Data Analytics for Capturing Marine Engine Operating Regions for Ship Performance Monitoring

Abstract

This study proposes marine engine centered data analytics as a part of the ship energy efficiency management plan (SEEMP) to overcome the current shipping industrial challenges. The SEEMP enforces various emission control measures, where ship energy efficiency should be evaluated by collecting vessel performance and navigation data. That information is used to develop the proposed data analytics that are implemented on the engine-propeller combinator diagram (i.e. one propeller shaft with its own direct drive main engine). Three marine engine operating regions from the initial data analysis are noted under the combinator diagram and the proposed data analytics (i.e. data clustering methodology) to capture the shape of these regions are implemented. That consists of implementing the Gaussian Mixture Models (GMMs) to classify the most frequent operating regions of the marine engine. Furthermore, the Expectation Maximization (EM) algorithm is used to calculate the respective parameters of the GMMs. This approach can also be seen as a data clustering algorithm that facilitated by an iterative process for capturing each operating region of the marine engine (i.e. the combinatory diagram) with the respective mean and covariance values. Hence, these data analytics can be used in the SEEMP to monitor the performance of a vessel with respect to the marine operating regions. Furthermore, it is expected to develop advanced mathematical models of ship performance monitoring under these operational regions of the marine engine as the future work.

Marine Engine Operating Regions under Principal Component Analysis to evaluate Ship Performance and Navigation Behavior

“…An approach based on Gaussian mixture models and an expectation-maximization algorithm is in this step implemented to cluster the respective data set in ship performance and navigation information. Three specific operating points of the marine engine is identified in this data set and an overview of the approach is presented in Perera and Mo (2016a). The results (i.e.…”

“…Since marine engines of vessels are operating around various operating points, operating points are appropriate mean values for such data clusters in ship performance and navigation information. The proposed data clustering approach consists of Gaussian mixture models (GMMs) with an expectation maximization (EM) algorithm and uses to identify such marine engine operating regions (Perera and Mo, 2016a). Then, the data set should equally be centered and scaled (i.e.…”

Marine Engine Centered Localized Models for Sensor Fault Detection under Ship Performance Monitoring

“…Both main engine (ME) power (kW -log scale) and propeller shaft speed (rpm) values are presented in this diagram. It is noted that three operating regions are frequently used by this vessel, therefore Gaussian mixture models (GMMs) with an expectation maximization (EM) algorithm is used to identify such regions (Perera and Mo, 2016d). These operating regions are classified as localized models and that use to create the respective data clusters.…”

“…Therefore, the boundaries of each operating region of the engine-propeller combinator diagram are determined and these regions classify localized models. The E-step is initiated by considering a multivariate GMM and denoted as (Perera and Mo, 2016d): …”

“…This structure is developed by considering the respective studies of ship performance and navigation monitoring (Perera and Mo, 2016a, 2016b, 2016c, 2016d and that consist of various data analysis and visualization tools and techniques. The figure consists of two layers to identify of the respective sensor fault situations in an onboard DAQ system.…”