Building large structures such as ships requires efficient management of production information. In shipyards, enterprise management information that includes production information is called master data, and which includes BOM (bill of material), WBS (work breakdown structure), basic unit and lead time. Master data related to production is closely related to time information. In the shipbuilding industry, however, the high variability of the shipbuilding process has made it difficult to gain the reliability of master data management. Low accuracy of master data can lead to financial losses as well as confusion of work.
To solve these problems, shipyards and related academia have been making various efforts to improve the master data system. However, most of the existing research is focused on traditional engineering perspectives and does not reflect the rapidly changing shipbuilding production environment.
Recently, machine learning methodologies have been widely applied to the manufacturing industry, along with the rapid development of big data related technologies. Machine learning is a correlation analysis technique of vast amounts of data, which is known to be able to overcome the limitations of causation analysis with traditional engineering methodologies. Research is also being conducted in the shipbuilding industry by applying various machine learning algorithms to systematically manage the production master data.
In this paper, I would like to introduce the study of applying ensemble learning, which is known to maximize the performance of machine learning, to the analysis of shipbuilding production master data. In addition, we would like to examine the applicability of production management tasks in actual shipyards by comparing the prediction results of the ensemble learning with the results of applying various learning algorithms and attaching a better performance learning model to the simulation.