Sequence prediction of missing data of spatio-temporal series of marine environmental elements based on ST-ResNet model

Yuanfeng Zhong

doi:10.54097/nzlf82ty

Authors

Yuanfeng Zhong

DOI:

https://doi.org/10.54097/nzlf82ty

Keywords:

Marine environmental elements, Deep learning, ST-ResNet model

Abstract

Spatial data of marine environmental elements are of great significance to the marine industry and marine safety, and they are also affected by external factors. This results in a new method for spatial and temporal sequence prediction of marine environmental elements based on the ST-ResNet model. Specifically, the temporal distance, period, and trend of marine regions are modelled by residual networks, and different residual sub-networks are used for each characteristic, in each sub-network, different weights are assigned to different branches and regions by simulating the changes of marine data, and the output is finally fitted to obtain the output. The ST-ResNet model was validated on the ERA5 dataset as an example, and compared with other prediction models, the ST-ResNet model achieved better results, which verified the effectiveness of the method in predicting the temporal and spatial sequences of marine environmental elements.

References

Dimoulkas I, Mazidi P, Herre L; EEM 2017 Prediction Competition: Using Autoregressive Models for Wind Power Generation Prediction. The 14th International Conference on European Energy Markets, 2017, (1-6).

Wang Jieruo, Yu Jing, Sun Huiying, et al. Spatial and temporal distribution characteristics of coordinated development between China's marine economy and resource environment [J/OL]. Notice on Marine Lakes and Marshes, 1-9 [2024-03-27].

Goodfellow I, Pouget Abadie J, Mirza M, etc. Generate adversarial networks [J]. ACM Communication, 2020, 63 (11): 139-144.

Kingma D P, Welling M. Automatic Encoding Variational Bayes [C]//International Conference on Learning Representation. Isaka, NYarXivorg, 2013.

Lu Xiaomin, Liu Fan, Cai Lihua, et al Integrated stack autoencoder and XGBoost for deep learning wave effective wave height prediction model J Computer and Modernization, 2021 (04): 32.

Srivastava N, Mansimov E, Salakhudinov R. Unsupervised Learning Using LSTM Video Representation: International Conference on Machine Learning [Z]. two thousand and fifteen

Zhu Q, Chen, Zhu, etc. Temporal and Spatial Wind Speed Prediction: A Deep Learning Method [J]. Energy, 2018, 11 (4): 705.

Bengio Y, Courville A, Vincent P. Representative Learning: A New Perspective and Review [J]. IEEE Journal of Pattern Analysis and Machine Intelligence, 2013, 35 (8): 1798-1828.

Zhang, Zheng Yong, Qi D. Deep spatiotemporal residual network for predicting urban crowd flow. Proceedings of the 31st AAAI Artificial Intelligence Conference. 2017, (1655-1661).

Shi, Chen, Wang. Convolutional LSTM network: a machine learning method for precipitation forecasting. Proceedings of the 28th International Conference on Neural Information Processing Systems, Volume 1. 2015, 11 (802-810).

Wang, Long, Wang, et al. Recurrent neural network predictive learning based on spatiotemporal LSTM [J]. Progress in Neural Information Processing Systems, 2017, 30.

Wang, Zhang, Zhu, et al. Memory in Memory: A Predictive Neural Network for Learning High Order Non stationarity from Spatiotemporal Dynamics [C]//IEEE/CFF Conference on Computer Vision and Pattern Recognition Proceedings. 2019: 9154-9162.