A review of intelligent ore sorting technology and equipment development

Xianping Luo, Kunzhong He, Yan Zhang, Pengyu He, and Yongbing Zhang

Cite this article as:

Xianping Luo, Kunzhong He, Yan Zhang, Pengyu He, and Yongbing Zhang, A review of intelligent ore sorting technology and
equipment development, Int. J. Miner. Metall. Mater., 29(2022), No. 9, pp. 1647-1655. https://doi.org/10.1007/s12613-022-2477-
5

View the article online at SpringerLink or IJMMM Webpage.

Articles you may be interested in

Hai-nan He, Xiao-chen Wang, Gong-zhuang Peng, Dong Xu, Yang Liu, Min Jiang, Ze-dong Wu, Da Zhang, and He Yan,
Intelligent logistics system of steel bar warehouse based on ubiquitous information, Int. J. Miner. Metall. Mater., 28(2021), No. 8,
pp. 1367-1377. https://doi.org/10.1007/s12613-021-2325-z

Lu Lin and Jia-qing Zeng, Consideration of green intelligent steel processes and narrow window stability control technology on
steel quality, Int. J. Miner. Metall. Mater., 28(2021), No. 8, pp. 1264-1273. https://doi.org/10.1007/s12613-020-2246-2

Xue-qiu He, Chao Zhou, Da-zhao Song, Zhen-lei Li, An-ye Cao, Shen-quan He, and Majid Khan, Mechanism and monitoring
and early warning technology for rockburst in coal mines, Int. J. Miner. Metall. Mater., 28(2021), No. 7, pp. 1097-1111.
https://doi.org/10.1007/s12613-021-2267-5

Rong Zhu, Bao-chen Han, Kai Dong, and Guang-sheng Wei, A review of carbon dioxide disposal technology in the converter
steelmaking process, Int. J. Miner. Metall. Mater., 27(2020), No. 11, pp. 1421-1429. https://doi.org/10.1007/s12613-020-2065-5

Wan-zhong Yin and Yuan Tang, Interactive effect of minerals on complex ore flotation: A brief review, Int. J. Miner. Metall.
Mater., 27(2020), No. 5, pp. 571-583. https://doi.org/10.1007/s12613-020-1999-y

Min Lin, Zhen-yu Pei, Yuan-yuan Liu, Zhang-jie Xia, Kang Xiong, Shao-min Lei, and En-wen Wang, High-efficiency trace Na
extraction from crystal quartz ore used for fused silica-A pretreatment technology, Int. J. Miner. Metall. Mater., 24(2017), No. 10,
pp. 1075-1086. https://doi.org/10.1007/s12613-017-1498-y

IJMMM WeChat QQ author group

International Journal of Minerals, Metallurgy and Materials
Volume 29, Number 9, September 2022, Page 1647
https://doi.org/10.1007/s12613-022-2477-5
Invited Review
A review of intelligent ore sorting technology and equipment development

Xianping Luo1,2,3), ✉, Kunzhong He3), Yan Zhang3), Pengyu He2), and Yongbing Zhang3)
1) Jiangxi Key Laboratory of High-efficiency Development and Application of Tungsten Resources, Jiangxi University of Science and Technology, Ganzhou
341000, China
2) Jiangxi Key Laboratory of Mining & Metallurgy Environmental Pollution Control, Jiangxi University of Science and Technology, Ganzhou 341000, China
3) Faculty of Resource and Environmental Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, China
(Received: 16 December 2021; revised: 21 March 2022; accepted: 21 March 2022)

Abstract: Under the background of increasingly scarce ore worldwide and increasingly fierce market competition, developing the mining in-
dustry could be strongly restricted. Intelligent ore sorting equipment not only improves ore use and enhances the economic benefits of enter-
prises but also increases the ore grade and lessens the grinding cost and tailings production. However, long-term research on intelligent ore
sorting equipment found that the factors affecting sorting efficiency mainly include ore information identification technology, equipment sort-
ing actuator, and information processing algorithm. The high precision, strong anti-interference capability, and high speed of these factors
guarantee the separation efficiency of intelligent ore sorting equipment. Color ore sorter, X-ray ore transmission sorter, dual-energy X-ray
transmission ore sorter, X-ray fluorescence ore sorter, and near-infrared ore sorter have been successfully developed in accordance with the dif-
ferent characteristics of minerals while ensuring the accuracy of equipment sorting and improving the equipment sorting efficiency. With the
continuous improvement of mine automation level, the application of online element rapid analysis technology with high speed, high precision,
and strong anti-interference capability in intelligent ore sorting equipment will become an inevitable trend of equipment development in the fu-
ture. Laser-induced breakdown spectroscopy, transient γ neutron activation analysis, online Fourier transform infrared spectroscopy, and nucle-
ar magnetic resonance techniques will promote the development of ore sorting equipment. In addition, the improvement and joint application of
additional high-speed and high-precision operation algorithms (such as peak area, principal component analysis, artificial neural network, par-
tial least squares, and Monte Carlo library least squares methods) are an essential part of the development of intelligent ore sorting equipment in
the future.
Keywords: intelligent ore sorting technology; sorting equipment; separation efficiency; online element rapid analysis technology

1. Introduction intelligent ore sorting equipment can improve resource utiliz-

ation and the grade of separated ore. Such equipment can re-
After 40 years of continuous high-intensity mining, the re- duce the grinding amount of ore and minimize the produc-
serves of high-grade and easy-to-separate ores have gradu- tion of fine tailings. This equipment can also improve the
ally decreased. Challenges in the separation of low-grade and economic benefits of enterprises, enhance the mining envir-
difficult-to-separate ores have also substantially increased onment, and prolong the service life of mining enterprises
[1–2]. The development of the mining industry could be [8]. Fig. 1 shows the beneficiation process using an ore sorter.
strongly restricted by several problems, such as the occur- The figure shows that the raw ore was crushed to the qual-
rence of a large number of refractory lean ores, increasingly ified particle size by the jaw crusher and fell into the buffer
fierce market competition, and serious environmental pollu- ore bin. The ore was then fed to the ore feeder by the convey-
tion in mines [3–4]. Traditional beneficiation methods solve or belt through the bucket feeder. The ore information ac-
the problem of ore recovery and use but can also cause the quisition system scanned the ore characteristic information
consumption of considerable amounts of water and increase (such as color, magnetism, and conductivity). The informa-
agent pollution of the environment [5]. With the progress and tion processing system then fed back the processed signal to
development of science and technology, mechanized and the computer, which sent the separation instruction to the
automatic production in nonferrous industry development separation execution system. Finally, the target ores fell into
has become an inevitable trend [6]. Considering the rising the designated ore bin after processing by the separation exe-
production costs of the mining industry, developing high-ef- cution system, thus realizing the purpose of separation.
ficiency intelligent ore sorting technology and equipment has The development history of intelligent ore sorting equip-
become the research focus of scholars and engineers at home ment can be traced back to 1947. Gunson Company manu-
and abroad [7]. Long-term production practice showed that factured the first color sorting equipment to solve the prob-

