Fault Diagnosis and Condition Monitoring of Marine Diesel Engines

Marine diesel engines are the primary drivers of global shipping, demanding exceptionally high reliability under harsh and variable operating conditions. Condition monitoring involves continuous acquisition and analysis of sensor data—such as vibration, in-cylinder pressure and exhaust temperature—to detect deviations from normal performance. Fault diagnosis encompasses the detection, isolation and identification of specific failure modes, ranging from injector malfunction to turbocharger fouling. Traditional model-based methods rely on physical and thermodynamic models to predict expected behaviour, whereas data-driven approaches exploit machine learning to extract latent features from high-dimensional signals. Prognostics and health management frameworks integrate these tasks to estimate remaining useful life and support predictive maintenance decisions. Advances in sensor technology, data fusion and artificial intelligence have enabled more accurate early warning systems, reducing unplanned downtime, optimising maintenance schedules and lowering environmental emissions. Despite progress, challenges remain in distinguishing sensor faults from engine faults, coping with variable load patterns and ensuring interpretability of complex algorithms in safety-critical maritime applications.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Fault Diagnosis and Condition Monitoring of Marine Diesel Engines publication trend

The graph below shows the total number of articles in fault diagnosis and condition monitoring of marine diesel engines across all publications each year (not limited to Nature Index journals).

Technical terms

Condition monitoring: Continuous tracking of engine parameters via sensors to detect performance deviations.

Fault diagnosis: Process of detecting, isolating and identifying specific engine failure modes.

Prognostics and health management (PHM): Integrated framework for monitoring health, diagnosing faults, predicting remaining useful life and guiding maintenance.

Remaining useful life (RUL): Estimated time period before an engine component fails or performance degrades below a threshold.

Convolutional neural network (CNN): Deep-learning architecture that extracts spatial or temporal features from multi-dimensional input.

Bidirectional long short-term memory (BiLSTM): Recurrent neural network that processes sequential data in forward and reverse temporal directions.

Attention mechanism: Neural network component that dynamically weights relevant features over time or space.

Explainable artificial intelligence (XAI): Methods that make machine-learning decision processes transparent and interpretable to users.

Shapley Additive Explanations (SHAP): Game-theoretic approach to quantify individual feature contributions to model outputs.

References

1. Marine Systems and Equipment Prognostics and Health Management: A Systematic Review from Health Condition Monitoring to Maintenance Strategy. Machines (2022).
2. Explainable Anomaly Detection Framework for Maritime Main Engine Sensor Data. Sensors (2021).
3. A Deep Learning-Based Fault Warning Model for Exhaust Temperature Prediction and Fault Warning of Marine Diesel Engine. Journal of Marine Science and Engineering (2023).