Abstract
Weld quality directly affects the performance and reliability of structural parts and can be critically ensured by online monitoring. In this study, local mean decomposition (LMD), an adaptive time-frequency analysis is used to analyze the current signals of welding, and then combined with deep belief network (DBN) to classify the weld quality. Firstly, the current signals are decomposed by LMD into a series of product functions. Each product function is a complex signal, and its complexity is calculated by multi-scale entropy to select the most relevant product function with weld quality. Finally, DBN is applied to classify weld quality into four types. This method has a higher classification and recognition rate compared with principal component analysis and extreme learning machine classification. Thus, LMD is a potentially effective method to diagnose weld quality.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call