Artificial Intelligence of Smart Agriculture with Multiple Cropping in NFT Hydroponics System

Abstract

This research aims to develop a model to predict plant growth in hydroponics using machine learning. Sensors are developed for measuring environmental factors such as temperature, humidity, temperature. Water, pH and EC by using the ESP 32 microcontroller board, receive data from sensors to send data to MySQL database. In system development, use C and C# for programming onto microcontroller boards, use PHP, Java Script and AJAX languages for web development and dashboard design, use Python for data analysis and prediction using algorithms. Machine learning algorithms include logistic regression, K-NN, random forest, decision tree, and bayesian network, and determine which model has the most accuracy for this dataset. The data stored in the database contains information that measures factors affecting plant growth, including temperature, humidity, water temperature, pH and EC. Light intensity uses artificial incandescent light with an appropriate range of light. plant The data to be stored in the database will store time based on the actual measurement, which is collected every 20 minutes. The results showed that the random forest model was the most accurate by using two precision measurement methods: split test and cross validation test, both of which used logistic regression model, K-NN, random forest, decision tree. and the Bayesian network. In the performance test, the most accurate model for this test dataset was the random forest with an accuracy of approximately 98%. The accuracy of these predictions depends on the data used. In addition, the sensor display is also available through the web page, which displays data in numerical, graph and table formats so that users can get the measurement data of the hydroponic environment which can be viewed. from any device connected to the Internet. This research aids in user decision making for growing in hydroponics systems and is able to effectively control crop yields and is also a model guideline for using the system. within the household to be able to grow crops for consumption and sale.