Document Type

Conference Proceeding

Publisher

Australian Society of Information and Communication Technologies in Agriculture

Faculty

Faculty of Health, Engineering and Science

School

School of Computer and Security Science/eAgriculture Research Group

RAS ID

18561

Comments

This article was originally published as: Leng, J. , Neuhaus, A., & Armstrong, L. (2014). A network that really works - the application of artificial neural networks to improve yield predictions and nitrogen management in Western Australia. Proceedings of Asian Federation for Information Technology in Agriculture. (pp. 298-306). Perth, W.A. Australian Society of Information and Communication Technologies in Agriculture. Original article available here

Abstract

Yield predictions are notorious for being difficult due to many interdependent factors such as rainfall, soil properties, plant health, plant density etc. This study is based upon the author’s previously published work and extends its findings by further investigating the best mathematical solution to this dilemma. Artificial intelligence (AI) techniques have been applied to a large set of soil, plant, rainfall, and yield data from CSBP’s field research trial program. Here we further differentiate by investigate two ANN techniques, a genetic algorithm with back propagation neural networks (GA-BP-NN) and a particle swarm optimization with back propagation neural networks (PSO-BP-NN). Results indicate that the GA-BP-NN technique offers a slightly better yield correlation. Our main conclusion is that this method would offer growers more confidence in making a better nitrogen decision during the season than currently available models due to more precise yield predictions. It also can also contribute to possibly better grain marketing decisions later in the season.

Access Rights

free_to_read

Share

 
COinS