TY - JOUR
T1 - Deep Learning-Based Dissolved Oxygen Forecasting for Intelligent Aquaculture Management
AU - Lukas Weber
AU - Sophie Janssen
AU - Jonas Richter
AU - Elena Fischer
JF - Journal of Intelligent Agriculture and Food Systems
Y1 - 2023
VL - 2
IS - 1
SP - 67
N2 - Precise tracking and forecasting of dissolved oxygen (DO) concentrations in aquaculture environments are essential for sustaining ideal water quality and safeguarding fish well-being. This research introduces an improved Long Short-Term Memory (LSTM)-based framework for DO forecasting that leverages historical records of DO concentrations, water temperature, and other environmental variables. In contrast to conventional techniques that rely on static assumptions, the proposed framework dynamically adapts to changing environmental conditions, delivering immediate, high-accuracy predictions. Experimental results show that the improved LSTM framework achieves a forecasting precision of 92.28%, surpassing current models, including IFP (81.97%), DOE (63.81%), and DOP (86.79%). The framework's exceptional precision and flexibility make it a reliable tool for aquaculture oversight, helping optimize DO levels and reduce the risk of fish deaths. By incorporating artificial intelligence into aquaculture surveillance, this method enhances system performance and long-term viability.
UR - https://airpublications.com/c561274146
ER -