Prediction Model of Greenhouse Eggplant Transpiration Based on RBF Neural Network

摘要

In order to reveal the greenhouse crop transpiration law, a test on eggplant transpiration rate was carried out in a Venlo-type greenhouse in North China University of Water Resources and Electric Power, Zhengzhou, China. Through the correlation anylsis of transpiration rate and the measured indoor conventional meteorological factors, four factors were selected as the main influence parameters to build a transpiration rate prediction model, including the indoor ground temperature, relative humidity, plant canopy temperature, and solar radiation. Using neural network toolbox in MATLAB, the neural network model based on RBF (Radial Basis Function) artificial neural network was built, which had a topological structure of 4-11-1. After model validation, the results indicated that this model based on RBF had a high prediction precision, the average relative error between the predicted and measured value is =1.01%, and the index of simulation effectiveness is EF=82.41%. This research has reference value for crop water demand calculation and crop water consumption regulation in greenhouse.