Performance monitoring I:
Pattern recognition and neural networks

Pattern recognition and neural networks may significantly improve the quality of performance monitoring.

Ph.D. student Benjamin Pjedsted Pedersen at FORCE Technology is looking into new statistical methods of neural network and pattern recognition for analysing operational data from ships. The aim is to use these methods to determine the hydrodynamic performance of ships with much higher accuracy than traditional empirical models.

In his study, Benjamin Pedersen has concluded that advanced statistical methods like neural networks and pattern recognition are suitable methods for analysis of performance data.

By applying advanced neural networks on a large set of data points for a tanker, Benjamin has verified that it is possible to predict the propulsion power with a relative error of less than 2.7% based on very few input parameters. 

In addition, Benjamin has verified and concluded that by using hindcast weather data from meteorological institutes instead of visual observations of the crew, it is possible to reduce the scatter of the performance analysis.

The advanced neural networks may also be applied to operational data to determine the relation between propulsive power and significant wave height. This is very useful for both design and operational purposes.

Papers on the topic:

	Modeling of Ship Propulsion Performance
	Prediction of Full Scale Propulsion Power Using Artificial Neural Networks