoring points and the long-term dependency features of time-series data to impute missing monitoring data.Engineering applications demonstrate that this method outperforms traditional interpolation techniques in terms of imputation accuracy and applicability，providing a novel technical solution for quality control of dam safety monitoring data.