✉ Corresponding author: Xianping Luo E-mail: [email protected]

© University of Science and Technology Beijing 2022
1648 Int. J. Miner. Metall. Mater., Vol. 29 , No. 9 , Sep. 2022

1
the electromagnetic plate was used for separation. However,
the first generation of sorting equipment had low efficiency
and poor accuracy, which could not meet the needs of mine
5 2 production. Therefore, researchers successively developed
3 the second and third generations, namely the magneto-optic-
2
al and flat belt optical sorting machines, respectively [5].
7 6 4 Compared with the first generation, the sorting accuracy and
efficiency of the second and third generations have been im-
9 proved to a certain extent. However, sorting equipment with
8 high separation efficiency has become the main research dir-
10 11 ection due to the increasing demand for mineral products
1—As-mined ores; 2—Conveyor belt; 3—Jaw crusher; 4—Buffer with industrial development.
orebin; 5—Scoop feeder; 6—Ore information detection device;
7—Computer; 8—Information processing system; 9—Separation 2. Influencing factors of sorting efficiency
execution system; 10—Concentrate bin; 11—Tailings bin

Fig. 1. Beneficiation process using an ore sorter. Intelligent ore sorting equipment uses the different phys-
ical and chemical properties of the ore, such as optical prop-
lem of labor shortages in seed processing [9]. In the early erties, electrical conductivity, magnetic properties, radio-
1960s, the UK conducted research on this equipment and activity, and the difference in reflection and absorption char-
transformed it for the separation of gypsum and marl, thus acteristics of various rays (e.g., γ-ray, β-ray, X-ray, infrared,
becoming the first application example of ore sorting equip- and radio waves) in different ores [11–13]. Fig. 2 shows the
ment in the mining industry [10]. Under the influence of the classification and advantages of ore sorters derived from the
research upsurge in foreign ore sorting equipment, research- physical and chemical properties of ore, which lead to the
ers developed China’s first-generation photoelectric ore sort- production of a large amount of ore sorting equipment.
er and high-frequency ore sorter in the 1960s [5]. These However, the efficient operation of sorting equipment is one
pieces of equipment use a traditional disk feeder for ore feed- of the main factors that determine the production cost of
ing. The detection signal could be controlled separately, and mines. Long-term production experience revealed that the

Daylight tube color sorter

Ore process mineralogy Optical sorter
LED light source color

Microwave light source color

Optical sorter Magnetic susceptibility Electrostatic field electric sorter

Radioactivity Electrostatic sorter
Electrical conductivity Corona electrostatic sorter

Reflection or absorption characteris- Tribo-electrostatic sorter

tics under different radiation

Magnetic sorter High intensity magnetic sorter

Low intensity magnetic

Radiometric sorter High gradient magnetic sorter

Ray ore sorter
Uranium radioactive sorter

X-ray based Thorium radioactive sorter

γ-ray ore sorter

Near-infrared ore sorter XRF-based sorter Advantages of ore sorter

XRT-based sorter (1) Improve resource utilization;

Double energy XRT sorter
(2) Upgrade grinding feed;

(3) Reduce fine tailings;

(4) Increase economic benefits;

(5) Improve the mining environment;

(6) Prolong the service life of mining enterprises.

Fig. 2. Sorting technologies and potential enabling advantages. XRT—X-ray transmission sorter, XRF—X-ray fluorescence sorter.
X.P. Luo et al., A review of intelligent ore sorting technology and equipment development 1649

factors affecting the separation efficiency of intelligent ore ment [15].

sorting equipment mainly include ore information identifica-
2.2. Equipment sorting actuator
tion technology, equipment sorting actuator, and information
processing algorithm. The equipment sorting actuator is the endpoint of intelli-
gent ore sorting. Herein, the target minerals are separated
2.1. Ore information identification technology
from gangue minerals. Fig. 3 shows that the main sorting ac-
Ore information identification technology is the core of in- tuators include the injection, percussion, and mechanical arm
telligent ore sorting equipment. This technology collects the sorting methods. (1) The injection method is the most widely
characteristic information of ore and serves as the basis for used sorting in intelligent ore sorting equipment [16]. This
ore sorting. The characteristic information of ore includes its method mainly comprises an air supply, high-speed solenoid
optical properties, electrical conductivity, magnetic proper- valve, and high-pressure nozzle (Fig. 3(a)). The air supply
ties, radioactivity, and the difference in reflection and ab- comprises an air compressor, air storage tank, and air purific-
sorption characteristics under various high-intensity rays. Ore ation. Moreover, the high-speed solenoid valve and high-
information identification technology comprised irradiation pressure nozzle comprise the injection actuator. The control
and detection systems. The irradiation system provides a light nozzle air opens one-to-one when the high-speed solenoid
or ray source for ore information detection. Meanwhile, the valve with a reaction frequency of thousands of times each
detection system mainly aims to use sensors to convert detec- second receives the driving signal of the information pro-
ted ore information, such as reflection and absorption of light cessing to eject high-pressure air through the nozzle, thus in-
sources or light rays, into electrical signals and send them to ducing the deviation of the ore from the original motion track
the computer. Therefore, the gain time, sensitivity, resolu- and realizing separation. (2) The percussion methods mainly
tion, and anti-interference strength of the sensor are the main comprise an electromagnetic coil, pushrod, striking plate,
factors of ore information identification technology that af- frame, and return spring (Fig. 3(b)) [15]. The actuator starts
fect the efficiency of intelligent ore sorting equipment [14]. the electromagnetic coil after receiving the processing signal
The length of gain time determines the efficiency for ore in- to charge and create a magnetic field to adsorb the pushrod.
formation collection by ore sorter. The sensitivity and resolu- This rod then drives the beating action to complete the beat-
tion of the sensor determine the sorting accuracy of the ing separation act of the ore. Finally, the beating plate could
equipment. Moreover, the intensity of anti-interference de- be controlled and reset by the return spring after completion
termines the wide application of intelligent ore sorting equip- of the beating. This method is mainly used in separated large

