An Integrated Fishery Meteorological Information Service Terminal Based on End-Side Deep Learning Technology

Authors

  • Xiaolong Zhao College of Information and Electrical Engineering and National Innovation Center for Digital Fishery, China Agricultural University, China
  • Xueqian Fu College of Information and Electrical Engineering and National Innovation Center for Digital Fishery, China Agricultural University, China https://orcid.org/0000-0001-7983-8700
  • Xiangrong Zeng College of Information and Electrical Engineering, China Agricultural University, China
  • Ningyi Zhang College of Information and Electrical Engineering, China Agricultural University, China

DOI:

https://doi.org/10.47852/bonviewAIA42021821

Keywords:

fishery energy internet, meteorology sensitivity, fishery informatization, fishery load, deep learning

Abstract

Fishery meteorology has multiple impacts on the fisheries industry, especially in modern fishery industrial parks where renewable energy is extensively utilized. Therefore, this study developed a comprehensive fishery meteorological information terminal, based on the Android system, that considers the requirements of fish farming, fishery load, and the characteristics of renewable energy for fishery meteorology. This terminal aims to provide convenient and comprehensive information services to aquaculturists actively involved in modernizing the fisheries industry. The system consists of two main subsystems: the fishery subsystem and the weather subsystem. In the fishery subsystem, real-time monitoring and recording of fishery meteorology and related parameters can be achieved. In the weather subsystem, the demand for photovoltaic energy in weather forecasting is emphasized. A weather prediction model based on LSTM is used for hourly weather forecasting. The model is trained on meteorological station data by default, and users can also upload photovoltaic station data to obtain a model trained on such data. The system can retain two models simultaneously, and when one of the datasets is unavailable, the available data is used to make predictions on the corresponding model to ensure service stability. Additionally, we conducted experiments to verify the performance loss brought by deploying the model on the edge using TensorFlow Lite. The results show that when the memory usage is reduced to 1/33 of the original, the model still retains over 99% of its performance. 

 

Received: 1 October 2023 | Revised: 19 November 2023 | Accepted: 25 January 2024

 

Conflicts of Interest

Xueqian Fu is the Lead Guest Editor for Special Issue on Utilizing Deep Learning and Statistical Theory in the Context of Smart Grids, a special issue of Artificial Intelligence and Applications. The authors declare that they have no conflicts of interest to this work.

Metrics

Metrics Loading ...

Downloads

Published

2024-02-02

How to Cite

Zhao, X., Fu, X., Zeng, X. ., & Zhang, N. . (2024). An Integrated Fishery Meteorological Information Service Terminal Based on End-Side Deep Learning Technology. Artificial Intelligence and Applications. https://doi.org/10.47852/bonviewAIA42021821

Issue

Section

Online First Articles