The Research On Acoustic Signal Processing For Laser Welding Quality Real-Time Detection

Laser welding has been used more and more widely for many specific advantages of itself in recent years. However, the process of laser welding is so complicated that we should monitor it with corresponding real-time monitoring system to evaluate the welding quality timely. When the defects came into being and where the positions were be in(great) request in time. So it is necessary to research and development a kind of monitor means to local the defects real time, so that the time for quality examination after the weld is been greatly reduced.At first, the text researched and designed the acquisition system based on the platform of Labview for collecting the acoustical signals of laser welding, which is able to change the acoustical signal to digital signal. Moreover, the system adopted array technology of microphones to collect the sound wave, accordingly increased the amount of retrievable information. The foundation of such system offer the possibility of further array signal processing.On the basis of the analysis work for the characteristic of welding acousitic field, the soundsource model correspond with the field which is of broadband and near field is builted. As the conventional algorithm about array acoustic signal processing are all builded on the suppose of far field and narrowband, improving the conventional algorithm is needed and indispensable.Because the acoustical signals are filled with interference signals and noisy signals, we can't obtain the effective information that reflect the welding quality directly from receipt signals. The core of this study is to research a arithmetic which can be used to pick up the acoustic component corresponded to the process of laser welding and fix position of serious defects.After the comprehend of principle of array signal processing, this text researched selectively the basic thought of Independent Component Analysis and Multiple signal classification algorithm, and apply them to digital signal process of this study by combining advantages of both. And expected that the algorithm can be successed in signal separation and locating the position of weld defects.The experimental process of this text mainly simulated the case of melt through severely which is realized by cutting a narrow groove on the back side of steel plate before welding. After the acquisition of acoustical signals of whole welding process, the verification work for algorithm is made by using the acquisition data. The work for data processing is made on Matlab. The computing result indicated that Independent Component Analysis extract effectively the acoustical signals about welding process, and on the basis of it, the DOA algorithm is also successed in locating the position of weld defects.