Model-Less Control - Real-Time Rendering: Computer Graphics with Control Engineering - page 93

FIGURE 6.4 (See colour insert.) Neural network model structure in ANFIS.

The fuzzy rule-set used in these experiments is shown in Table 6.1. The selection
of an appropriate membership function is non-automatic and we begin with a generic
non-linear segment of a parabolic curve. The next step is testing the fuzzy controller with
the actual rendering process over several iterations to ensure that both the gradient of the
selected range and membership enrollment (curve) functions are suitable to perform
the tracking function correctly. The membership functions are shown in Figure 6.6.

Control
Output

370

Reference

Switch

Fuzzy Logic
Controller

+
-

+
+

Out1
Out2
Out3

In1
In2

Subtract

Plant Output

Add

Feedback FPS 4.4e4

Error

To Plant

Feedback Input Trim point

4.4e4 gives 380fps

From Plant

Plant Output vs
Reference

InputToPlant

To Workspace

RefVsOutput

To Workspace2

FIGURE 6.5 Using ANFIS for controlling real-time rendering process.