Prediction of Pakcoy ( Brassica rapa L.) Plant Canopy Area using ANN in Aeroponic System at Indoor Farming

Abstract The growth of Pakcoy ( Brassica rapa L.) plants in the aeroponic system at indoor farming requires accurate monitoring of environmental parameters to optimize production yields. One modern approach that can be used to predict growth parameters, such as canopy area, is the Artificial Neural Network (ANN), which can handle non-linear relationships between environmental variables and plant growth. This study aims to build a predictive model of the canopy area of the Pakcoy plant based on temperature, humidity, and light intensity data. The research methodology used a quantitative approach with 54 Pakcoy plants as a population, and 18 samples were observed for 30 days after transplantation (DAT). The data was divided into training and test data with an 80:20 ratio and analyzed using the feedforward ANN model with backpropagation. The results showed that the ANN model of the 3-5-1 architecture provided the best performance with an accuracy of 0.99 and an RMSE value of 0.09. A comparison graph between actual and predicted data shows an increase in predictive proximity as the number of neurons increases, with the highest R² value of 0.9904 in the five neurons. These findings confirm that the optimal ANN structure can improve the accuracy of plant growth predictions and can be a reliable tool in technology-based farming systems.