Image Processing Reference
In-Depth Information
Without Smith predictor
2000
50
0
1500
-50
1000
-100
500
-150
0
-200
-500
-250
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
Iteration #
Iteration #
FIGURE 9.37 TC and actuator response for a step change to TC target with time delay of 10
cycles and without Smith predictor (dashed curve: TC target input; solid curve: TC output
[left], dispensed mass [right]).
W i th Smith predictor
4.4
7
4.2
6
4
5
3.8
4
3.6
3
3.4
2
3.2
1
3
0
2.8
-1
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
Iteration #
Iteration #
FIGURE 9.38 TC and actuator response for a step change to TC target with time delay of 10
cycles and with Smith predictor (dashed curve: TC target input; solid curve: TC output [left],
dispensed mass [right]).
predictor improves the stability of the loop (Figures 9.36 and 9.38), while at the
same time injects some overshoot when compared to the response without any
delay as in Figure 9.33.
9.11.4 F EEDFORWARD C OMPENSATION FOR I MAGE D ISTURBANCE
Generally, pixel count is used to estimate the amount of toner required for develop-
ing the images. Since images can have varying degrees of area coverage, it can
represent a large disturbance to the TC control system, depending on how the area
coverages are distributed while printing. Particularly, if the jobs contain a mixture of
low and high area coverages and the change is rapid, the system can be overloaded
with disturbance before it has time to recover. This can induce undesirable
uctu-
ations in TC and even a loss of stability. Proper feedforward compensation of the
control input,
u, from the pixel counter can potentially minimize the impact.
Search WWH ::




Custom Search