Abstract
Digitization has made its way into the shipping industry in recent years. Consequently, a vast amount of data is being generated, recorded, and utilized in the development of Digital Twins (DTs) designed for various applications, such as condition monitoring and efficiency improvements in industrial systems. Due to different reasons, such as malfunctions in onboard data acquisition systems in vessels, an assortment of anomalies and missing values can be observed among measured performance and navigation data in vessels. These issues can degrade the performance of the DT models and, therefore, the credibility of the resulting analysis. Consequently, cleaning the dataset is a crucial step in developing any data-driven model development framework, such as DTs in shipping. This process includes the identification, isolation, and ultimately, treatment of anomalies and missing values in datasets from ocean-going vessels. The main contribution of this study is the introduction of a data quality improvement framework for identifying and isolating anomalies, as well as recovering the missing values by estimating appropriate values. To better illustrate the proposed framework, a dataset from an oceangoing vessel is used as a sample case study, and the results of applying the proposed methods to it are presented.