Neural Networks for Dynamical Crop Growth Model Reduction and Optimization
A dynamic crop growth model of undeterminate tomato variety (TOMGRO) consists of a large number (69) of difference equations that describe age classes of different organs. In order to find a fast working and low-dimensional equivalent with dynamic Neural Networks (NN), several model reduction approaches has been tried, including aggregation, PCA transformation, and bottleneck NN compression. Reduced data were used to train dynamic NN. Simulations with the NN model produced trajectories which agreed well with the original trajectories of TOMGRO.
KeywordsNeural Network Model Reduction Mature Fruit Inverse Transformation Green Fruit
Unable to display preview. Download preview PDF.
- 2.Demuth, H., Beale, M.: Neural Network Toolbox for Use with Matlab. Users Guide, Natick, MA 01760, USA: MathWorks, Inc., 1992.Google Scholar
- 4.Arkin, H., Colton, G.: Statistical Methods, New York, USA: Barnes and Noble, Inc. 1953.Google Scholar
- 5.Sejnowski, T., Lawrence, D. and Golomb, B.: Sexnet: A Neural Network Identifies Sex from Human faces, in Advances in Neural Information processing systems 3. Touretzky, D., and Lippman, R., (eds.), San Mateo, CA: Morgan Kaufmann 1991.Google Scholar
- 6.Polyak, B.T.: Introduction to Optimization, New York: Optimization Software 1987.Google Scholar