Development of adaptive network-based fuzzy inference system (ANFIS) in hydroponic

Hydroponic is increasingly important in changing world conditions, with its contribution to the economy, employment, productivity time and many other areas. In this study, climate and stock food tank used in hydroponic greenhouses are controlled by the Adaptive Network-based fuzzy Inference System (ANFIS). The ANFIS model was created in MATLAB with the specified membership functions and datasets. The model created has been tested with with trimf, trampmf, gbellmf, gas2mf, dsigmf, psigmf functions for each membership function, and the hyperparameters that give the best results have been determined. In addition, the model of the ANFIS structure created in tensorflow was implemented. Training has been carried out on data with the Multi-variable Regression (MVR), K nearest neighbors (KNN), Support Vector Regression (SVR), Multi-layer Perceptron (MLP), Random Forest (RF) and XGBOOST. The most appropriate machine learning method is determined based on mean absolute error and R2 values. Using the Raspberry Pi card and various sensors, the hydroponic prototype was developed and the suitable model was tested in this prototype.