Design of a Chayote Jelly Candy Stirring System Based on Color Sensors and Artificial Neural Networks

Abstract

The utilization of chayote (Sechium edule) as a raw material for jelly candy production has potential to increase agricultural value. However, the production process still faces challenges during the stirring stage and in determining the dough maturity level, which are commonly performed manually. This manual approach is subjective, inconsistent, and reduces process efficiency and product quality. This study aims to design and implement an intelligent control based chayote jelly candy stirring device using an artificial neural network to determine the dough maturity level automatically. The system was developed using a TCS3200 color sensor to detect changes in dough color in RGB components, a MAX6675 thermocouple sensor to monitor cooking temperature, and an Arduino Uno microcontroller as the main for data processing and actuator control. The RGB color data were processed using a backpropagation artificial neural network method to classify the dough condition. Network training was conducted using 170 training datasets with 10,000 epochs. Experimental results show that the system can identify the dough maturity level with 95.25% accuracy. The mature dough condition was detected at average color values of R = 154.2, G = 167, and B = 86.5 within a cooking temperature range of 70-80

°C. The system also displays information on temperature, color values, and dough maturity status, while automatically controlling the stirring motor. Therefore, the developed device improves consistency, efficiency, and reliability in chayote jelly candy manufacturing.