XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 899
Amankwaa-Kyeremeh, B., Greet, C., Skinner, W.,
Asamoah, R.K., 2021b. Correlating process mineral-
ogy and pulp chemistry for quick ore variability diag-
nosis, ed. International Future Mining Conference
Online, D. Australia Australia Australasian Institute of
Mining and Metallurgy, Australia, pp. 1–10.
Amankwaa-Kyeremeh, B., Greet, C., Zanin, M., Skinner,
W., Asamoah, R.K., 2020a. Predictability of rougher
flotation copper recovery using Gaussian process regres-
sion algorithm, eds. th, U.B.I.M., Mineral Conference
Tarkwa, G.A. Ghana UMaT, Ghana, pp. 1–8.
Amankwaa-Kyeremeh, B., Greet, C., Zanin, M., Skinner,
W., Asamoah, R.K., 2020b. Selecting key predic-
tor parameters for regression modelling using modi-
fied Neighbourhood Component Analysis (NCA)
Algorithm, eds. th, U.B.I.M., Mineral Conference
Tarkwa, G.A. Ghana UMaT, Ghana, pp. 320–325.
Amankwaa-Kyeremeh, B., McCamley, C., Zanin, M.,
Greet, C., Ehrig, K., Asamoah, R.K., Prediction and
Optimisation of Copper Recovery in the Rougher
Flotation Circuit. Minerals, 2024, 14(1), 36.
Amankwaa-Kyeremeh, B., Skinner, W., Asamoah, R.K.,
2021c. Comparative study on rougher copper recov-
ery prediction using selected predictive algorithms,
ed. International Future Mining Conference Online,
D. Australia Australia Australasian Institute of Mining
and Metallurgy, Australia, pp. 1–10.
Amankwaa-Kyeremeh, B., Zhang, J., Zanin, M., Skinner,
W., Asamoah, R.K., Feature selection and Gaussian
process prediction of rougher copper recovery. Minerals
Engineering, 2021d, 170(107041), 1–14.
Arthur, C.K., Temeng, V.A., Ziggah, Y.Y., Novel approach
to predicting blast-induced ground vibration using
Gaussian process regression. Engineering with
Computers, 2020, 36(1), 29–42.
Asamoah, R.K., Baawuah, E., Greet, C., Skinner, W.,
Characterisation of metal debris in grinding and flota-
tion circuits. Minerals Engineering, 2021, 171.
Bhavsar, H., Panchal, M.H., A review on support vector
machine for data classification. International Journal
of Advanced Research in Computer Engineering and
Technology, 2012, 1(10), 185–189.
Breiman, L., Cutler, A., Manual–Setting Up, Using, and
Understanding Random Forests, v 4.0, URL: ftp://ftp.
stat. berkeley. edu/pub/users/breiman. 2003.
Dankwah, J.B., Asamoah, R.K., Zanin, M., Skinner, W.,
Dense liquid flotation: Can coarse particle flotation
performance be enhanced by controlling fluid density?
Minerals Engineering, 2022a, 180, 107513.
Dankwah, J.B., Asamoah, R.K., Zanin, M., Skinner, W.,
Influence of water rate, gas rate, and bed particle size
on bed-level and coarse particle flotation performance.
Minerals Engineering, 2022b, 183, 107622.
Forson, P., Zanin, M., Abaka-Wood, G., Skinner, W.,
Asamoah, R., Flotation of auriferous arsenopyrite from
pyrite using thionocarbamate. Minerals Engineering,
2022, 181, 107524.
Forson, P., Zanin, M., Skinner, W., Asamoah, 2020. A brief
review of auriferous sulphide flotation concentration:.
pyrite and arsenopyrite mineral separation, In UMaT
Biennial International Mining Mineral Conference
Tarkwa, Ghana August. UMaT Ghana, Ghana,
pp. 1–12.
Gomez-Flores, A., Heyes, G.W., Ilyas, S., Kim, H.,
Prediction of grade and recovery in flotation from
physicochemical and operational aspects using machine
learning models. Minerals Engineering, 2022, 183,
107627.
Grömping, U., Relative importance for linear regression in
R: the package relaimpo. Journal of statistical software,
2006, 17(1), 1–27.
Hodouin, D., Methods for automatic control, observation,
and optimization in mineral processing plants. Journal
of Process Control, 2011, 21(2), 211–225.
Kesler, S.E., Simon, A.C., Simon, A.F., Mineral resources,
economics and the environment. 2015, Cambridge
University Press.
Kuipers, K.J.J., van Oers, L.F.C.M., Verboon, M., van der
Voet, E., Assessing environmental implications associ-
ated with global copper demand and supply scenarios
from 2010 to 2050. Global Environmental Change,
2018, 49, 106–115.
Latorre, M., Troncoso, R., Uauy, R., 2019. Chapter
4 -Biological Aspects of Copper, In Clinical and
Translational Perspectives on WILSON DISEASE, eds.
Kerkar, N., Roberts, E.A. Academic Press, pp. 2531.
Lee, Y., Oh, S.-H., Kim, M.W., 1991. The effect of initial
weights on premature saturation in back-propagation
learning, In IJCNN-91-Seattle international joint con-
ference on neural networks. IEEE, pp. 765–770.
Mathe, Z., Harris, M., O’Connor, C., A review of methods
to model the froth phase in non-steady state flotation
systems. Minerals Engineering, 2000, 13(2), 127–140.
Nakhaei, F., Irannajad, M.J.P.P.o.M.P., Comparison
between neural networks and multiple regression
methods in metallurgical performance modeling of flo-
tation column. 2013, 49.

