A Fish Biomass Prediction Model for Aquaponics System Using Machine Learning Algorithms

Most of the fish biomass estimation process till now, is based on human intervention, making the process time-consuming, expensive, imprecise, and laborious. Also, no studies have been done on fish biomass estimation for aquaponics systems. This paper proposes two prediction models for fish weight estimation of the aquaponics system using the Artificial neural network and its hybrid with fuzzy logic i.e., Adaptive neuro-fuzzy inference system (ANFIS). The Feed-forward backpropagation network is used in the ANN model, and for the fuzzification process in ANFIS model, the Gaussian membership function is employed with Sugeno structure as the fuzzy inference system (FIS). The performance evaluation indicated that ANFIS model had attained the best prediction accuracy in terms of MAE of 0.3141 and RMSE of 0.6379, and R-2 of 0.9808 in comparison with the conventional ANN model. This prediction model can help predict the fish growth of the aquaponics system simply and cost-effectively, supporting farmers in avoiding imbalances in market supply, and demand and economic management.