Real-Time Classification of Diesel Marine Engine Loads Using Machine Learning.

Area of Science:

• Internal Combustion Engine Systems
• Artificial Intelligence in Engineering
• Signal Processing

Background:

• Engine control systems optimize combustion parameters for efficiency.
• Engine load significantly impacts optimal parameter settings.
• Real-time engine load information is crucial for operational efficiency.

Purpose of the Study:

• To develop a real-time engine load classification method.
• To improve internal combustion engine efficiency through accurate load detection.
• To utilize sensed signals for dynamic parameter adjustment.

Main Methods:

• A magnetic pickup sensor measures rotational speed from a V12 marine diesel engine.
• Measured signals are converted into crank angle degree (CAD) signals.
• An artificial neural network is trained using processed CAD signals for classification.

Main Results:

• The artificial neural network achieved 99.4% accuracy in classifying five engine load classes.
• The system effectively classifies engine loads in real-time.
• Crank angle degree signals proved to be effective input features.

Conclusions:

• Real-time engine load classification is feasible using artificial neural networks and CAD signals.
• This method enables dynamic optimization of engine control parameters.
• The proposed system enhances internal combustion engine efficiency.