Previous Page Next Page

Extracted Text (may have errors)

XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 899
Amankwaa-Kyeremeh, B., Greet, C., Skinner, W.,
Asamoah, R.K., 2021b. Correlating process mineral-
ogy and pulp chemistry for quick ore variability diag-
nosis, ed. International Future Mining Conference
Online, D. Australia Australia Australasian Institute of
Mining and Metallurgy, Australia, pp. 1–10.
Amankwaa-Kyeremeh, B., Greet, C., Zanin, M., Skinner,
W., Asamoah, R.K., 2020a. Predictability of rougher
flotation copper recovery using Gaussian process regres-
sion algorithm, eds. th, U.B.I.M., Mineral Conference
Tarkwa, G.A. Ghana UMaT, Ghana, pp. 1–8.
Amankwaa-Kyeremeh, B., Greet, C., Zanin, M., Skinner,
W., Asamoah, R.K., 2020b. Selecting key predic-
tor parameters for regression modelling using modi-
fied Neighbourhood Component Analysis (NCA)
Algorithm, eds. th, U.B.I.M., Mineral Conference
Tarkwa, G.A. Ghana UMaT, Ghana, pp. 320–325.
Amankwaa-Kyeremeh, B., McCamley, C., Zanin, M.,
Greet, C., Ehrig, K., Asamoah, R.K., Prediction and
Optimisation of Copper Recovery in the Rougher
Flotation Circuit. Minerals, 2024, 14(1), 36.
Amankwaa-Kyeremeh, B., Skinner, W., Asamoah, R.K.,
2021c. Comparative study on rougher copper recov-
ery prediction using selected predictive algorithms,
ed. International Future Mining Conference Online,
D. Australia Australia Australasian Institute of Mining
and Metallurgy, Australia, pp. 1–10.
Amankwaa-Kyeremeh, B., Zhang, J., Zanin, M., Skinner,
W., Asamoah, R.K., Feature selection and Gaussian
process prediction of rougher copper recovery. Minerals
Engineering, 2021d, 170(107041), 1–14.
Arthur, C.K., Temeng, V.A., Ziggah, Y.Y., Novel approach
to predicting blast-induced ground vibration using
Gaussian process regression. Engineering with
Computers, 2020, 36(1), 29–42.
Asamoah, R.K., Baawuah, E., Greet, C., Skinner, W.,
Characterisation of metal debris in grinding and flota-
tion circuits. Minerals Engineering, 2021, 171.
Bhavsar, H., Panchal, M.H., A review on support vector
machine for data classification. International Journal
of Advanced Research in Computer Engineering and
Technology, 2012, 1(10), 185–189.
Breiman, L., Cutler, A., Manual–Setting Up, Using, and
Understanding Random Forests, v 4.0, URL: ftp://ftp.
stat. berkeley. edu/pub/users/breiman. 2003.
Dankwah, J.B., Asamoah, R.K., Zanin, M., Skinner, W.,
