ENHANCING CNC MILLING PROCESS EFFICIENCY THROUGH DATA- DRIVEN PREDICTIVE MAINTENANCE

• Main Author: Harpin Asrori, Achmad
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/80975
• Institution: Institut Teknologi Bandung

In Industry 4.0, several techniques are used: IoT, machine learning, artificial intelligence, etc. This technology is tested to improve industrial quality and productivity. Several manufacturing industries are working on smart manufacturing techniques integrated with multiple sensors in the machine. These manufacturing sensors are synchronized with various systems through IoT and used for different predictive management. The manufacturing industry strives to improve operational efficiency, particularly using Computer Numerical Control (CNC) milling machines, which play an important role in modern production processes. The research centers on data- driven predictive maintenance to optimize machine performance and reduce unexpected downtime. This research aims to obtain the best prediction model to classify the condition of production machines by comparing machine learning models. Predictive maintenance models are expected to be able to estimate machine maintenance schedules, extend the life of production machines, and show maintenance costs. The analysis method revolves around classification analysis, comparing six classification models: Decision Tree, Random Forest, K-Nearest Neighborhood, Gradient Boosting, Gaussian Naïve Bayes, and Multi-Layer Perceptron. This algorithm comparison aims to identify the optimal classification model for the case of production machines. Among the six algorithms, the best model was obtained from the Random Forest Model with an accuracy of 97.9%.