Development steps of the Data Analytics Toolbox: Ship.AI
The development steps of the proposed data analytics toolbox, Ship.AI, are presented in Figure 1. The main objective of this toolbox is to extract, visualize and analyze information from Big Data sets of ship performance and navigation parameters. That consists of various data driven models to transform such data sets of ship performance and navigation parameters into digital information that can be used towards digitizing the shipping industry.
The deployment layers of Ship.AI toolbox can be categorized as: Functions. Function Handler, Function Learner, Decision Support and Ship Intelligence.
The first layer of this toolbox consists of various functions. The respective features of such functions can be categorized as: Data Driven Models (Data Classification and Structural Identification), Sensor & DAQ (Data Acquisition Systems) Fault Detection, Parameter Reduction/Error Compression, Parameter Expansion/Data Recovery, Integrity Verification & Regression and Data Visualization & Decision Supporting.
The Function Handler layer is developed to execute these functions in a more organized format. It is believed that the respective Functions and Function Handler can be used by ship owners to analyze the respective data sets. The Function Learner layer is developed to adapt such functions with respective to various vessel types and operational conditions. That can also introduce additional flexibility in extracting, visualizing and analyzing the respective data sets. It is believed that the respective Function Handler and Function Learner can be used by service companies to analyze the respective data sets and that can improve their services.
The Decision Support layers is developed to accommodate the information that are extracted, visualized and analyzed through the respective data sets in real-time and that can be used to support navigator's decisions. It is believed that the respective Function Learner and Decision Support can be used by system developers to integrate such features into their systems to improve the system performance. The Ship Intelligence layer is developed not only to extract, visualize and analyze through the respective data sets in real-time but also to make appropriate navigation decisions on-board vessels. Such features can support towards remote and autonomous vessels and their operations. It is believed that the respective Decision Support and Ship Intelligence layers can be used by solution developers to integrate such features into their products to improve various remote and autonomous functionalities in vessels.
