Model-Less Control - Real-Time Rendering: Computer Graphics with Control Engineering - page 90

Control
Output

370

Reference

Switch

Fuzzy Logic
Controller

+
-

+
+

Out1
Out2
Out3

Subtract In1

In2

Plant Output

Add

4.4e4

Feedback FPS

Error

To Plant

Feedback Input Trim point

4.4e4 gives 380fps

From Plant

Plant Output vs
Reference

InputToPlant

To Workspace

RefVsOutput

To Workspace2

FIGURE 6.1 Fuzzy control system in Simulink/MATLAB.

TABLE 6.1
Fuzzy Inference Rule Set
If

Then

fps_error IS High vertex_count IS DecreaseHigh

fps_error IS Low vertex_count IS IncreaseHigh

implementation of fuzzy control. Figure 6.2 illustrates the graphical user interface
that allows the creation of membership functions, the set-up of the rule base, and
several other parameters that may be changed.
Fuzzy logic deals with non-crisp values. Thus the approach to tuning fuzzy logic
controller parameters relies on heuristics and iterative processes that allow easy
observation of the effects on simulation performance from changes in fuzzy con-
troller parameters. Some parameters that may be changed include the membership
function and the membership input and output ranges.
The fuzzy logic toolbox provides a step-through functionality in simulation
time. This allows a user to observe how a defuzzified decision is derived by view-
ing the fuzzified inputs and how they are combined to produce the output via the
firing function. This tool is important for helping a user decide the appropriate
membership function to use by analysing the output of the fuzzy controller over a
series of steps.

6.3 ADAPTIVE NEURAL FUZZY CONTROL
We described the basic structure of the type of fuzzy inference system in a sys-
tematic manner. In brief, it consists of multi-tier relationships that first map input