A neural network for electromagnetic based ore sorting. Issue 2 (5th March 2018)

• Record Type: Journal Article
• Authors: Li, Min; Caushaj, Arber; Silva, Rodrigo; Lowther, David
• Is Part Of: Compel. Volume 37:Issue 2(2018)
• Page Start: 691
• Page End: 703
• Publication Date: 2018-03-05
• Subjects: Sensors -- Neural networks -- Eddy currents -- Nondestructive testing -- Pattern recognition -- Electromagnetic sensor; Electrical engineering -- Data Processing -- Periodicals; Electrical engineering -- Mathematics -- Periodicals; Electrical engineering -- Periodicals; Electronics -- Data Processing -- Periodicals; Electronics -- Mathematics -- Periodicals; 621.3
• DOI: 10.1108/COMPEL-12-2016-0529
• Languages: English
• ISSNs: 0332-1649

Abstract

Abstract : Purpose: This paper aims to presents a novel application of neural network (NN) pattern recognition to ore rock sorting using inductive electromagnetic (EM) sensors. Design/methodology/approach: The impedance of a metallic rock can be measured with an inductive method based on Faraday's law and eddy current theory. A virtual rock model is then created for the simulation of the EM measurements. An NN is trained to differentiate between waste and useful ore samples (containing high amount of minerals) based on the EM sensor signals produced by the rocks. Findings: The NN solution showed high accuracy of rock classification and produced relatively robust results from signals with noise. Originality/value: A pattern recognition NN was applied to classify low- and high-grade ore samples. It has the potential to determine the approximate amount of conductive materials inside ore rocks through multiple classes. This method can be used to improve the performance of EM-based ore sorting for mineral pre-concentration.