Development of an Artificial Intelligence-Based Decision Support System for Engine Room Watchkeeping Using Real-Time Machinery Parameters

Abstract

Engine room watchkeeping requires continuous monitoring and rapid decision-making to ensure machinery reliability and ship safety. Conventional watchkeeping practices depend heavily on officers' experience and manual interpretation of machinery parameters, leading to delayed responses and increased risk of machinery failure. This study proposes an Artificial Intelligence-Based Decision Support System (AI-DSS) for engine room watchkeeping using real-time machinery parameters. The system integrates machine learning algorithms with operational data, including lubricating oil temperature, cooling water temperature, exhaust gas temperature, fuel consumption, and vibration signals. Data were collected from a marine engine simulator and ship operational records. Random Forest and Artificial Neural Network models were employed to predict machinery abnormalities and generate early warnings. The results indicate that the proposed system achieved a prediction accuracy of 94.6%, significantly improving anomaly detection compared to conventional monitoring methods. Furthermore, the developed Engine Room Watchkeeping Index (ERWI) provides a quantitative assessment of machinery conditions during watchkeeping operations. The proposed model can support marine engineers in making timely and accurate decisions, thereby improving operational safety and machinery reliability.