1 1
5
4
3 4 3
2 6 7 2 5 7
6

(a) Injection method of ore sorter. 1—Computer; 2—Ore (b) Percussion method of ore sorter. 1—Computer; 2—Ore
information collection system; 3—Traveling belt; 4—Air- information collection system; 3—Traveling belt;
pump; 5—Gas ejector pipe; 6—Tailings bin; 7—Concen- 4—High-frequency electromagnetic coil; 5—Hit the board;
trate bin. 6—Tailings bin; 7—Concentrate bin.

1
3

4
2

6
5
(c) Mechanical arm method of ore sorter. 1—Computer; 2—Ore
information collection system; 3—Mechanical arm; 4—Traveling belt;
5—Concentrate bin; 6—Tailings bin.
Fig. 3. Ore sorter sorting actuator: (a) injection method of ore sorter, (b) percussion method of ore sorter, and (c) mechanical arm
method of ore sorter.
1650 Int. J. Miner. Metall. Mater., Vol. 29 , No. 9 , Sep. 2022

particle size ore blocks. (3) The main parts of the mechanical tin mine of Guangxi Guihuacheng Co., Ltd., the German Mo-
arm sorting method include the sorting manipulator and mo- gensen AP 1200 color sorter is used to separate wolframite
bile (Fig. 3(c)). The identified and positioned ore enters the and cassiterite with 14–31 mm particle size. A separation rate
multiarm sorting region, and multiple sorting manipulators of 98.9% could be obtained when sorting black materials and
select the target ore stably and quickly. This method is up to 95.2% when sorting white materials. The grade of the
mainly used to sort coal gangue or large ore due to its slow corresponding qualified ore is expected to be increased by
sorting efficiency [17]. The applicable environment of sort- approximately 16%, and the rejection rate will be increased
ing actuators of various equipment was different. The devel- by approximately 14% compared with the original [22]. A
opment of a responsive, energy-saving, and widely applic- talc mine in Guangxi uses the British Sortex-621 photoelec-
able equipment sorting actuator is one of the important tric sorter to sort talc and gangue. A +50 mm particle size
factors to promote the development of intelligent ore sorting fraction could be recovered, and the talc concentrate grade
equipment. obtained by one-time sorting was 91.47%. Unexpectedly, a
67.30% concentrate yield could be obtained [23]. Thus, the
2.3. Information processing algorithm
application of color sorters can markedly improve the effi-
An information processing algorithm controls intelligent ciency of ore sorting. However, these sorters can only separ-
ore sorting equipment, and its main purpose is analysis and ate ores with large color differences, which limits the applic-
decision making. Therefore, the calculation accuracy and ation of color sorters.
running speed of the information processing algorithm XRT applied in mining has brought a qualitative leap to
largely determine the sorting efficiency of intelligent ore intelligent ore sorting equipment, not only improving the
sorting equipment. High-speed processing on the obtained sorting efficiency but also ensuring the sorting accuracy of
characteristic information performs basic noise processing on the equipment [24]. Fig. 4 shows the working principle of the
graphics or data [18]. A series of algorithms is used to ana- XRT intelligent sorter.
lyze the processed images or data, and the analysis results are
compared with the predetermined threshold to recognize
1
gangue and the ore. A driving signal will decide the separa-
tion action. The algorithm will then drive the sorting actuator
after delay and power enlargement to separate the arm ores or
gangues accurately. For example, the artificial neural net- 2 4
work method is a mathematical model similar to the synaptic 3
connection structure of the brain for information processing
[19–20]. This method is a nonlinear and adaptive informa- 1—X-ray 1aser occurs; 2—X-ray receiver;
tion processing system comprising a large number of pro- 3—Air pump; 4—Gas ejector pipe.
cessing units based on the research results of modern neuros- Fig. 4. Schematic of XRT intelligent sorter.
cience. The artificial neural network method is capable of
self-study, associative storage, and finding optimal solutions At the beginning of the 21st century, a special meeting on
at high speed. ore sorting technology and equipment was held in Russia and
Germany. Many discussions on X-ray radiation technology
3. Development and application of intelligent and equipment took place, attracting the attention of the X-
ore sorting equipment ray sorter equipment industry [25]. The main equipment sup-
pliers include Redwave in Austria, Mogensen in Germany,
Since the first sorter came out in 1947, a variety of intelli- Rados in Russia, and Tomra in Norway. Foreign countries
gent ore sorters have been continuously developed with the have also invested considerably into the research and devel-
increasing demand for ore products and the advancement of opment of X-ray sensor waste disposal technology, and the
science and technology. The mechanical and working prin- equipment application is relatively mature. For example, a
ciples of most intelligent ore sorting equipment are the same. 66.67% concentrate index with yield, P2O5 recovery of
Therefore, the main difference in equipment classification 85.49%, and SiO2 stripping rate of 60.20% could be obtained
lies in ore information identification technology. Many dif- with the introduction of a COM Tertiary XRT B2400 photo-
ferent sorters based on information recognition technology, electric sorter in a phosphate mine in Saudi Arabia [26]. An
such as color sorter, X-ray transmission sorter (XRT), dual- XRT intelligent sorter is used in the crushing section of a
energy X-ray transmission sorter (DE-XRT), X-ray fluores- scheelite concentrator in Hunan to pre-discard raw ore with a
cence sorter (XRF), and near-infrared sorter (NIR), have particle size of 15–45 mm. The grade of discarded waste rock
been developed. WO3 was less than 0.05%, while the feed grade of WO3 was
The main application of the color sorter is to separate the approximately 0.3%. The grade of discarded concentrate
ores, which have evident color differences under light source WO3 was approximately 0.80%, the rejection rate could
irradiation in arm ore and gangue [21]. Research on color reach 68%, and the recovery rate of discarded concentrate
sorters in foreign countries is early, while ore sorting techno- could reach more than 88% [27]. However, domestic R&D
logy is relatively mature. For example, in the coral tungsten investment in X-ray sorters is limited despite the flourishing
X.P. Luo et al., A review of intelligent ore sorting technology and equipment development 1651

