Neural Network Modeling and Identification of Naturally Ventilated Tropical Greenhouse Climates

Naturally ventilated tropical greenhouse is classified as a complex system because it involves with nonlinear process and multivariable system. The purpose of this study is to determine the mathematical model of NVTG climates to in order to describe and predict the dynamic behavior of temperature and humidity inside NVTG for development of its control system. The modeling of the system is divided into two parts namely parametric and nonparametric modeling. The auto regressive with exogenous input, ARX and nonlinear auto regressive with exogenous input, NARX model were used for the parametric models while neural network auto regressive with exogenous input, NNARX model was used for the nonparametric model. The recursive least square estimation, RLSE, was used for the parameter estimation of the parametric model while the artificial neural network, ANN used the Levenberg-Marquardt method for predicting the performance parameter of NVTG system for the nonparametric model. All the models established were validated using statistical validation method such as mean square error (MSE), root mean square error (RMSE), error index (EI) and for the nonparametric model it added with the correlation coefficient, (R). From the result, the best model for parametric model is identified with the error index of 0.0573 for temperature and 0.0362 for humidity. The best non-parametric model gives error indexes of 0.0025 for temperature and 0.0024 for humidity. Key-Words: Parametric, Non-parametric, Identification, Neural Network, Naturally Ventilated Tropical Greenhouse.