Informer-LSTM-FECAM Model for Coastal Dynamic Data Estimation

This paper proposes a deep learning-based approach for time series prediction to address the limitations of traditional models in handling such data. The article primarily introduces the applications of Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory networks (LSTM), and Transformer-based models, with a particular focus on the recently proposed Informer model. By analyzing the strengths and weaknesses of these models in time series prediction tasks, an improved model-a hybrid Informer-LSTM-FECAM model based on Discrete Cosine Transform—is proposed. This model enhances the accuracy and efficiency of long sequence prediction in Informer by incorporating a Frequency Enhanced Channel Attention Mechanism and LSTM layers. Through experiments on multiple real-world datasets, the paper demonstrates that the improved model outperforms traditional models in prediction accuracy, particularly excelling in long time series tasks such as coastal dynamic data.