foreign XRT technology. Ganzhou Good Friend Technology rays irradiate ores, and the outer electrons will transit to the
Co., Ltd., which manufactures XRT series ray intelligent inner shell. The excess energy in electron transition is re-
sorters, is a leading domestic technology and is widely used leased in the form of characteristic atomic X-rays. The de-
in many mines in China. For example, the Hunan Zhazixi tector of the X-ray sorter analyzes and identifies the ore and
Antimony Mine has started using the XRT-1200 X-ray intel- gangue after receiving the characteristic X-rays released by
ligent sorter of this company. The antimony grade of waste various elements and separates them through the actuator
rock is lower than 0.04%, and the recovery rate of fine anti- [36–37]. In Rados, Russia, XRF has been used in ore sorting
mony in the pre-separation block is more than 98.5% [28]. and achieved certain results; however, XRF has not been ap-
The Taoxikeng mine, which is part of the Ganzhou Zhangy- plied on a large scale due to the unstable production index
uan tungsten industry, has used the HPY-XRT-1400 intelli- [38]. Shenyang Longji also developed an XRF surface radi-
gent sorter and AWS-100-c2s1 single-sided image sorter. ation sorter. The SRF-3-150 sorter produced by the Russian
Good indicators of waste rock WO3 grade could be obtained company Krasnoyarsk was used by Northeastern University
when the raw ore WO3 grade is 0.31% and the selected to separate the low-grade molybdenum ore of Chaoyang
particle size is −50+10 mm through preselection and tailings. Tianma Group. The results show that the tailings rate of raw
The concentrate WO3 grade of 0.95%, concentrate WO3 re- molybdenum ore is 33.2%, the enrichment ratio is 1.44, the
covery rate of 96.20%, enrichment ratio of 3.1, and operation recovery rate of molybdenum concentrate is 95.87%, and the
rejection rate of 68.70% could be obtained [29]. XRT intelli- grade of molybdenum concentrate is 0.092% [39]. XRF tech-
gent ore sorting equipment is one of the most successful be- nology developed the ore separation characteristic informa-
neficiation equipment in the market. However, the applica- tion from the surface color and the reflection or transmission
tion of XRT intelligent ore sorting equipment presents the property of rays to the type and content of elements con-
current situation of low intensification and limited applica- tained in the ore. Such technology considerably improved the
tion range of automation systems and equipment due to the accuracy of the sorting equipment.
complex nature of the ore and the remarkable difficulty in NIR ore sorter uses the NIR vibration spectrum to identify
separation. ore characteristics and separate gangue from the ore; it per-
DE-XRT is based on XRT technology. The technology forms better than conventional surface detection methods
adopts the ore DE-XRT intelligent identification and collects [40]. NIR separation technology was first applied in mining
the high and low-energy transmission signals of ore to X-ray. in 2008; thus, separating kimberlite and gangue minerals is
This technology uses a control box for real-time data opera- feasible [41]. The purification results of kimberlite by NIR
tion and processing and outputs the different information of technology in Cullinan mine, South Africa, by the Canadian
target elements of massive ores to provide an effective basis company Renald revealed that the coarse particle separation
for ore separation [30–31]. For example, a Ksslxt-100 intelli- of NIR active minerals can form identifiable and usable char-
gent sensor sorter was introduced into a low-grade lead–zinc acteristics through direct and positive detection [42]. RWTH
mine in Guangxi to discard 15–60 mm particle size lead–zinc Aachen University implemented NIR separation after wash-
ore samples in advance. The concentrate yield was 52.73%, ing and cleaning the surface of borate mines with different
and the grades of Pb and Zn were 1.06% and 3.97%, respect- particle sizes. The grade of borate was increased from
ively. Meanwhile, the recoveries of Pb and Zn were 86.00% 38.00% to 45.45% and the yield was 85.40%, effectively im-
and 86.15%, respectively. Calculated from the raw ore, the proving the follow-up flotation effect [43]. The application of
comprehensive rejection rate is 34.05%, and the loss rates of the NIR ore sorter shows that the NIR spectrum with the ad-
Pb and Zn are 2.41% and 1.61%, respectively. The test in- vantage of vibration spectrum has a surface activation thick-
dexes were also good [32]. Yan Tong of the University of ness of 30 um, which can withstand a thin dust layer to a cer-
British Columbia, Canada, used pseudo DE-XRT techno- tain extent. When it comes to detection, NIR detection and X-
logy to separate Mississippi Valley-type lead–zinc deposits ray detection technology have certain advantages over color
in the Pend Oreille mine. The lead grade of lead concentrate detection [44].
is 9.92% and the lead recovery is 99% based on the lead and Manufacturers have also made some developments in the
zinc grades of 3.65% and 9.52%, respectively. The zinc grade algorithm of the information processing mechanism of intel-
of zinc concentrate is 13.56%, and the zinc recovery is ligent ore sorting equipment to improve the sorting effi-
96.1%. The lead grade of tailings is 0.17%, and the zinc ciency of the equipment. Huang et al. [45] developed a re-
grade is 0.88%. The overall rejection rate is 38.1% [33]. DE- gional block scrap metal identification algorithm based on
XRT technology is unaffected by factors such as material the combination of low-energy transmission signal value and
moisture change, surface pollution, material shape, or size. dual-energy R-value, improving the applicable thickness
This technology can improve detection accuracy and sorting range of the identification algorithm. In enterprises, the OSX
efficiency and is suitable for sorting wide particle size mater- photoelectric sorter of the Norwegian company Comex
ials [34]. works with a computer and a field-programmable gate array
An XRF sorter separates gangue from the ore by analyz- (FPGA) programmable device. The computer completes a
ing the types and contents of elements in ores [35]. The inner series of work, such as detection result reception, image
electrons of atoms receive energy and produce holes when X- formation and change-over, noise filtering, image algorithm
1652 Int. J. Miner. Metall. Mater., Vol. 29 , No. 9 , Sep. 2022

