Research On Arc Sound Signal Of Aluminum Alloy MIG Welding

Welding process is a complex physical and chemical process containing arc sound, arc light, and electrical signal, etc. Each kind of signals provides important information for welding process control and quality control. As a concomitant of welding process, arc sound signal contains rich information about arc and has a close relevance with arc behavior, metal transfer, stability of arc, welded joint quality and so on. It is an important information resource for studying the stability of welding process and monitoring and controlling welding quality. Arc sound in aluminum alloy MIG welding process was studied on the relativity between arc sound signal and metal transfer in this thesis. Arc sound patterns of different metal transfer were established by analyzing arc sound in different metal transfer with ARMA two-spectra, and then were used to do pattern recognition though SVM. The relativity between arc sound signal and the subsidence of aluminum alloy MIG welding bead was also studied in the thesis. The main works were done as follows:The signal gathering experiment system was developed to gather arc sound, arc light and electrical signal synchronously and MATLAB software was used to process data in the thesis.Through analyzing the signal data gathered by the signal synchronous gathering system, the relativity of arc sound and metal transfer that arc signal frequency is the same as metal transfer frequency is found. After analyzing arc sound by different power spectrum estimation methods, AR parameterization estimation power spectrum estimation method is better than others and can gain metal transfer frequency fast. Then arc sound in different metal transfer modes were analyzed by AR parameterization estimation power spectrum estimation method.The double-spectrum information of arc sound in different metal transfer mode was obtained by ARMA double-spectrum estimation method and the feature values could be detected from these double-spectrum information. SVM pattern recognition with the feature value can classify arc sound in different metal transfer mode and realize fast recognizing metal transfer mode by arc sound.The studies on the relativity between arc sound and the subsidence of welding bead show that the arc sound signal energy has some relation with welding bead. The total energy of arc sound signal increases along with the subsidence of welding bead and change largely before the subsidence and after the subsidence. Arc sound signal was processed by wavelet packet and the energy distribution rule of arc sound signal indifferent frequency range was obtained. The energy distribution of arc sound signal shows that the subsidence of welding bead can be detected accurately from the change of arc sound energy within frequency range from 2.75kHz to 27.5kHz.