1106 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
average content of mineral in the product, and thus, opera-
tional regime directly affects the quality of the flotational
products.
Another interesting issue is that variability of Pb con-
tent is different for each test, and its changeability may be,
to some extent, connected to the level of Pb content in
the froth. For example, for low Pb content, the standard
deviation is low and equals 0.47 percent. Together with
increasing of the average Pb content (test 2), the standard
deviation increases either, but further increase in Pb con-
tent decreases the standard deviation. Therefore, it is pos-
sible to conclude, that the relationship between variability
of useful mineral content in flotation product and quality
of flotation product, measured through the average content
of Pb, can be described through a parabolic function. It
should to be pointed out that it is not a firm state, and
such relationship should be verified through more detailed
investigation, prior to acceptance.
Table 2. summarizes the results obtained during image
recognition and classification into specific groups. For the
purposes of this article it was assumed that three groups of
images, reflecting three levels of Pb content, were registered.
It can be seen from the Table how many images (in percent-
ages) were assigned to individual groups. For example, for
test no. 1, 60% of images were assigned to group 1, 29.8%
to group 2 and 10.2% to group 3. The results should be
interpreted in rows, not columns.
DISCUSSION AND SUMMARY
The results of experiments presented in the paper show that
it is possible to link the characteristics of flotation froth
with the quality of individual beneficiation products. The
results also showed that the characteristics of froth for indi-
vidual tests are different and each test was carried out at
different operational conditions. Thus, the results obtained
through the vision system can be correlated with the pro-
cess course and, as a result, it can be possible to control the
content of the useful mineral in the product through the
control of machine work.
It was also shown that implemented algorithms worked
efficiently and were able to adjust the process of image rec-
ognition. It should be noted, however, that the process of
machine learning is more efficient to conduct for stable
conditions of Pb contents and low variations. Operational
Figure 3. Raw results of testing programme
Table 1. Results of experiments
Test 1 Test 2 Test 3
Average content of Pb 65.00 73.04 76.79
St dev. 0.47 1.49 0.80
Range 1.6 4.1 2.1
Table 2. Recognition of images from individual groups
Group 1 Group 2 Group 3
Group 1 60.0% 29.8% 10.2%
Group 2 25.9% 74.1% 0.0%
Group 3 1.5% 24.9% 73.6%
average content of mineral in the product, and thus, opera-
tional regime directly affects the quality of the flotational
products.
Another interesting issue is that variability of Pb con-
tent is different for each test, and its changeability may be,
to some extent, connected to the level of Pb content in
the froth. For example, for low Pb content, the standard
deviation is low and equals 0.47 percent. Together with
increasing of the average Pb content (test 2), the standard
deviation increases either, but further increase in Pb con-
tent decreases the standard deviation. Therefore, it is pos-
sible to conclude, that the relationship between variability
of useful mineral content in flotation product and quality
of flotation product, measured through the average content
of Pb, can be described through a parabolic function. It
should to be pointed out that it is not a firm state, and
such relationship should be verified through more detailed
investigation, prior to acceptance.
Table 2. summarizes the results obtained during image
recognition and classification into specific groups. For the
purposes of this article it was assumed that three groups of
images, reflecting three levels of Pb content, were registered.
It can be seen from the Table how many images (in percent-
ages) were assigned to individual groups. For example, for
test no. 1, 60% of images were assigned to group 1, 29.8%
to group 2 and 10.2% to group 3. The results should be
interpreted in rows, not columns.
DISCUSSION AND SUMMARY
The results of experiments presented in the paper show that
it is possible to link the characteristics of flotation froth
with the quality of individual beneficiation products. The
results also showed that the characteristics of froth for indi-
vidual tests are different and each test was carried out at
different operational conditions. Thus, the results obtained
through the vision system can be correlated with the pro-
cess course and, as a result, it can be possible to control the
content of the useful mineral in the product through the
control of machine work.
It was also shown that implemented algorithms worked
efficiently and were able to adjust the process of image rec-
ognition. It should be noted, however, that the process of
machine learning is more efficient to conduct for stable
conditions of Pb contents and low variations. Operational
Figure 3. Raw results of testing programme
Table 1. Results of experiments
Test 1 Test 2 Test 3
Average content of Pb 65.00 73.04 76.79
St dev. 0.47 1.49 0.80
Range 1.6 4.1 2.1
Table 2. Recognition of images from individual groups
Group 1 Group 2 Group 3
Group 1 60.0% 29.8% 10.2%
Group 2 25.9% 74.1% 0.0%
Group 3 1.5% 24.9% 73.6%