An Automated Evaluation Method of Tool Surface Condition in Composites Manufacturing Using Machine Learning and Sparse Sensing

Abstract

The interaction between a tool and part during composites manufacturing may significantly contribute to the formation of residual stresses and process-induced deformations (PIDs). Aerospace manufacturers typically treat tool surfaces with release coatings to minimize tool-part interaction. During autoclave processing, tool surfaces undergo physicochemical changes due to the aging of release coating and cure-induced contamination. The aging of release coating and surface contamination may strongly influence the level of tool-part interaction and PIDs in subsequent processing cycles. Production tools are frequently retreated with fresh layers of release coating and intermittently cleaned to mitigate aging and contamination effects. However, release coating reapplications and tool surface cleaning are often performed based on know-how in production environments, leading to cost-deficient tool preparation schedules, and in some cases excessive PIDs. This paper presents an automated method to evaluate the surface condition of large production tools for composites manufacturing using machine learning (ML) and sparse sensing. ML methods are used in conjunction with multiple measurement techniques for fast global digital mapping and sparse local physicochemical evaluation of a tool surface. The technology can be integrated into a robotic platform for automatic inspection of large tools to improve production efficiency and potentially mitigate PIDs in composites manufacturing.