In-situ surface inspection for wire-arc directed energy deposition integrating 3D topography reconstruction, defect detection and roughness measurement

Abstract

In wire-arc directed energy deposition (WA-DED), the in-situ surface inspection is imperative for acquiring 3D topographies, detecting weld bead surface defects, and measuring surface roughness to facilitate autonomous quality evaluation and repair path generation. However, existing offline inspection methods fail to provide comprehensive, real-time surface characterization. To address this limitation, a laser vision-based in-situ inspection system, where 3D topography reconstruction, defect detection and roughness measurement are innovatively integrated into a unified real-time scanning task, is developed. The inspection system consists of three subprocesses running simultaneously. By employing the novel laser centerline extraction algorithm CIICEA alongside the novel spatial curve fitting algorithm K-QCBFA, the complete 3D topographies of the printed geometries are reconstructed with a high precision of 0.247 mm. Surface hump and depression defects on the weld bead are identified and classified with a high accuracy of 97.44 % by applying the specific defect detection criterion derived from the analysis of the weld bead height difference curve. Utilizing the surface datum plane determined by the RANSAC algorithm, the geometrical indices for surface roughness, including $R_z$, $R_a$, $S_z$ and $S_a$, are quantitatively evaluated with errors of ≤ 0.1558 mm. Experimental validations demonstrate that the system provides comprehensive multi-source surface information during the WA-DED process, capabilities not achievable with single-function inspection technologies, thus enabling sufficient data for effective part quality improvement.