Project: Sub Project 4 - Performance in a Seaway
Author(s): Lokukaluge P. Perera, Brage Mo
Journal/conference: Proceedings of the 3rd International Conference on Maritime Technology and Engineering (MARTECH 2016)
Machine Intelligence for Energy Efficient Ships: A Big Data Solution
Appropriate navigation strategies are developed to overcome the challenges that are encountered by the shipping industry due to various emission control based energy efficiency measures.
Ship performance and navigation data are collected to develop such navigation strategies as an integrated part of the ship energy efficiency management plan (SEEMP). It is believed that the SEEMP with various navigation strategies will play an important part of e-Navigation under modern integrated bridge systems. Hence, onboard large scale data handling processes are implemented to facilitate in such situations, where effective navigation strategies should be based on accurate ship performance and navigation information. This study proposes a marine engine centered data flow chart for ship performance and navigation data monitoring to improve the quality of the respective navigation strategies as a big data solution. The proposed data flow chart is divided into two main sections of pre and post processing. The data pre-processing is an onboard applica-tion and that consists of sensor faults detection, data classification and data compression steps. Various sensor fault situations of ship performance and navigation data are identified in the first step of this process, where a machine learning (principal component analysis (PCA) is proposed. The data classification along the operat-ing regions of a marine engine is considered in the second step of this process, where a MI technique of Gaussian mixture models (GMMs) with an expectation maximization (EM) algorithm is proposed. Then, the classified data sets are compressed to reduce the number of parameters that communicated to shore based data centers. The data post-processing is a shore based application (i.e. in data centers) and that consists of data expansion, integrity verification and data regression steps. The compressed data set in ship performance and navigation information is expanded to its original dimensions at the first step of this process. Both steps of da-ta compression and expansion are conducted under another MI technique of Autoencoder (i.e. deep learning). Even though the compressed data set is expanded into its original dimensions, the new data set consists of the estimated parameters of ship performance and navigation information. Therefore, a comparison between both actual and estimated data sets is conducted to any variations. The second step of this process is to verify the integrity of estimated ship performance and navigation data, where the AIS data can be used. The final step in this process is data regression, where the estimated data points are used to estimate the required parameters of ship performance and navigation information.