analysis, sorting threshold definition, calculation, or setting An optical system is used to collect the luminous signal, di-
of the delay time of separation execution action. The FPGA vide the light by the spectrometer, and record the spectrum
mainly stores the driver of the sorting system, which drives using the detector. The composition and content of the ele-
the separation execution system rather than the computer. ments in the sample are qualitatively and quantitatively ana-
The computer could have additional memory space to com- lyzed through spectral analysis. LIBS can also improve de-
plete image processing to realize the efficiency of the entire tection accuracy and is suitable for measuring gas, liquid, and
signal processing system. The Titech sorter adopts the Mi- solid substances. To date, the detection limit of LIBS for
crosoft pack sorting control system, which integrates the con- most elements has reached the order of PPM, while that for
trol and ingredients needed for sorting. This system is used to some elements has even reached the order of PPB [50]. Va-
analyze ore images, store material characteristics and posi- dillo et al. [51] designed a long-distance LIBS detection
tions, adjust the action time of injection systems, and test and device to realize online monitoring of Cr and Ni concentra-
adjust high-speed solenoid valves [46]. The progress of other tions in molten steel. Mohamad et al. [52] applied LIBS tech-
types of ore separation equipment is relatively slow due to nology to the analysis of target elements of aluminum alloy,
technology and market application prospects and few reports obtained temperature through a Boltzmann diagram, and also
on relevant progress in recent years. Therefore, developing acquired the calibration curves of magnesium, manganese,
information processing mechanism algorithms with sensitive copper, and silicon. Peng et al. [53] analyzed the heavy met-
responses, fast information processing speed, and strong al chromium in fresh orange leaves under laboratory condi-
compatibility will promote the further enhancement of intel- tions to verify the feasibility of LIBS in environmental pollu-
ligent ore sorting equipment. tion; they confirmed that LIBS can quickly detect the heavy
metal elements in fresh plant leaves, which is an effective de-
4. Prospect of intelligent ore sorting equipment tection technology. The research of LIBS demonstrates its
real-time, online, fast, remote, and noncontact advantages,
The application of intelligent ore sorting equipment has achieving multielement simultaneous in-situ measurement
ushered in a new situation in the development of the mining and analysis without pre-sample preparation.
industry. The research and development of intelligent ore PGNAA uses the neutron stream generated by a neutron
sorting equipment with fast processing capacity and accurate source to bombard the nuclei of various elements in the
sorting accuracy will promote improvement in the mining sample. The element produces a neutron capture reaction and
field. The ore characteristic information relied on by tradi- is de-excited in 10−14 s–10−12 s, releasing γ rays with an en-
tional sorting equipment has certain detection errors, such as ergy of 2 keV–10 MeV. An analysis system according to
overlapping detection spectra, fuzzy equipment identifica- each characteristic γ online qualitative and quantitative ana-
tion, similar physical properties of ores and gangues, and dif- lysis of elements is conducted in accordance with the energy
ficulties in separation, thereby reducing separation accuracy and intensity of rays [54]. PGNAA technology is fast, accur-
and affecting separation efficiency of equipment [47]. XRF ate, and comprehensive, simultaneously realizing multiele-
separation equipment has been used to separate target and ment analysis. This technology is highly suitable for the on-
gangue minerals by analyzing the types and content of ele- line detection of main ore elements and can solve the prob-
ments contained in the ore. Unfortunately, XRF is an off-line lems of low accuracy and low separation efficiency of cur-
analysis method to control lag, and its running speed is rent sorting equipment.
markedly affected by the environment. Equipment X-ray Online FTIR is a spectral analysis method based on the se-
leakage would also be harmful to the human body [48]. On- lective absorption of elements to infrared radiation. The
line element rapid analysis technology has the advantages of transition of vibrational and rotational energy levels occurs
fast, strong online processing capability, high identification after elemental molecules absorb infrared radiation; there-
accuracy, and strong adaptability. Therefore, using online fore, the infrared spectrum is also called the molecular vibra-
element rapid analysis technology, which is more effective tional-rotational spectrum [55]. The scanning speed of FTIR
than XRF, will become an unavoidable trend in developing equipment can reach 200 spectra/s, and the time resolution of
intelligent ore sorting equipment in the future. Tthe online step scanning technology can reach 250 × 10−12 s. Online
element rapid analysis system, which could be used in the in- FTIR does not require sample preparation and is fast and
dustry at present, mainly includes laser-induced breakdown simple; it can measure multiple ingredients online simultan-
spectroscopy (LIBS), instant emission spectroscopy γ neut- eously [56].
ron activation analysis (PGNAA), online Fourier transform Nuclear MR technology has the advantage of strong pen-
infrared spectroscopy (FTIR), and nuclear magnetic reson- etration; it is also nondestructive, online, fast, and real-time.
ance (MR). This technology is suitable for massive ore sorting. In addi-
LIBS uses a focused, high-energy laser source to release a tion, the MR sensor measurement response time is short, al-
laser with a high power density (larger than 1 GW/cm2) in a lowing simultaneous processing of different grades of ore
remarkably short time (several ns) [49]. High-intensity during the sorting process [57]. The application of nuclear
plasma is formed on the surface of the analytical material MR technology is important for ore sorting, which could sub-
sample, and the atoms in the plasma are excited to emit light. stantially promote the development of intelligent ore sorting
X.P. Luo et al., A review of intelligent ore sorting technology and equipment development 1653

