Mutual Separation of Aluminum Alloys by Combining Appearance Recognition with Neural Network Analysis (ARENNA) and Laser-induced Breakdown Spectroscopy (LIBS) Sorters

Therefore, aluminum circulation is unstable and some recycled aluminum has to be exported. Harita Metal Co., Ltd., one of the most sophisticated metal recycling companies in Japan, is improving high-grade horizontal material recycling. It is involved in a national project in collaboration with universities and national research institutes to achieve mutual separation of scrap aluminum alloys by combining an appearance recognition system with a neural network analysis (ARENNA) sorter and a laser-induced breakdown spectroscopy (LIBS) sorter. The ARENNA sorter was developed at the National Institute of Advanced Industrial Science and Technology (AIST). ARENNA performs 3D image analysis and density measurement with a neural-network system to identify shredded scrap aluminum alloy particles. The sorter is expected to separate wrought aluminum alloys from casting aluminum alloys. The LIBS sorter was designed originally in collaboration with the German company SECOPTA and several Japanese companies. The LIBS sorter detects the elements Cu, Zn, Fe, Mn, Fe, Si, and Mg in aluminum alloys. Samples of seven kinds of wrought aluminum alloys and one kind of casting alloy were used in the separation tests with (1) only the ARENNA sorter, (2) only the LIBS sorter, and (3) both sorters in series. In experiment (1), all the wrought alloys were separated with a separation probability of over 90%, although the separation probability was about 86% for the casting alloy. For experiment (2), all the wrought alloys except the 2000 series were separated with over 90% separation probability; however, the 2000 series alloy was misrecognized as a casting alloy, which should be removed by the ARENNA sorter before feeding to LIBS sorter. In experiment (3), all the wrought alloys except the 3000 series were separated with a separation probability of over 90%, and the separation probability for the 3000 series alloys was about 85%. The reasons for the misrecognition were also clarified. Harita Metal Co., Ltd. intends to develop more precise mutual separation of scrap aluminum alloys in a pilot-scale plant with a capacity of more than 1 t/h in accordance with customer requirements. We also intend to create a precise, energy-saving sensor-based sorting process for all scrap metals and metal alloys, which could replace part of conventional energy-intensive smelting processes.