RESUMO
Although the risk of some hazardous operations has been greatly reduced, the changes in air conditions caused by them cannot be ignored. A lot of smoke is produced in some hazardous workplaces, which is very easy to cause air pollution. Exposed to this condition for a long time, human health is also greatly endangered. Therefore, the online detection of air conditions in hazardous workplaces is of great significance. In this work, electric welding operation is taken as an example to study the air conditions in hazardous workplaces. The real-time detection of smoke generated from electric welding operation is carried out by Laser induced breakdown spectroscopy (LIBS). The spectral line of lead is found in the smoke spectrum of lead tin wire. Some other metal elements are also detected in the smoke of the tin wire, such as Sn, Cu, Ca and K. Consequently, it is recommended to use lead-free tin wire. In addition, two welding scenarios are simulated. The results show that the generated smoke contains different element information with different welding objects. Then, the quantitative analysis of lead in smoke is carried out. It shows that the concentration of Pb can be calculated by the spectral line intensity in the spectrum. Finally, combined with the error back propagation artificial neural network (BP-ANN), different kinds of smoke are classified, and the recognition accuracy is more than 99%, which proves that the combination of LIBS and BP-ANN is developed into an important method in the monitoring of different air pollutants.
