Research on the prediction of dissolved oxygen content in aquaculture water

doi:10.16178/j.issn.0528-9017.20240561

Abstract

Abstract:

Dissolved oxygen, as a key water quality factor in aquaculture, is closely related to the yield and quality of aquatic products, and accurate prediction of dissolved oxygen content and changes is of great significance for ensuring the safety of aquaculture. In this study, the dissolved oxygen data of the collected aquaculture water were preprocessed, and then combined with the long and short-term memory (LSTM) neural network algorithm, the prediction model of the dissolved oxygen content of the aquaculture water was constructed to predict the dissolved oxygen concentration data at different times in the future, and the prediction accuracy of the prediction model of the dissolved oxygen content of aquaculture water was verified through different prediction accuracy indicators, in order to provide reference for the subsequent research on the prediction of dissolved oxygen content in aquaculture water.

Key words: aquaculture water, dissolved oxygen content, long and short-term memory (LSTM), prediction model

CLC Number:

S959
TP183

XU Hui, DENG Haoran, WANG Zhongpei, ZHOU Lele, QIAN Rong. Research on the prediction of dissolved oxygen content in aquaculture water[J]. Journal of Zhejiang Agricultural Sciences, 2025, 66(1): 184-188.

Figures/Tables 6

Fig.1 Dissolved oxygen content data from June 1 to 30, 2023

Table 1 Prediction model of dissolved oxygen content in aquaculture water

模型	时间步长	预测时间/ min	决定系数 (R²)	均方根误差 (RMSE)	平均绝对值误差(MAE)
LSTM	5	30	0.954	0.346	0.228
		60	0.907	0.492	0.333
		90	0.840	0.648	0.441
		120	0.671	0.926	0.622
	10	30	0.952	0.355	0.246
		60	0.898	0.516	0.362
		90	0.839	0.647	0.448
		120	0.745	0.815	0.569
	15	30	0.952	0.353	0.221
		60	0.895	0.523	0.363
		90	0.833	0.660	0.447
		120	0.693	0.892	0.615

Fig.2 Line chart of the measured and predicted dissolved oxygen contents in the next 2 hours under different time step

Fig.3 Line chart of the measured and predicted dissolved oxygen contents in the next 90 minutes under different time step

Fig.4 Line chart of the measured and predicted dissolved oxygen contents in the next 1 hours under different time step

Fig.5 Line chart of the measured and predicted dissolved oxygen contents in the next 30 minutes under different time step

References 15

[1]	郑谦, 李其琪. 基于CiteSpace的中国智慧农业研究趋势可视化分析[J]. 安徽科技学院学报, 2023, 37(5): 109-116.
[2]	张梦瑶, 郑谦. 长三角城市群农业信息化发展水平测度[J]. 安徽科技学院学报, 2022, 36(3): 97-102.
[3]	冯国富, 卢胜涛, 陈明, 等. 基于自注意力机制和改进的K-BiLSTM的水产养殖水体溶解氧含量预测模型[J]. 江苏农业学报, 2024, 40(3): 490-499.
[4]	胡继杰, 朱练峰, 钟楚, 等. 溶解氧对稻田土壤氮素转化及水稻氮代谢影响研究进展[J]. 生态学杂志, 2017, 36(7): 2019-2028.
[5]	彭巧, 王长平, 王冬梅, 等. 养殖环境因子对克氏原螯虾影响[J]. 中国科技信息, 2023(17): 128-130.
[6]	SHAHID F, ZAMEER A, MUNEEB M. A novel genetic LSTM model for wind power forecast[J]. Energy, 2021, 223: 120069.
[7]	李莉, 李文军, 马德新, 等. 基于LSTM的温室番茄蒸腾量预测模型研究[J]. 农业机械学报, 2021, 52(10): 369-376.
[8]	徐一帆, 韩云春, 黄刚, 等. 基于LSTM的矿井突水风险预测研究[J]. 科学技术创新, 2024(10): 183-186.
[9]	高源, 黄犚. 基于深度学习的股票价格指数预测[J]. 软件工程, 2024, 27(5): 7-13.
[10]	姚亚. 数据预处理和直方图时间序列在水质预测中的应用[D]. 杭州: 浙江大学, 2013.
[11]	HATZIKOS E, HÄTÖNEN J, BASSILIADES N, et al. Applying adaptive prediction to sea-water quality measurements[J]. Expert Systems with Applications, 2009, 36(3): 6773-6779.
[12]	卢毅敏, 张红. 结合时间序列分解和神经网络的河流溶解氧预测[J]. 华侨大学学报(自然科学版), 2020, 41(5): 659-666.
[13]	朱南阳, 吴昊, 尹达恒, 等. 基于长短时记忆网络(LSTM)的蟹塘溶解氧估算优化方法[J]. 智慧农业, 2019, 1(3): 67-76.
[14]	李建文, 孟连子, 刘娜, 等. 基于EEMD-SVR的渔业养殖水质预测模型[J]. 天津理工大学学报, 2018, 34(5): 14-19, 24.
[15]	YU Y, SI X S, HU C H, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Computation, 2019, 31(7): 1235-1270.