Improvement of Fault Finding Techniques in Industrial Robots: Using Unsupervised Learning for Predictive Maintenance

A detailed comparative analysis of various machine learning models for predictive maintenance in industrial robotic systems is conducted. The focus is on constructing accurate models to forecast industrial robot breakdowns, enabling timely repairs, and reducing downtime. Data from industrial robots is gathered, feature engineering is performed, and multiple machine learning models are analyzed using the Silhouette Score. The goal is to identify the most effective model for predictive maintenance, aligning with the need for optimal system performance.