Purpose: The marine oil purifier is a critical piece of auxiliary equipment on autonomous ships that can cause major engine failures if not maintained properly. Diagnosis of auxiliary equipment failures, especially those caused by degradation of seals, is difficult using current diagnostic techniques. The aim of this study is to develop a database and algorithm that will identify areas where condition-based maintenance is necessary for rubber seals.
Methods: A multistep accelerated thermal aging test was performed on the marine oil purifier’s rubber seal. Using the purifier’s failure-simulation testbed, diagnostic data indicating the vibration-based condition due to the level of seal degradation was collected. Time and frequency domain analysis was performed using the established database, and a failure classification method using STFT and CNN was proposed.
Results: Tensile and hardness test results indicated that rubber seals experienced decreasing tensile properties when undergoing the accelerated thermal aging process. A valid difference in the frequency domain was observed in the equipment condition monitoring data. Failure classification methods indicated an accuracy of 99%.
Conclusion: In this study, a database and algorithm were developed for diagnosing the condition of rubber seals in marine oil purifiers. Their effectiveness was verified using time and frequency domain analysis and failure classification methods. The validity of the developed data-driven failure diagnosis method was successfully confirmed.