Prediction of Selected Water Quality and Macronutrients Parameters in an Aquaponic System Using Artificial Neural Network (ANN)

Aquaponics, essentially incorporates multiple aquaculture and hydroponics into a single closed-loop system, is an excellent alternative method for generating reduced environmental waste by recycling nutrients (fish waste) for crop development. The pH, dissolved oxygen (DO), total ammoniacal nitrogen (TAN), and total percentage sludge of phosphorus (P) and nitrogen (N) of empirical experiments obtained from the aquaponic system have been investigated (N). To obtain the predicted value, an artificial neural network (ANN) modelling function employing the Levenberg–Marquardt (LM) and scaled conjugate gradient (SCG) training functions was utilized in this research. The outputs of the training function model suggested the optimal number of neurons for each pH parameter, DO, and TAN for neurons 6, respectively. In terms of total sludge (nitrogen and phosphorus), the optimal number of neurons is three. On the other hand, the most optimum neuron number at neurons 4 for parameters pH, DO, and TAN, while for the SCG training function was the most optimal neuron number is 4 for the total sludge parameter (N and P), respectively. Meanwhile, in trained neural network’s evaluated prediction parameters accuracies, LM outperformed SCG in terms of correlation coefficient (R) and Mean Square Error (MSE).