The study on the appearance of deformation defects in the yacht lamination process using an AI algorithm and expert knowledge

TL;DR

This paper uses AI to identify patterns in yacht lamination defects, improving production efficiency and reducing errors.

Contribution

The novel use of the A-priori algorithm to derive defect relationships in yacht lamination, validated against expert knowledge.

Findings

The A-priori algorithm identified association rules between lamination defects and deformation issues in yacht shells.

AI-derived rules were consistent with expert opinions and revealed additional root causes overlooked by departments.

Applying AI improved objectivity in defect analysis, reducing errors and production time.

Abstract

This article describes the application of the A-priori algorithm for defining the rule-based relationships between individual defects caused during the lamination process, affecting the deformation defect of the yacht shell. The data from 542 yachts were collected and evaluated. For the proper development of the algorithm, a technological process of the yacht lamination supported by expert decisions was described. The laminating technology is a complex process of a sequential application of individual laminates according to a special strategy. The A-priori algorithm allowed for obtaining the set of association rules defining the relationships between the defects resulting from the lamination process and influencing the deformation defect of the yacht shell, which is one of the most common errors in yacht production. The obtained aggregated rules were compared with the expert knowledge of the employees of the production, quality control, mould regeneration, and technology departments of the yacht yard. The use of the proposed A-priori algorithm allowed for the generation of relationship rules consistent with the general opinion of experts. Associative rules additionally took into account detailed causes of a specific error, which were not always noticed by employees of specific departments. The assessment of the lamination process using an artificial intelligence algorithm turned out to be more objective, which made it possible to gradually reduce the total number of errors occurring in the yacht shell lamination process, and thus shorten the time needed to repair errors and the total time of producing the yacht.

Figures