equipment. development prospects. Developing intelligent ore sorting

In addition, the algorithm improvement of information and beneficiation equipment can improve the utilization of
processing of intelligent ore sorting equipment could be pro- resources, enhance the economic benefits of enterprises, in-
moted to develop equipment. Common spectral analysis crease the ore grade, reduce the grinding cost, and minimize
methods include the peak discussion method, principal in- tailings production.
gredient analysis, artificial neural network method, partial (2) Ore information recognition technology is the core of
least squares (PLS), and Monte Carlo spectral library least intelligent ore sorting equipment. The ore information identi-
squares (MCLLS) method. The peak discussion method and fication technology with fast speed, high precision, and
principal ingredient analysis are both quantitative analyses strong anti-interference capability can promote the wide ap-
based on the scale between the net or total discussion of ele- plication of ore sorters. The sorting actuator is the front-end
ment characteristic peaks and concentrations. Additional equipment of intelligent ore sorting, with sensitive responses
spectrum processing techniques are necessary to overcome and strong execution capabilities, ensuring efficient ore sep-
the overlapping peak, absorption improvement effect, and aration. Moreover, the application of high-speed, high-resol-
peak influence between elements [58]. The artificial neural ution, and high compatibility information processing al-
network method directly uses the content and intensity data gorithms will further improve the sorting efficiency of intelli-
of known samples in a large number of training samples to gent ore sorting equipment.
obtain a specific quantitative function. Furthermore, this (3) With the continuous innovation of science and techno-
method does not need to consider the complex relationship logy, many achievements have been realized in the develop-
between content and ray intensity to overcome the overlap- ment of intelligent ore sorting equipment in recent years.
ping peaks and absorption enhancement effects between ele- Color, XRT, DE-XRT, XRF, and NIR ore sorters have been
ments indirectly [59–60]. The neural network method has successfully developed in accordance with the different char-
better results and processing capability for nonlinearity com- acteristics of minerals while ensuring the accuracy of equip-
pared with the commonly used peak area method. However, ment sorting and improving equipment sorting efficiency.
the artificial neural network method requires a large number Device application instances revealed that the sorting recov-
of experimental data as the basis of model training; thus, a ery rate of the aforementioned equipment is generally high,
large number of experimental standard samples is required to while some can also obtain high-quality concentrate.
(4) With the continuous improvement of ore product de-
facilitate measurement in different complex environments,
mand and mine automation level, the application of online
which is a time- and labor-consuming task [61]. PLS is a
element rapid analysis technology with high speed, high pre-
method that has attracted increasing attention in recent years.
cision, and strong anti-interference capability in intelligent
Man et al. [62] measured the main elements of stainless steel
ore sorting equipment will become an inevitable trend of
by PLS. Luo et al. [63] improved PLS and successfully ap-
equipment development in the future. At present, the online
plied it to the analysis of heavy metals and copper mines.
element rapid analysis technology mainly includes LIBS,
Urbanski and Kowalska [64] applied PLS to analyze the
PGNAA, online FTIR, and MR. In addition, the improve-
composition and thickness of iron and zinc coatings in ore
ment and joint application of additional high-speed and high-
powder. MCLLS mainly uses the Monte Carlo method to
precision operation algorithms are an essential part of the de-
simulate and calculate the single element spectrum library
velopment of intelligent ore sorting equipment in the future.
[65]. A large number of research results show that the applic-
The peak area method, principal component analysis method,
ation of PLS can considerably improve the efficiency of in-
artificial neural network method, PLS method, and MCLLS
telligent ore sorting equipment. The experimental spectrum is will further promote the development of intelligent ore sort-
fitted and calculated through multiple cyclic iterations until ing equipment.
the optimal solution is obtained. This method has the advant-
ages of automatic energy peak identification, full-spectrum Acknowledgements
fitting, and automatic correction of the matrix effect. MCLLS
has been successfully applied to the rapid detection of Pb in This work was financially supported by the National Sci-
bone. Overall, the improvement and joint application of addi- ence and Technology Support Program of China (No.
tional high-speed and high-precision operation algorithms are 2012BAC11B07) and the Jiangxi Science and Technology
an essential part of the development of intelligent ore sorting Innovation Base Plan (No. 20212BCD42017).
equipment in the future. The application of a rapid online ele-
ment analysis system will promote the development of intel- Conflict of Interest
ligent ore sorting equipment to a new level.
The authors declare no potential conflict of interest.
5. Conclusions
References
(1) This study responds to the call for green mines at a na-
tional level and realizes the sustainable development of the [1] L. Panda, P.K. Banerjee, S.K. Biswal, R. Venugopal, and N.R.
mining industry. Intelligent ore sorting equipment has broad Mandre, Artificial neural network approach to assess selective
1654 Int. J. Miner. Metall. Mater., Vol. 29 , No. 9 , Sep. 2022

