An accurate tool wear prediction method based on machining feature informed data for NC machining of complex aircraft parts

Abstract

Accurate tool wear prediction is indeed a critical aspect of ensuring quality, improving efficiency, and reducing costs in the manufacturing process of aircraft parts. Due to the complex geometry and machining technology, the working conditions change continuously and in a large range during the machining process, which poses substantial challenges to traditional data-driven tool wear prediction methods. This article presents an accurate tool wear prediction methodology that leverages machining feature-informed data for NC machining of aircraft parts. The monitoring data is collected and classified using machining features as carriers. As a result, the coupled influences of cutting conditions and tool wear on the monitoring data could be informed into machining feature information. Independent prediction models of different machining features are trained as fundamental models with machining feature informed data, and finally, the meta-invariant feature space (MIFS) learns the natural laws governing the evolution of these fundamental models, enabling precise tool wear prediction even amidst substantial and continuous variations in working conditions. The experimental results indicate a 35% enhancement in prediction accuracy for the proposed method, in comparison to existing data-driven approaches in the NC machining of a typical complex structure part.