Supervised Learning for Microclimatic parameter Estimation in a Greenhouse environment for productive Agronomics

Abstract

Maximum crop returns are essential in modern agriculture due to various challenges caused by water, climatic conditions, pests and so on. These production uncertainties are to be overcome by appropriate evaluation of microclimate parameters at commercial scale for cultivation of crops in a closed-field and emission free environment. Internet of Things (IoT) based sensors are used for learning the parameters of the closed environment. These parameters are further analyzed using supervised learning algorithms under MATLAB Simulink environment. Three greenhouse crop production systems as well as the outdoor environment are analyzed for comparison and model-based evaluation of the microclimate parameters using the IoT sensors. This analysis prior to cultivation enables creating better environment and thus increase the productivity and harvest. The supervised learning algorithm offers self-tuning reference inputs based on the crop selected. This offers a flexible architecture and easy analysis and modeling of the crop growth stages. On comparison of three greenhouse environment as well as outdoor settings, the functional reliability as well as accuracy of the sensors are tested for performance and validated. Solar radiation, vapor pressure deficit, relative humidity, temperature and soil fertility are the raw data processed by this model. Based on this estimation, the plant growth stages are analyzed by the comfort ratio. The different growth stages, light conditions and time frames are considered for determining the reference borders for categorizing the variation in each parameter. The microclimate parameters can be assessed dynamically with comfort ratio index as the indicator when multiple greenhouses are considered. The crop growth environment is interpreted better with the Simulink model and IoT sensor nodes. The result of supervised learning leads to improved efficiency in crop production developing optimal control strategies in the greenhouse environment.