The Development of Hydroponic Nutrient Solutions Control Using Fuzzy and BPNN for Celery Plant

Abstract

As the increasing number of human populations, most live in urban areas with limited farmlands. Hydroponic is one of the solutions to grow crops in urban areas. Electrical Conductivity (EC) and scale of acidity (pH) in the hydroponic nutrient solution are the important things to be controlled. Controlling EC and pH in hydroponic can increase the quantity and quality of the crop. This research suggested a new method to merge fuzzy and Backpropagation Neural Network (BPNN) to control nutrient solutions in a Nutrient Film Technique hydroponic, with sensors EC and pH as input. The training data of BPPN are obtained from the implementation of the fuzzy technique. Controlling nutrient solutions can use fuzzy methods, but it has a weakness: use greater power because sensors require continuous detection. By using BPNN method, sensors only detect once to perform the same control action. In this research, the outputs of both methods are the duration of pumps in active conditions to optimize the nutrient solution. Based on experiments, the best BPNN model has eight hidden layers with a learning rate of 0.8. The result accuracies which had been obtained by alkaline solution (pump A) was 90.77 %, 91.93% for acid solution (pump B), and 91.13% for nutrient fertilizer (pumps C and D). The result showed that the use of power for BPNN is less than fuzzy. The average total power used for BPNN method is 68.43% lower than the fuzzy method.