Pendekatan Hibrida Rule-Based System dan Support Vector Machine untuk Monitoring Serta Rekomendasi Kondisi Tanah pada Budidaya Kakao Berbasis IoT

Abstract

Agriculture modernization through smart farming requires a robust software infrastructure to process environmental data into actionable decisions. This study introduces a hybrid software framework combining a rule-based system (RBS) and a Support Vector Machine (SVM) integrated within an IoT-based monitoring system for soil classification and irrigation recommendation on cocoa farms in Jembrana Regency. The developed system utilizes an ESP32 microcontroller to acquire real-time streaming data from soil moisture, temperature, and pH sensors. The SVM algorithm is deployed on the cloud layer to classify multi-parameter soil conditions into non-linear agricultural states, while the rule-based system engineered at the edge layer triggers real-time irrigation recommendations for the solenoid valve actuators. Software engineering testing focused on SVM classification accuracy, system functionality, and mobile application response latency. The experimental results demonstrate that the hybrid software architecture achieved an SVM classification accuracy of 92.4% and 100% logic execution for actuator controls, with an average latency for data transmission of 2.4 seconds. This hybrid engine successfully optimizes watering decisions based on structural soil data, providing an efficient framework for automated green agriculture infrastructure