Published in Scientific Papers. Series E. Land Reclamation, Earth Observation & Surveying, Environmental Engineering, Vol. XIII
Written by Stefan-Mihai PETREA, Ira-Adeline SIMIONOV, Alina ANTACHE, Aurelia NICA, Madalina CALMUC, Valentina CALMUC, Dragos CRISTEA, Puiu Lucian GEORGESCU
The paper aims to use machine-learning-based algorithms in order to enable and empower the integration of soft sensors for improving the economic sustainability of integrated multi-trophic recirculating aquaculture systems (IMRAS) through efficient and accurate water quality monitoring of nitrate (NO3), the main key parameter for maintaining the sustainability of the IMRAS in various production scenarios. A 30-day trial was conducted in a sturgeon–tarragon IMRAS to develop a NO3 soft sensor, based on a series of predictors such as pH, temperature, NH4, NO2, NO3, conductivity (EC), P2O5, Ca and Mg, as well as to identify the prediction model peculiarities in various exploitation scenarios generated by the crops culture density. The results reveal the effectiveness of different learning algorithms as MLR and XGBoost (>80% accuracy) in developing solutions for supporting the water quality monitoring process in IMRASs, concluding that the intensity of production technologies must be considered as a determinant factor in upscaling the solutions to industrial level.
[Read full article] [Citation]
