IMPLEMENTATION OF REAL-TIME MONITORING SYSTEM TO DETERMINE RISK ANALYSIS AND MACHINE CONDITION OF MIXING MACHINE IN FACTORY

• Main Author: Ridlo, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73699
• Institution: Institut Teknologi Bandung

Implementation of real-time monitoring system for mixing machine aims to provide risk analysis of machine fault types and maintenance schedule recommendation to users. The users consist of maintenance supervisor and technician, operator, and product control. This thesis explains the details of data processing, which consists of gateway processing and cloud processing. The data processing involves preprocessing, vibration feature extraction, abnormality detection, vibration fault types detection, and maintenance schedule recommendations. The entire data processing requires input data consisting of three parameters: temperature, vibration, and RPM (Revolutions Per Minute), and produces output warnings to be displayed to the user through the interface, as well as providing recommendations for machine repair schedules. The deep learning model for vibration fault prediction involves RNN with the JANet cell. Based on the verification results, data processing is able to provide end-to-end processing with total execution time of 586,6252 ms. Besides, It achieves 100% accuracy of miscalculation and 98% accuracy of data acquisition for temperature and RPM, and achieves 63% accuracy for vibration. For the mounting design, the total weight of the components that is installed on the machine is 303 grams. Furthermore, the user test explains that the warning readings capability level scores 3,03 of 5, and the schedule description readings capability level scores 4,53 of 5 using the likert scale.