HCS12 datasheet(373/452 Pages) FREESCALE

373 / 452 page

S12CPUV2 Reference Manual, Rev. 4.0

Freescale Semiconductor

373

instructions would form the working part of a 16-bit resolution fuzzy

inference routine.

There are many other methods of performing inference, but none of these

are as widely used as the min-max method. Since the CPU12 is a

general-purpose microcontroller, the programmer has complete freedom to

program any algorithm desired. A custom programmed algorithm would

typically take more code space and execution time than a routine that used

the built-in REV or REVW instructions.

9.7.3 Defuzzification Variations

Other CPU12 instructions can help with custom defuzzification routines in

two main areas:

• The first case is working with operands that are more than eight bits.

• The second case involves using an entirely different approach than

weighted average of singletons.

The primary part of the WAV instruction is a multiply and accumulate

operation to get the numerator for the weighted average calculation. When

working with operands as large as 16 bits, the EMACS instruction could at

least be used to automate the multiply and accumulate function. The CPU12

has extended math capabilities, including the EMACS instruction which

uses 16-bit input operands and accumulates the sum to a 32-bit memory

location and 32-bit by 16-bit divide instructions.

One benefit of the WAV instruction is that both a sum of products and a sum

of weights are maintained, while the fuzzy output operand is only accessed

from memory once. Since memory access time is such a significant part of

execution time, this provides a speed advantage compared to conventional

instructions.

The weighted average of singletons is the most commonly used technique

in microcontrollers because it is computationally less difficult than most

other methods. The simplest method is called max defuzzification, which

simply uses the largest fuzzy output as the system result. However, this

approach does not take into account any other fuzzy outputs, even when

they are almost as true as the chosen max output. Max defuzzification is not

a good general choice because it only works for a subset of fuzzy logic

applications.

HCS12 데이터시트(PDF) 373 Page - Freescale Semiconductor, Inc