Abstract: Aluminum alloy is widely used in industrial manufacturing by virtue of its excellent characteristics,for which a lot of waste aluminum alloy is produced.In addition,the new production of aluminum alloy would consume a lot of energy and cause environmental pollution.Therefore,recycling waste aluminum alloy play a very important role,which requires to find an efficient classification method to recycle aluminum alloy.Based on laser-induced breakdown spectroscopy(LIBS),a surface component detection technology,the spectral information of aluminum alloy surface components was collected.Then,combined with ELMAN neural network,artificial neural network(ANN) and random forest(RF) model,the series classification and brand classification of aluminum alloy were carried out for the full spectral data and characteristic spectral line data.The results show that 6 series and 11 brands of aluminum alloy can be classified more quickly and accurately by LIBS combined with RF model.The results provide a reference method for the rapid classification of waste aluminum alloy series and brands.

Key words: Aluminum alloy, Laser-induced breakdown spectroscopy(LIBS), Random forest(RF), Artificial neural network(ANN), ELMAN neural network