Prediction of sea ice area based on the CEEMDAN-SO-BiLSTM model

• Published in: PeerJ
• Main Authors: Guo, Qiao, Zhang, Haoyu, Zhang, Yuhao, Jiang, Xuchu
• Format: Text
• Language: English
• Published: PeerJ Inc. 2023
• Subjects: Computational Science; Sea ice
• Online Access: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10362850/; http://www.ncbi.nlm.nih.gov/pubmed/37483978; https://doi.org/10.7717/peerj.15748

This article proposes a combined prediction model based on a bidirectional long short-term memory (BiLSTM) neural network optimized by the snake optimizer (SO) under complete ensemble empirical mode decomposition with adaptive noise. First, complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) was used to decompose the sea ice area time series data into a series of eigenmodes and perform noise reduction to enhance the stationarity and smoothness of the time series. Second, this article used a bidirectional long short-term memory neural network optimized by the snake optimizer to fully exploit the characteristics of each eigenmode of the time series to achieve the prediction of each. Finally, the predicted values of each mode are superimposed and reconstructed as the final prediction values. Our model achieves a good score of RMSE: 1.047, MAE: 0.815, and SMAPE: 3.938 on the test set.