Dense liquid flotation: Can coarse particle flotation
performance be enhanced by controlling fluid density?
Minerals Engineering, 2022a, 180, 107513.
Dankwah, J.B., Asamoah, R.K., Zanin, M., Skinner, W.,
Influence of water rate, gas rate, and bed particle size
on bed-level and coarse particle flotation performance.
Minerals Engineering, 2022b, 183, 107622.
Forson, P., Zanin, M., Abaka-Wood, G., Skinner, W.,
Asamoah, R., Flotation of auriferous arsenopyrite from
pyrite using thionocarbamate. Minerals Engineering,
2022, 181, 107524.
Forson, P., Zanin, M., Skinner, W., Asamoah, 2020. A brief
review of auriferous sulphide flotation concentration:.
pyrite and arsenopyrite mineral separation, In UMaT
Biennial International Mining Mineral Conference
Tarkwa, Ghana August. UMaT Ghana, Ghana,
pp. 1–12.
Gomez-Flores, A., Heyes, G.W., Ilyas, S., Kim, H.,
Prediction of grade and recovery in flotation from
physicochemical and operational aspects using machine
learning models. Minerals Engineering, 2022, 183,
107627.
Grömping, U., Relative importance for linear regression in
R: the package relaimpo. Journal of statistical software,
2006, 17(1), 1–27.
Hodouin, D., Methods for automatic control, observation,
and optimization in mineral processing plants. Journal
of Process Control, 2011, 21(2), 211–225.
Kesler, S.E., Simon, A.C., Simon, A.F., Mineral resources,
economics and the environment. 2015, Cambridge
University Press.
Kuipers, K.J.J., van Oers, L.F.C.M., Verboon, M., van der
Voet, E., Assessing environmental implications associ-
ated with global copper demand and supply scenarios
from 2010 to 2050. Global Environmental Change,
2018, 49, 106–115.
Latorre, M., Troncoso, R., Uauy, R., 2019. Chapter
4 -Biological Aspects of Copper, In Clinical and
Translational Perspectives on WILSON DISEASE, eds.
Kerkar, N., Roberts, E.A. Academic Press, pp. 2531.
Lee, Y., Oh, S.-H., Kim, M.W., 1991. The effect of initial
weights on premature saturation in back-propagation
learning, In IJCNN-91-Seattle international joint con-
ference on neural networks. IEEE, pp. 765–770.
Mathe, Z., Harris, M., O’Connor, C., A review of methods
to model the froth phase in non-steady state flotation
systems. Minerals Engineering, 2000, 13(2), 127–140.
Nakhaei, F., Irannajad, M.J.P.P.o.M.P., Comparison
between neural networks and multiple regression
methods in metallurgical performance modeling of flo-
tation column. 2013, 49.

