Applying Receiver-Operating-Characteristic (ROC) to bulk ore sorting using XRF. (15th January 2020)

• Record Type: Journal Article
• Authors: Li, Genzhuang; Klein, Bern; Sun, Chunbao; Kou, Jue
• Is Part Of: Minerals engineering. Volume 146(2020)
• Publication Date: 2020-01-15
• Subjects: Bulk ore sorting -- Classification -- Simple linear regression -- Receiver Operating Characteristic; Mines and mineral resources -- Periodicals; Ressources minérales -- Périodiques; Mines and mineral resources; Periodicals; Electronic journals; 622
• DOI: 10.1016/j.mineng.2019.106117
• Languages: English
• ISSNs: 0892-6875

Abstract

Highlights: Receiver Operating Characteristic (ROC) was applied in ore classification for bulk ore sorting. ROC approach was found to result in more accurate classification compared to SLR. Economics performance was enhanced by ROC approach compared to that of SLR. Abstract: As high grade deposits become depleted, the mining industry is challenged by mining lower grades that require the use of non-selective bulk mining methods. Bulk ore sorting can reduce the dilution of the mined ore thereby preventing costly processing of low grade waste while improving the quality of material reporting to the processing plant. X-Ray Fluorescence (XRF) is one of the sensing technologies used for bulk ore sorting. The grade of the bulk material is predicted based on XRF measurement that is correlated to a grade using a calibrated algorithm. The effectiveness of XRF bulk sorting depends on the accuracy of the grade classification of the scanned material. False classification is attributed to poor performance of the sorting algorithm to correlate the measurements with the assay grades. The paper presents an approach to analyze XRF measurements that reduce misclassification of material. Receiver Operating Characteristic (ROC) was applied as the tool in classification of bulk materials. The performance of this approach was compared to conventional sorting algorithm, namely, simple linear regression (SLR), where XRF responses was linearly correlated to the assay grades. The ROC approach was found to result in more accurate classification and enhanced economics performance of the bulk ore sorting system.