Optimal Control and Neural Networks Applied to Peanut Irrigation Management

Abstract

A method was developed to capture the results of a computationally intensive irrigation optimization routine through the use of neural networks. The PNUTGRO peanut crop growth simulation model was modified and incorporated into a routine to search for optimal irrigation decisions using the Sequential Control Search approach. The daily environmental conditions and crop state variables associated with these optimal irrigation sequences were used to train a neural network. The resulting neural network was incorporated as a subroutine in PNUTGRO and was used as an irrigation scheduling policy. The simulated net returns above irrigation costs using the neural network irrigation control policy were only $4/ha less than the average for the same seasons using the Sequential Control Search optimization.