The maintenance of sensors in Alfa Laval’s PureBallast 3 ballast water treatment system presents significant challenges, particularly due to the unpredictable degradation of sensor accuracy over time. This thesis explores the use of unsupervised methodologies to detect and analyze sensor measurement errors, aiming to enhance operational efficiency and safety while reducing maintenance costs. Using sensor data from 39 machines, the study focuses on the UV Light Intensity(ULI) and Flow Transmitter (FT) sensors, leveraging exploratory data analysis, correlation analysis, the Mann-Kendall Tau test, and LOESS smoothing to identify trends indicative of sensor degradation. The results demonstrate that trend analysis can effectively uncover sensor degradation issues, providing statistically significant evidence for these trends. By comparing the behaviour of potentially faulty sensors to those of healthy ones, this research highlights the potential for a condition-based maintenance strategy, which could offer operational and economic benefits. Despite limitations such as the reliance on correlation analysis and the absence of labelled data, the study sets a new standard in the maintenance of ballast water treatment systems, ensuring safer maritime operations and preserving the integrity of marine ecosystems.