PREDICTIVE MAINTENANCE OF MARINE DIESEL ENGINES: CURRENT APPROACHES TO CONDITION MONITORING AND SHIPBOARD INTEGRATION

Ünver B., Sönmez H. İ., Ekin F.

ASES INTERNATIONAL EUROPEAN SCIENTIFIC RESEARCH CONGRESS, Munich, Almanya, 21 - 23 Haziran 2026, ss.1-20, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Munich
  • Basıldığı Ülke: Almanya
  • Sayfa Sayıları: ss.1-20
  • Van Yüzüncü Yıl Üniversitesi Adresli: Evet

Özet

Predictive maintenance in marine diesel engines has developed from a monitoring-centered objective into a broader problem of converting machinery data into reliable maintenance action under operational shipboard conditions. This study reviews the current literature on condition monitoring, fault diagnosis, prognostic development, and onboard implementation for marine diesel engines. The source base includes thermodynamic monitoring, vibration-based diagnosis, oil and tribological analysis, multi-sensor fusion, machine-learning classification, deep-learning prediction, digital-twin frameworks, and real-time decision-support applications. Across these studies, the strongest evidence appears in areas where diagnostic interpretation remains close to physically meaningful signals and well-bounded component faults such as injector malfunction, valve leakage, lubrication-related degradation, or turbocharger abnormalities. The literature is less mature where the task shifts from diagnosis to prognostics, especially in early warning calibration, remaining useful life estimation, uncertainty handling, and maintenance-decision integration. Many studies report technically strong classification or detection performance, yet fewer demonstrate stable behavior under variable loading, sensor noise, incomplete data, and real fleet operating constraints. The review indicates that the main bottleneck is no longer sensing alone. It is the construction of trustworthy, low-burden maintenance logic from heterogeneous evidence streams. The most valuable next step for the field is therefore not another isolated high-accuracy model, but integrated shipboard systems that combine data quality control, multi-sensor reasoning, physics-informed interpretation, and execution-ready maintenance support.