The Concept Design of Rice Quality Detection System Using Model-Based System Engineering Approach

Quality of rice is determined by several factors such as water content, broken grains, and whiteness. The approach often used for the measurement is manual, time-consuming, and prone to error. Therefore, this research proposes a faster and more accurate rice quality detection system using Model-Based System Engineering (MBSE) approach. System was based on the needs analysis presented through an activity diagram showing the components and activities flow. Logical architecture diagrams were also used to structurally describe system logic to be further abstracted to the physical architecture stage. Moreover, machine learning techniques were used to simulate rice quality data analysis using the decision tree classification with the Iterative Dichotomizer 3 (ID3) algorithm. The simulation was applied to 200 supervised random datasets divided into 80% training and 20% test data with a focus on three attributes, including water content, broken grains, and whiteness. System design was developed using Visual Paradigm Community Edition software and the data were analyzed through the application of R software. The ID3 algorithm simulation produced rice quality detection system with a 92.5% accuracy rate, where 53% of rice was classified as good and 47% as bad. The proposed conceptual design for rice quality detection can be a starting point for the development of an industrial-scale system design.