Artificial Neural Network Model Use for Particulate Matter Evaluation from Ships in Klaipeda Port

Abstract

This publication deals with the evaluation of forecasting the emissions of ships operating in the port through neural networks. Analyzed particulate matter (PM1, PM2.5, PM10, TSP) emissions from ships at various parts of the port. The research is based on usage of AIS system data, the technical database of the ship, the ambient air measurement data and the ambient air pollution measuring data for the use of neural network training. Results showed that trained neuronal networks could be sufficiently accurate (the correlation coefficient amounted from 0.82 to 0.92 depending on pollutant) to use for ship operating in the port emissions evaluation.