Highly efficient industrial sorting lines require fast and reliable classification methods. Various types of sensors are used to measure the features of an object to determine which output class it belongs to. One technique involves the use of an RGB camera and a machine learning classifier. The paper is focused on protecting the sorting process against prohibited and dangerous items potentially present in the sorted material that pose a threat to the sorting process or the subsequent metallurgical process. To achieve this, a convolutional neural network classifier was applied under real-life conditions to detect forbidden elements in copper-based metal scrap. A laboratory stand simulating the working conditions in a high-speed scrap sorting line was prepared. Using this custom stand, training and test sets for machine learning were gathered and labeled. An image preprocessing algorithm was designed to increase the robustness of the resulting forbidden element detector system. The performance of multiple neural network architectures and data set augmentations was analyzed. The highest accuracy of 98.03% and F1-score of 97.16% were achieved with a DenseNet-based classifier. The results of this paper show the feasibility of using the presented solution on a high-speed industrial line.
The processing of copper-bearing scrap makes it possible to reduce the costs and energy consumption of obtaining copper alloy products compared to producing them from primary raw materials. To achieve this, it is necessary to quickly and accurately determine the content of alloying elements in individual scrap elements. However, the copper-bearing scrap obtained at secondary raw materials collection points consists of elements with various surface conditions (due to contamination, shape, paint coatings, roughness, etc.). The paper contains research results on the influence of surface roughness and paint coatings on the measurement result of the content of alloying elements in copper-bearing scrap. Three mobile spectrometers were used for measurements: spark-induced optical emission spectroscopy (OES), X-ray fluorescence spectrometry (XRF) and laser-induced breakdown spectroscopy (LIBS). The tests used elements with different surface roughness (from Ra = 0.03 μm to 6.7 μm) and covered with various types of varnish (alkyd, water-based, oil-phthalic, acrylic, oil-alkyd). It was found that the roughness of Ra < 2 μm does not significantly affect the results of the measurements with the OES and LIBS spectrometers, and a larger scatter of measurement results was observed for the XRF spectrometer compared to OES and LIBS. For Ra > 2 μm, a significant impact of roughness was found on the measurement results (this may result in the erroneous classification of the scrap as an incorrect material group). The influence of paint coatings on the measurement is much stronger compared to surface roughness. Even a single layer of paint can cause a change in the measurement result of the content of alloying elements by more than 10%. In the case of an OES spectrometer, paint coatings may prevent the measurement from being performed (which means that no measurement result can be acquired).