Design of ship monitoring system based on unsupervised learning

Abstract

This paper aims to present an integrated methodology for designing of the ship monitoring system using machine learning algorithms. We collected ship monitoring data such as engine room monitoring system and AIS, and integrate and fuse data to form integrated ship information platform, the proposed methodology will train models using complete voyage data and then classify new data points using the improved Gaussian Mixture Model to get the most frequent operating regions of the main engine. Finally, we propose a scheme for performance evaluation of equipment using principal component analysis (PCA). This work will provide a flexible but robust framework for the early detection of emerging machinery faults. And provides a new way of thinking for the design of engine room monitoring system.