5
running since the raw ore of the plant comes from its own mines. The other parameters such as 1
feed flow rate, pulp concentration and particle size of solid are controlled in the grinding circuit. 2
When the fluctuation of the parameters directly impact on good flotation running, the reagent 3
dosage need to be regulated to stabilize flotation running. The parameters should be necessary 4
condition for successful flotation such that the fluctuation of the parameters are considered to be 5
disturbance variables for flotation. 6
During the production of flotation, the workers regulate the reagent dosage based on two
7
principles. Firstly, the workers determines primary reagent dosage according to the feed grade and
8
the handling capacity of ore when the flotation starts running. Secondly, the workers regulate the
9
reagent dosage and pulp level again according to their observable bubble size each time regular
10
inspection during flotation running. The reagent dosage and the pulp level need to be repeatedly
11
regulated until the output bubble size meet their criteria. However, abnormal flotation states such
12
as froth outflow, pulp upturn and low pulp level, etc. sometimes occur for some reasons when the
13
workers don't inspect the flotation process during flotation running. This cannot result in
14
immediate respond to this situation, which is one of drawbacks of manual control mode.
15
Considering flotation mechanism based robust reagent dosage control scheme is poorly
16
modeled because of inherent complexity, dynamics and nonlinearity of flotation and the lack of
17
testing equipments such as X-ray fluorescence analyzers, it is possible to build bubble size
18
distribution based reagent dosage control algorithm since bubble size structure reflects important
19
process characteristics and responds to changes in the reagent dosage. By employing bubble size
20
as process outputs, the output bubble size is capable to be predicted from the given values of the
21
reagent dosage and the historical data of bubble size. Instead of using a concentrate grade or
22
recovery, the PDF of bubble size can be used as a target variable. In addition, a feedforward
23
compensator is added to regulate the reagent dosage in advance according to the feed disturbances
24
in order to reduce the influence of large disturbances on flotation. Generally, for the use of
25
feedforward control, two somewhat separate issues have to be solved. The first is the question of
26
how suitable disturbance information can be obtained. The second is how this information can be
27
used best to improve control performance. Fortunately, the solid percent, the slurry flow rate, the
28
pulp level and the reagent dosage are capable to be obtained all the time and the feed grade is
29
capable to be obtained by assay every 2 hours.
30
3 Combined
FFB
reagent dosage control
31
3.1 Control scheme
32
As discussed above, it is shown in Fig. 2 that the proposed control scheme mainly consists of 33
a feedforward compensator and a feedback predictive controller. Here the reagent dosage stands 34
for the manipulated variables. The feed flow rate, the pulp concentration and the feed grade stand 35
for the measured input disturbances, whose effects are compensated by the feedforward part of the 36
controller. The PDF of the output bubble size is taken as controlled variable, which closely related 37
to the indicators. It is noticed that there are always some measurement error that will degrade the 38
performance of the controller for a real flotation process. However, as long as the information 39
about measured disturbances are well provided above just assuming the feed grade to remain 40
constant for some time, the feedforward compensator can be expected to improve control 41
performance. Meanwhile a BSD based feedback predictive control is used to assure the control of the 42