Optimization of ANN Adaptation Time for the Modeling of Greenhouse Climate Using Wavelet Transform

Abstract

In this paper we optimize the adaptation time taken by an Artificial Neural Network (ANN) to model the indoor climate greenhouse system. For this purpose the discrete wavelet transform is applied to transform outdoor/indoor climatic signals of a greenhouse. The considered greenhouse is a nonlinear MIMO process with four inputs and two outputs. Results simulation had shown that ANN metrics (mean square error and adaptation time) are improved using Haar wavelet with one decomposition level.