Ice coating on overhead transmission lines poses a significant threat to the safe and stable operation of power grids. Accurate prediction of ice thickness is crucial for disaster prevention and mitigation. However, traditional prediction methods often rely on single models and empirical parameters, making it difficult to adapt to complex and variable meteorological conditions and line environments, resulting in limited prediction accuracy. To address these issues, this paper proposes an accurate prediction method for ice coating on overhead transmission lines based on hyperparameter optimization and ensemble models. Firstly, a multi-dimensional feature space incorporating meteorological factors, line parameters, and historical ice data is constructed, and key features are selected using feature engineering methods. Secondly, considering the nonlinear nature of ice data, the SSA algorithm is employed to iteratively optimize the hyperparameters of VMD, enhancing its decomposition performance. Subsequently, three single prediction models are applied to predict the decomposed components, and the results are superimposed and reconstructed to obtain the prediction results of the single models. Finally, the entropy weight method is used to assign weights to the prediction results of different single models, and the final prediction value is calculated based on the weight results. Simulation results demonstrate that, compared to traditional single models, the proposed method can effectively improve the prediction accuracy of ice thickness, providing a scientific basis for ice disaster warning and prevention in power grids.