DESIGN OF SMART FARMING ARCHITECTURE FOR CONTROL PROCESSES IN GREENHOUSES TO SUPPORT THE IMPROVEMENT OF AGRICULTURAL PRODUCTIVITY IN WEST BANDUNG REGENCY

• Main Author: Tasyrikah Fauziah, Rifda
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/84074
• Institution: Institut Teknologi Bandung

Indonesia is an agrarian country where agriculture serves as the backbone of the nation's economy. According to data obtained from the Badan Pusat Statistik (BPS), the number of farmers using agricultural land in Indonesia is 27,799,280. West Java, with 3,135,522 farmers, is among the three provinces with the highest number of farmers in Indonesia. West Bandung Regency is one of the suppliers of agricultural products from West Java Province, with 155,537 farmers. However, data of agricultural production in West Bandung Regency from 2020 to 2024 shows a decline. Factors contributing to this are climate, pests, and diseases. A review of the literature reveals several developments previously undertaken by academic practitioners, such as the development of environmental condition control systems for cultivating tomato or spinach plants. There have also been developments in machine learning models for detecting diseases affecting Pinaceae plants. However, no comprehensive solution has been proposed that integrates both functionality to address the needs of various crop varieties. Therefore, it is necessary to design a smart system that can help solve these problems. The methodology used in this project is DSRM. The final outcome is a system architecture design. The design was developed through a design process using the ADM method. The resulting architecture includes the business, data, application, and technology layers. The system has two key features. The first feature involves controlling temperature, humidity, and light intensity by utilizing IoT sensor technology and actuators connected to devices such as exhaust fans, sprinklers, and shading nets to control environmental conditions. The second feature is pest and disease detection using machine learning technology to identify threats based on visual data captured by cameras installed in the greenhouse. Guidance on how to handle the detected threats will then be provided from the system. iv The design results are evaluated through two stages: verification and validation. Verification was conducted by assessing whether the design met the identified requirements and objectives. Each system requirement and architectural objective will be evaluated to determine whether it was fully met, partially met, or not met. The evaluation results indicate that the design has met 80% of the system requirements and 100% of the architectural objectives. Additionally, since the architectural design was created by dividing the architecture into four layers, as in enterprise architecture, the design was also evaluated using the EA scorecard. The evaluation results indicated that all architectural layers received scores above 50%, confirming that the design was well-constructed. Validation was carried out through simulation. The simulation is performed by creating scenarios that represent potential situations in the field. The system’s impact on these situations was then assessed. The impact to be evaluated is the efficiency of the process, with the metric used being time. The simulation results concluded that the control activities during plant cultivation in the greenhouse became more efficient and improved accuracy in ensuring that the planting conditions were always optimal. This, in turn, has implications for increasing agricultural productivity in West Bandung Regency.