Help

Destination page number Search scope Search Text

loading

898 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
VAF indicator was also used to verify the correctness
of the models and how well they could make predictions.
These results indicate that, GPR model which outper-
formed SVM, ANN, LR and RF can explain about 99.99%,
94.70% and 94.65% of the potential variance in the pre-
dicted rougher copper recovery values from the training,
validation and testing data sets, respectively. Again, random
forest followed GPR for the VAF results.
From Table 2, it can clearly be seen that GPR model
produced the most precise rougher copper recovery predic-
tion as compared to SVM, ANN, LR and RF. This GPR
performance may be attributed to its intrinsic ability to
add a prior knowledge and specification about the shape of
the model, capturing the uncertainties using noise variance
hyperparameter during formulation stage. The next best
model from Table 2 is the RF model. The performance of
RF model could be linked to the fact that it is an ensemble
learner which builds multiple predictor trees and finds the
average prediction value in solving a problem. This ensem-
ble technique in RF makes it better than most standalone
predictive models. Again, from Table 2, it is obvious that
SVM and ANN models had a similar performance which
was below the performance of both GPR and RF models.
This is mainly because both SVM and ANN models had
very similar learning ability during the training phase as
well as a similar generalisation capability during the vali-
dation and testing phases. The LR performance may be
attributed to its inability to capture the complex nonlinear
relationship. Using large data timeframe aimed at captur-
ing significant variabilities which may occur during that
period, allowing the model to be more robust as complex
correlation structure may be built. This could also reduce
the drifting on the model as new data becomes available. In
the case of drifting, the model could be re-trained to cap-
ture the new correlation structure. Our future studies will
investigate the generalisation of the models for other rel-
evant systems. Ensemble learning could be used to address
disturbances that could not be captured by a single data-
driven model.
CONCLUSION
Copper recovery prediction using support vector machine,
Gaussian process regression, multi-layer perceptron artifi-
cial neural network, linear regression, and random forest
algorithms has been investigated. The predictive model
performance was assessed using correlation coefficient (𝑟),
root mean square error (RMSE), mean absolute percent-
age error (MAPE) and variance accounted for (VAF). The
results showed that, except LR model, all the investigated
predictive models make good copper recovery prediction
using feed grade, feed particle size, throughput, xanthate
dosage, frother dosage, airflow rate and froth depth as input
variables. The performance assessment showed that GPR
model makes the most precise rougher copper recovery
prediction obtaining 𝑟 values 0.96, RMSE values 0.42,
MAPE values 0.25% and VAF values 94% from train-
ing, validation and testing data sets. It should however be
noted that, this model will continue to maintain its pre-
diction accuracy provided the characteristics of the feed in
terms of ore complexity and variability does not change
significantly. Once this happens, retraining of the model
may be required. The significance of this work is that the
developed GPR model could be used to simulate rougher
copper recovery. Our future work will seek to develop pre-
dictive models for the other sections of the flotation circuit
(cleaners and scavengers) with the aim having an integrated
predictive model for the whole flotation circuit, integrating
ensemble learning for better predictive performance.
ACKNOWLEDGMENT
This research has been supported by the South Australian
Government and BHP Olympic Dam through the PRIF
RCP Industry Consortium. The authors would also like
to appreciate BHP Olympic Dam for approving the pub-
lication of this article. Financial support from the Future
Industries Institute of the University of South Australia
is also acknowledged. Support from the Australia-
India Strategic Research Fund for the Recovery of the
Battery Materials and REE from Ores and Wastes is also
acknowledged.
REFERENCES
Ali, D., Hayat, M.B., Alagha, L., Molatlhegi, O.K., An
evaluation of machine learning and artificial intel-
ligence models for predicting the flotation behavior
of fine high-ash coal. Advanced Powder Technology,
2018, 29(12), 3493–3506.
Amankwaa-Kyeremeh, B., Ehrig, K., Greet, C., Asamoah,
R., Pulp Chemistry Variables for Gaussian Process
Prediction of Rougher Copper Recovery. Minerals,
2023, 13(6), 731.
Amankwaa-Kyeremeh, B., Greet, C., Skinner, W., Asamoah,
R., 2021a. A brief review of pulp chemistry parameters
in relation to flotation feed variation, In International
Mineral Processing Congress, Cape Town, South Africa,
October. The Southern African Institute of Mining and
Metallurgy, South Africa, pp. 1855–1863.

Previous Page Next Page

Extracted Text (may have errors)

