A Framework on Optimizing Greenhouse Conditions for Plant Growth using Machine Learning Based on Time-Series Data

Abstract

This study presents a machine learning (ML) framework for optimizing greenhouse environments to promote efficient plant growth. The approach utilizes time-series data to analyze and predict optimal growing conditions by integrating morphological traits—such as plant height, leaf size, and canopy structure—with dynamic environmental variables, including temperature, humidity, light intensity, and CO₂ levels. The proposed framework follows a structured, six-stage workflow comprising: data acquisition, data preprocessing, feature engineering, model selection and training, model evaluation, and prediction and decision support.. High-resolution sensors continuously collect environmental and plant-related data, which are then cleaned and formatted to handle missing values and temporal inconsistencies. Feature engineering techniques are applied to extract relevant growth indicators and environmental patterns. This framework lays the foundation for an intelligent decision support system that can assist growers in making informed environmental control decisions. By aligning greenhouse microclimates with the specific needs of crops, this approach aims to maximize yield, reduce energy consumption, and minimize resource waste. The adaptability of the framework also allows for its application across various plant species and cultivation setups, making it a scalable solution for modern agricultural operations.