Analysis Track Condition Trends Through Sensor Data and Clustering Techniques

Railway safety is critical for urban and suburban transit, where increasing train frequency, capacity, and speed heighten operational risks. This study uses sensor data, including velocity, latitude, longitude, and acceleration, from Bandung commuter lines between August 2023 and July 2024 to analyze track conditions with clustering techniques. K-Means with k=2 grouped tracks into two categories. Cluster 0 consists of winding, longer tracks with higher average velocity and acceleration, indicating complex and higher-risk routes. Cluster 1 includes shorter, straighter tracks with lower velocity and acceleration, reflecting safer conditions. A comparative analysis with Gaussian Mixture Models confirmed K-Means' effectiveness, achieving a silhouette score of 0.480. The findings highlight the need for interventions such as speed limits and increased monitoring on high-risk tracks, while safer routes require fewer measures. This approach demonstrates how clustering can support proactive railway safety management.