A multi-type riveting defect detection and classification method based on the enhancement of force-displacement curve timing features

Abstract

Aiming at the key requirements of riveting quality inspection for aircraft thin-walled structures, this study proposes an intelligent detection method for multi-type defects based on riveting timing feature enhancement. Addressing the issues of low efficiency and high subjectivity in traditional manual detection, the research achieves technical breakthroughs through three stages: First, based on the size measurement of the high-precision driven head, a classification system for riveting defects is constructed, and the visual mapping relationship between geometric parameters and quality standards is established. Second, the force-displacement dynamic characteristics and their change rates of the riveting process are incorporated into the analysis framework. The correlation mechanism between specific defects and mechanical responses is revealed by decoupling the characteristics of the process stages, and the relationship between the process characteristics and quality indicators is identified. Finally, an intelligent recognition model based on BiLSTM is constructed, and its superior performance in multi-type defect classification is validated through comparative experiments. The experimental results show that the four evaluation indexes of this method exceed 98 %, providing a quality monitoring solution for the field of aviation manufacturing.