898 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
VAF indicator was also used to verify the correctness
of the models and how well they could make predictions.
These results indicate that, GPR model which outper-
formed SVM, ANN, LR and RF can explain about 99.99%,
94.70% and 94.65% of the potential variance in the pre-
dicted rougher copper recovery values from the training,
validation and testing data sets, respectively. Again, random
forest followed GPR for the VAF results.
From Table 2, it can clearly be seen that GPR model
produced the most precise rougher copper recovery predic-
tion as compared to SVM, ANN, LR and RF. This GPR
performance may be attributed to its intrinsic ability to
add a prior knowledge and specification about the shape of
the model, capturing the uncertainties using noise variance
hyperparameter during formulation stage. The next best
model from Table 2 is the RF model. The performance of
RF model could be linked to the fact that it is an ensemble
learner which builds multiple predictor trees and finds the
average prediction value in solving a problem. This ensem-
ble technique in RF makes it better than most standalone
predictive models. Again, from Table 2, it is obvious that
SVM and ANN models had a similar performance which
was below the performance of both GPR and RF models.
This is mainly because both SVM and ANN models had
very similar learning ability during the training phase as
well as a similar generalisation capability during the vali-
dation and testing phases. The LR performance may be
attributed to its inability to capture the complex nonlinear
relationship. Using large data timeframe aimed at captur-
ing significant variabilities which may occur during that
period, allowing the model to be more robust as complex
correlation structure may be built. This could also reduce
the drifting on the model as new data becomes available. In
the case of drifting, the model could be re-trained to cap-
ture the new correlation structure. Our future studies will
investigate the generalisation of the models for other rel-
evant systems. Ensemble learning could be used to address
disturbances that could not be captured by a single data-
driven model.
CONCLUSION
Copper recovery prediction using support vector machine,
Gaussian process regression, multi-layer perceptron artifi-
cial neural network, linear regression, and random forest
algorithms has been investigated. The predictive model
performance was assessed using correlation coefficient (𝑟),
root mean square error (RMSE), mean absolute percent-
age error (MAPE) and variance accounted for (VAF). The
results showed that, except LR model, all the investigated
predictive models make good copper recovery prediction
using feed grade, feed particle size, throughput, xanthate
dosage, frother dosage, airflow rate and froth depth as input
variables. The performance assessment showed that GPR
model makes the most precise rougher copper recovery
prediction obtaining 𝑟 values 0.96, RMSE values 0.42,
MAPE values 0.25% and VAF values 94% from train-
ing, validation and testing data sets. It should however be
noted that, this model will continue to maintain its pre-
diction accuracy provided the characteristics of the feed in
terms of ore complexity and variability does not change
significantly. Once this happens, retraining of the model
may be required. The significance of this work is that the
developed GPR model could be used to simulate rougher
copper recovery. Our future work will seek to develop pre-
dictive models for the other sections of the flotation circuit
(cleaners and scavengers) with the aim having an integrated
predictive model for the whole flotation circuit, integrating
ensemble learning for better predictive performance.
ACKNOWLEDGMENT
This research has been supported by the South Australian
Government and BHP Olympic Dam through the PRIF
RCP Industry Consortium. The authors would also like
to appreciate BHP Olympic Dam for approving the pub-
lication of this article. Financial support from the Future
Industries Institute of the University of South Australia
is also acknowledged. Support from the Australia-
India Strategic Research Fund for the Recovery of the
Battery Materials and REE from Ores and Wastes is also
acknowledged.
REFERENCES
Ali, D., Hayat, M.B., Alagha, L., Molatlhegi, O.K., An
evaluation of machine learning and artificial intel-
ligence models for predicting the flotation behavior
of fine high-ash coal. Advanced Powder Technology,
2018, 29(12), 3493–3506.
Amankwaa-Kyeremeh, B., Ehrig, K., Greet, C., Asamoah,
R., Pulp Chemistry Variables for Gaussian Process
Prediction of Rougher Copper Recovery. Minerals,
2023, 13(6), 731.
Amankwaa-Kyeremeh, B., Greet, C., Skinner, W., Asamoah,
R., 2021a. A brief review of pulp chemistry parameters
in relation to flotation feed variation, In International
Mineral Processing Congress, Cape Town, South Africa,
October. The Southern African Institute of Mining and
Metallurgy, South Africa, pp. 1855–1863.

900 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
Pu, Y., Szmigiel, A., Chen, J., Apel, D.B., FlotationNet: A
hierarchical deep learning network for froth flotation
recovery prediction. Powder Technology, 2020, 375,
317–326.
Quintanilla, P., Neethling, S.J., Brito-Parada, P.R.,
Modelling for froth flotation control: A review.
Minerals Engineering, 2021, 162, 106718.
Shahbazi, B., Chehreh Chelgani, S., Matin, S.S., Prediction
of froth flotation responses based on various condition-
ing parameters by Random Forest method. Colloids
and Surfaces A: Physicochemical and Engineering
Aspects, 2017, 529, 936–941.
Sverdrup, H.U., Ragnarsdottir, K.V., Koca, D., On mod-
elling the global copper mining rates, market supply,
copper price and the end of copper reserves. Resources,
Conservation and Recycling, 2014, 87, 158–174.

Previous Page Next Page

Extracted Text (may have errors)