On the Classification of a Greenhouse Environment for a Rose Crop Based on AI-Based Surrogate Models

A precise microclimate control for dynamic climate changes in greenhouses allows the industry and researchers to develop a simple, robust, reliable, intelligent model. Accordingly, objective of this investigation was method that can accurately define most suitable environment greenhouse an optimal yield roses. Herein, highly accurate BO-DNN surrogate model developed (based on 300 experimental data points) quick reliable classification rose considering some influential variables including soil humidity, temperature humidity air, CO2 concentration, light intensity (lux) into its architecture. Initially, two BO techniques (GP GBRT) are used tuning process hyper-parameters (such as learning rate, batch size, number dense nodes, neurons, input activation function, etc.). After that, simple combination selected DNN algorithm based points, which further classify (the environments were classified four classes such without water, correct environment, too hot, very cold environments). The high accuracy proposed (0.98) originated from introduction vital meteorological parameters inputs help identifying efficient crop yields.