Dam Displacement Prediction Method Combining Image Sequence Input with Residual Networks

In this paper, a new dam displacement prediction method is proposed to solve the problems of traditional recurrent neural networks that require a large amount of data for training to avoid overfitting, poor generalization ability, and difficulty in capturing distant local information in the sequence. In this method, the time series is transformed into image series under fixed coordinates, and the residual network is used to predict, and the predicted images are restored to the specific values under the original coordinates. Compared with traditional deep learning models such as LSTM, GRU and RNN, the proposed method has significant advantages in prediction accuracy. The experimental results show that the proposed method has significant advantages in the face of limited training data, noise interference, and dealing with distant dependencies. The proposed method provides a new and effective solution for dam displacement prediction.