flocculation on hematite and kaolinite, Int. J. Miner. Metall. World Nonferrous Met., 2016, No. 17, p. 31.
Mater., 21(2014), No. 7, p. 637. [23] X.X. Shao and X.P. Ji, Separation of talc by photoelectric sep-
[2] G. Qian, On the development and utilization of low grade min- aration technology, Nonmetallic Min., 1989, No. 5, p. 22.
eral resources, China Met. Bulletin., 2020, No. 1, p. 49. [24] C. Robben, P. Condori, A. Pinto, R. Machaca, and A. Takala,
[3] T. Henckens, Scarce mineral resources: Extraction, consump- X-ray-transmission based ore sorting at the San Rafael tin mine,
tion and limits of sustainability, Resour. Conserv. Recycl., Miner. Eng., 145(2020), art. No. 105870.
169(2021), art. No. 105511. [25] X.W. Yu, M. Yang, S.R. Mao, F.T. Yu and T. He, Research on
[4] J. Lessard, J. de Bakker, and L. McHugh, Development of ore application of X-ray separation technology in beneficiation of
sorting and its impact on mineral processing economics, Miner. phosphate rock, Ind. Miner. Process., 49(2020), No. 9, p. 31.
Eng., 65(2014), p. 88. [26] N. Li, S.H. Zhang, H. Peng, L. Fu, and Y.J. Lu, Application
[5] X.P. Luo, X.H. Ning, T. Wang, P.C. Wang, and P.Y. He, De- practice and evaluation of photoelectric separation in a phos-
velopment and application of intelligent picking technology, phate mining industry, Non Met. Mines, 41(2018), No. 2, p. 73.
Met. Mine, 2019, No. 7, p. 113. [27] N. Jia, Discarding experiment on XRT ray brainpower separat-
[6] B. Sun, J.T. Dai, K.K. Huang, C.H. Yang, and W.H. Gui, Smart or in a scheelite concentrator in Hunan Province, Mod. Min.,
manufacturing of nonferrous metallurgical processes: Review 34(2018), No. 7, p. 154.
and perspectives, Int. J. Miner. Metall. Mater., 29(2022), No. 4, [28] W. Peng and P.Y. He, Pre-selection test and practice of a anti-
p. 611. mony ore using the X-ray intelligent concentrator, Met. Mine,
[7] H. Wotruba and C. Robben, Sensor-based ore sorting in 2020, 2019, No. 9, p. 92.
Automatisierungstechnik, 68(2020), No. 4, p. 231. [29] W.P. Di and Z.H. Wu, Preconcentration and discarding techno-
[8] J. Lessard, W. Sweetser, K. Bartram, J. Figueroa, and L. logy of intelligent photoelectric dressing equipment, Nonfer-
McHugh, Bridging the gap: Understanding the economic im- rous Met. Miner. Process., 2021, No. 1, p. 117.
pact of ore sorting on a mineral processing circuit, Miner. Eng., [30] H. Gao, J.Y. Wang, X.F. Zhang, and D.S. Zhao, Image pro-
91(2016), p. 92. cessing design of mineral identification system based on dual-
[9] F. Zheng, Development of photoelectric sorter abroad, Nonfer- energy X-ray transmission, Nonferrous Met. Miner. Process.,
rous Met. Mineral Process., 1980, No. 2, p. 38. 2021, No. 1, p. 101.
[10] T. Ding, H.P. Xu, J.S. Wang, D.R. Liu and Z.W. Wang, Re- [31] M.G. Xu and X.D. Bao, Object classification suppressed thick-
search of online ore sorter based on vision recognition techno- ness effect based on pseudo dual-energy X-ray transmission
logy, Equip. Manuf. Technol., 2014, No. 7, p. 106. imaging system, Chin. J. Electron Devices, 30(2007), No. 1, p.
[11] Z.H. Wu, Ore pre concentration and discarding waste techno- 219.
logy and selection of intelligent photoelectric beneficiation [32] L.N. Cui and X.Q. Peng, Test study on dual energy X-ray trans-
equipment, World Nonferrous Met., 2020, No. 16, p. 202. mission pre-concentration of a low-grade lead–zinc mine in
[12] Robben and Wotruba, Sensor-based ore sorting technology in Guangxi, Min. Eng., 18(2020), No. 4, p. 30.
mining—Past, present and future, Minerals, 9(2019), No. 9, art. [33] Y. Tong, Technical Amenability Study of Laboratory-scale
No. 523. Sensor-based Ore Sorting On a Mississippi Valley Type
[13] H. Wotruba and H. Harbeck, Sensor-based sorting, [in] Ull- Lead–Zinc Ore [Dissertation], University of British Columbia,
mann’s Encyclopedia of Industrial Chemistry, Wiley-VCH Ver- Vancouver, 2012, p. 41.
lag GmbH & Co. KgaA, Weinheim, 2012. [34] V. Rebuffel and J.M. Dinten, Dual-energy X-ray imaging: Be-
[14] A. Cardenas-Vera, M. Hesse, R. Möckel, R. Gerhard Merker, T. nefits and limits, Insight Non Destr. Test. Cond. Monit.,
Heinig, and Q.V. Phan, Investigation of Sensor-Based sorting 49(2007), No. 10, p. 589.
and selective comminution for pre-concentration of an unusual [35] G.Z. Li, B. Klein, C.B. Sun, and J. Kou, Lab-scale error analys-
parisite-rich REE ore, South Namxe, Vietnam, Miner. Eng., is on X-ray fluorecence sensing for bulk ore sorting, Miner.
177(2022), art. No. 107371. Eng., 164(2021), art. No. 106812.
[15] X.S. Wu, W. Zhang, X.K. Zhao, X. Gao and W.B. Huang, New [36] L.X. Li, G.Z. Li, H.Z. Li, G.Q. Li, D. Zhang, and B. Klein,
development trend of sensor-based ore pre-concentration tech- Bench-scale insight into the amenability of case barren copper
nology, China Min. Mag., 31(2022), No. 6, p. 10. ores towards XRF-based bulk sorting, Miner. Eng., 121(2018),
[16] D.Y. Zhu, D.Y. Liang, P.W. Shi, H.X. Ran, and H.L. Shang, p. 129.
Research and development of ore sorting technology and equip- [37] G.Z. Li, B. Klein, C.B. Sun, and J. Kou, Applying
ments, Min. Process. Equip., 44(2016), No. 7, p. 5. Receiver–Operating–Characteristic (ROC) to bulk ore sorting
[17] H.J. Wang, X.P. Liu, G. Wang, and S. Han, Optimization of using XRF, Miner. Eng., 146(2020), art. No. 106117.
mobile manipulator sorting path based on improved genetic al- [38] W.Z. Yin, Y. Wu, Y.X. Han, et al., The theory and application
gorithm, J. Beijing Univ. Posts Telecommun., 43(2020), No. 5, of X-ray separation technology, China Min. Mag., 20(2011),
p. 34. No. 12, p. 88.
[18] Y.F. Wu, L.R. Gao, B. Zhang, H.N. Zhao, and J. Li, Real-time [39] M.B. Liu, W.Z. Yin, T.X. Han, and Z.Q. Sun, The study on the
implementation of optimized maximum noise fraction trans- X-ray separator and its pre-sorting test of the low grade ore of
form for feature extraction of hyperspectral images, J. Appl. Re- Mo and Ni, Min. Metall., 21(2012), No. 4, p. 26.
mote Sensing, 8(2014), No. 1, art. No. 084797. [40] M. Dalm, M.W.N. Buxton, and F.J.A. Ruitenbeek, Ore–waste
[19] S.W. Wu, J. Yang, and G.M. Cao, Prediction of the Charpy V- discrimination in epithermal deposits using near-infrared to
Notch impact energy of low carbon steel using a shallow neural short-wavelength infrared (NIR-SWIR) hyperspectral imagery,
network and deep learning, Int. J. Miner. Metall. Mater., Math. Geosci., 51(2019), No. 7, p. 849.
28(2021), No. 8, p. 1309. [41] E. Gülcan, A novel approach for sensor based sorting perform-
[20] M. Li, A. Caushaj, R. Silva, and D. Lowther, A neural network ance determination, Miner. Eng., 146(2020), art. No. 106130.
for electromagnetic based ore sorting, COMPEL Int. J. Comput. [42] Y.M. Liu, D. Chen, Q.F. Li, and K.X. Xu, Novel multi-scale
Math. Electr. Electron. Eng., 37(2018), No. 2, p. 691. modeling method for near infrared spectral measurement, Nan-
[21] C. Bergmann, Developments in Ore Sorting Technologies, otechnol. Precis. Eng., 12(2014), No. 5, p. 381.
Council for Mineral Technology, Randburg, 2009, p. 75. [43] M.R. Robben, H. Wotruba, and J. Heizmann, Sensor-based sep-
[22] C.Y. Xu, The application of photoelectric color separator and aration of carbonates, [in] IMPC 2012: XXVI International Min-
the optimization in mineral separation process of some mine, eral Processing Congress, New Delhi, 2012, p. 04496.
X.P. Luo et al., A review of intelligent ore sorting technology and equipment development 1655

