Abstract: In order to construct a high-precision dam deformation prediction model, a feature screening method based on maximal information coefficient（MIC-CFS） was introduced to screen the dam deformation influencing factors, which reduced redundant information and the complexity of the model. At the same time, the gray wolf optimizer（GWO） was used to optimize the regularization coefficient and kernel parameters of the kernel extreme learning machine（KELM）, to improve the prediction accuracy of the model, and a dam deformation prediction model based on MIC-CFS-GWO-KELM was established. The measured data of a concrete double-curved arch dam were used to test the model, and the results show that the proposed model is better than GWO-KELM, MIC-CFS-KELM, PCA-GWO-KLEM and MIC-CFS-BP models in root mean square error, mean absolute percentage error and R-squared, and the prediction accuracy is high, which can provide a reference for the safety analysis of dam deformation.