Identification Method of Iron-Ore Type in Open Pit Based on Measured Spectra and Random Forest

ABSTRACT:Iron ore is the main mineral resources to support China's national economy, and open-pit mining is the main way. Chemical analysis has its disadvantages in terms of the determination of iron-ore type, such as slow speed and high cost. Therefore this study has put forward a method to determine iron-ore type through field spectrum analysis. First various types of iron-ore samples were collected in Anqian mine of Liaoning Province, China, including 48 magnetite and 84 hematite samples. Then, through the hole drilling and cutting, the circular sheets of 1~2 cm thick are formed. After that the chemical composition, visible/near-infrared spectra of these samples were tested and analyzed. The principal component analysis is used to reduce the data dimension of spectrum, and the random forest algorithm is applied to classify the iron ore type of samples, while the accuracy of classification results is validated. The results show that the random forest method can classify different types of iron ore, and the successful rate is over 95%, which lay the foundation for the determination of the ironore type in open pit based on the portable spectroradiometer or the airborne imaging spectrometer.