[44] S. Iyakwari, H.J. Glass, and P.B. Kowalczuk, Potential for near by use of Fourier-transform infrared spectrometry, [in] Light
infrared sensor-based sorting of hydrothermally-formed miner- Metals 2014, Springer, Cham, 2014, p. 647-652.
als, J. Infrared Spectrosc., 21(2013), No. 3, p. 223. [56] Y. Qiu, B. Chen, and D.S. Jia, Development of application of
[45] H. Huang, W.H. Ye, T.Z. Xiong, P.L. Hu, and J.S. Huang, Zone infrared spectrometry, Environ. Sci. Surv., 27(2008), No. S1, p.
division scrap non-ferrous metal recognition algorithm based on 23.
dual energy X-ray transmission, Mach. Build. Autom., [57] X.F. Zhang, W. Ni, J.Y. Wu, and L.P. Zhu, Hydration mechan-
48(2019), No. 4, p. 26. ism of a cementitious material prepared with Si−Mn slag, Int. J.
[46] O. Udoudo, Modelling the Efficiency of an Automated Sensor- Miner. Metall. Mater., 18(2011), No. 2, p. 234.
based Sorter [Dissertation], University of Exeter, Exeter, 2010. [58] J. Moros, A. Gredilla, S. Fdez-Ortiz de Vallejuelo, A. de Diego,
[47] N.G. Cutmore, Y. Liu, and A.G. Middleton, On-line ore charac- J.M. Madariaga, S. Garrigues, and M. de la Guardia, Partial
terisation and sorting, Miner. Eng., 11(1998), No. 9, p. 843. least squares X-ray fluorescence determination of trace ele-
[48] M. Gaft, R. Reisfeld, and G. Panczer, Minerals radiometric sort- ments in sediments from the estuary of Nerbioi-Ibaizabal River,
ing and online process control, [in] Modern Luminescence Talanta, 82(2010), No. 4, p. 1254.
Spectroscopy of Minerals and Materials, Springer, Cham, 2015, [59] L.Q. Luo, G.Z. Ma, A. Ji, C.L. Guo, X.C. Zhan, and G.L. Li-
p. 499. ang, Neural cluster structure with multiple component predic-
[49] T.F. Wang and J.X. Yang, Online intelligent sorting method of
tion based on back error propagation in X-ray fluorescence
gold ore based on LIBS technology, [in] H. Haeri, ed., Materi-
spectrometry, J. Anal. Sci., 14(1998), No. 3, p. 177.
als in Environmental Engineering, De Gruyter, 2017, p. 753.
[60] W.Y. Li, W.H. Ye, and T.Z. Xiong, Metal identification al-
[50] R. Noll, H. Bette, A. Brysch, M. Kraushaar, I. Mönch, L. Peter,
gorithm based on bp neural network for dual energy X-ray
and V. Sturm, Laser-induced breakdown spectrometry—Ap-
transmission, Nonferrous Met. Eng., 10(2020), No. 8, p. 124.
plications for production control and quality assurance in the
[61] L.Q. Luo, L. Gan, X.J. Wu, A. Ji, and G.L. Liang, Correction of
steel industry, Spectrochim. Acta B: At. Spectrosc., 56(2001),
nonlinear matrix effects by using an algorithm of combining
No. 6, p. 637.
neural networks with fundamental parameters, Chin. J. Anal.
[51] J.M. Vadillo, S. Palanco, M.D. Romero, and J.J. Laserna, Ap-
plications of laser-induced breakdown spectrometry (LIBS) in Lab., 20(2001), No. 1, p. 1.
surface analysis, Fresenius J. Anal. Chem., 355(1996), No. 7-8, [62] R.L. Man, X.N. Zhao, and A. Ji, Application of partial least-
p. 909. squares method in multicomponent analysis by portable ra-
[52] M. Sabsabi and P. Cielo, Quantitative analysis of aluminum al- dioisotope XRFS, Spectrosc. Spectr. Anal., 11(1991), No. 4, p.
loys by laser-induced breakdown spectroscopy and plasma 50.
characterization, Appl. Spectrosc., 49(1995), No. 4, p. 499. [63] L.Q. Luo, G.Z. Ma, and A. Ji, Study of partial least squares re-
[53] Q.M. Peng, M.Y. Yao, M.H. Liu, Z.J. Lei, Y. Xu, and T.B. gression and fundamental parameters method in X-ray fluores-
Chen, Analysis of metal element Cr in fresh orange leaf using cence analysis, Chin. J. Anal. Chem., 20(1992), No. 9, p. 1074.
laser-induced breakdown spectroscopy, Acta Agric. Univ. Ji- [64] P. Urbański and E. Kowalska, Application of partial least-
angxiensis, 34(2012), No. 2, p. 397. squares calibration methods in low-resolution EDXRS, X Ray
[54] L.Z. Zahng, B.F. Ni, W.Z. Tian, et al., Status and development Spectrom., 24(1995), No. 2, p. 70.
of prompt γ-ray neutron activation analysis, Atom. Energy Sci. [65] L. Wielopolski and R.P. Gardner, Development of the detector
Technol., 2005, No. 3, p. 282. response function approach in the least-squares analysis of X-
[55] T.A. Aarhaug, A. Ferber, O. Kjos, and H. Gaertner, Online ray fluorescence spectra, Nucl. Instrum. Methods, 165(1979),
monitoring of aluminium primary production gas composition No. 2, p. 297.