PREDICTIVE NEURAL NETWORK MODELING OF pH AND ELECTRICAL CONDUCTIVITY IN DEEP–TROUGH HYDROPONICS

Abstract

A model is presented that predicts pH and electrical conductivity (EC) changes in the root zone of lettuce (Lactucasativa cv. Vivaldi) grown in a deeptrough hydroponic system. A feedforward neural network is the basis of that modeling.The neural network model has nine inputs (pH, EC, nutrient solution temperature, air temperature, relative humidity, lightintensity, plant age, amount of added acid, and amount of added base) and two outputs (pH and EC at the next time step).The most suitable and accurate combination of network architecture and training method was one hidden layer with ninehidden nodes, trained with the quasiNewton backpropagation algorithm. The model proved capable of predicting pH at thenext 20minute time step within 0.01 pH units and EC within 5 .Scm1. Simpler prediction methods, such as linearextrapolation and the lazy man prediction (in which prediction is the value of the previous time step), gave comparableaccuracy much of the time. However, they performed poorly in situations where the control actions of the system had beenactivated and produced relatively rapid changes in the predicted parameters. In those cases, the neural network model didnot encounter any difficulties predicting the rapid changes. Thus, the developed model successfully identified dynamicprocesses in the root zone of the hydroponic system and accurately predicted onestepahead values of pH and EC.