Condition detection in Al alloy welding process based on deep mining of arc spectrum

Condition detection during the welding process of aluminum alloy is of great significance both for guaranteeing the quality stability of aerospace structure and promoting the robotic intelligent welding manufacturing (IWM). The following research was carried out considering being lack of the effective knowledge mining method of arc spectrum and the unclear correlation between arc spectrum and weld defects. The sensitivity position interval of spectrum probe was experimentally determined to ensure the reliability of the collected spectrum information. By means of principle component analysis of metal spectrum, FeI(407.84 nm), MgI(383.83 nm) and AlI(369.15 nm) were selected, and their correlation to wire feeding state was qualitatively and quantitatively evaluated. Subsequently, based on the feature of principle component coefficients of the metal spectral line, the dynamic response rules of different metal elements were analyzed, and then, the strong correlation was found between FeI spectral line and the wire feeding state. Status detection of wire feeding was performed based on the proposed feature of FeI line spectrum. After repeat verification through different welding tests, the results showed that the method had high stability and strong anti-interference ability.