A Study on the Application of Discrete Wavelet Decomposition for Fault Diagnosis on a Ship Oil Purifier

With the development of Internet things, big data, and AI leading 4th industrial revolution, it has become possible to acquire, manage, analyze vast diverse condition signals from various machinery facilities. In addition, been revealed that large amounts acquired facilities can be utilized for fault diagnosis. Currently, while data-driven diagnosis techniques applicable are being developed, tried apply fully autonomous ships in shipbuilding shipping industry. Since must able detect diagnose failures on their own real time, overall research is required how acquire ship use them failures. this study, a framework was proposed condition-based maintenance (CBM) oil purifiers, which an auxiliary facility engine system ship. First, purifier test-bed simulating faults built obtain data state equipment. After extracting features using discrete wavelet decomposition were visualized by t-distributed stochastic neighbor embedding, used train support vector machine-based diagnostic models. Finally, trained models evaluated with Accuracy F1 score, some scored 0.99 or higher, confirming high performance. This study as reference establishing CBM system. Furthermore, expected improve safety reliability purifiers Degree 4 MASS.