AK-ILS: An Active learning method based on Kriging for the Inspection of Large Surfaces

Tolerance verification permits to check the product conformity and to verify assumptions made by the designer. For conformity assessment, the uncertainty associated with the values of the measurands must be known. In fact, to evaluate form characteristics of large aircraft structure workpieces, sampling is required, so a measurement error is present: exact estimation of form characteristics would require complete knowledge of the surface. To minimise this measurement error, this paper presents a Kriging-based procedure to identify the minimum of measured points to check the conformity with a given confidence level. The proposed method is validated on a simple example of orientation tolerance and then performed to inspect the form defect on three large aircraft workpieces.

► We model a surface using the Kriging estimation technique. ► A design of experiments is improved iteratively. ► The chosen point is the most probable point outside the tolerance zone. ► Academic validation with the orientation inspection of a flat surface. ► Three test cases based on aeronautic structures.