Design, train, and test Sugeno-type fuzzy inference systems
The Neuro-Fuzzy Designer app lets you design, train, and test adaptive neuro-fuzzy inference systems (ANFIS) using input/output training data.
Using this app, you can:
Tune membership function parameters of Sugeno-type fuzzy inference systems.
Automatically generate an initial inference system structure based on your training data.
Modify the inference system structure before tuning.
Prevent overfitting to the training data using additional checking data.
Test the generalization ability of your tuned system using testing data.
Export your tuned fuzzy inference system to the MATLAB® workspace.
You can use the Neuro-Fuzzy Designer to train a type-1 Sugeno-type fuzzy inference system that:
Has a single output.
Uses weighted average defuzzification.
Has output membership functions all of the same type, for example
Has complete rule coverage with no rule sharing; that is, the number of rules must match the number of output membership functions, and every rule must have a different consequent.
Has unity weight for each rule.
Does not use custom membership functions.
Open the Neuro-Fuzzy Designer App
MATLAB Toolstrip: On the Apps tab, under Control System Design and Analysis, click the app icon.
MATLAB command prompt: Enter
neuroFuzzyDesigner opens the Neuro-Fuzzy
neuroFuzzyDesigner( opens the app and loads the
fuzzy inference system
fis can be any
sugfis object in the MATLAB workspace.
You can create an initial Sugeno-type fuzzy inference system
from training data using the
the app and loads a fuzzy inference system.
the name of a
.fis file on the MATLAB path.
To save a fuzzy inference system to a
In the Fuzzy Logic Designer, select File > Export > To File
At the command line, use
Support for representing fuzzy inference systems as structures will be removed
Warns starting in R2019b
Support for representing fuzzy inference systems as structures will be removed in a future
instead. There are differences between these representations that require updates to your
code. These differences include:
Object property names that differ from the corresponding structure fields.
Objects store text data as strings rather than as character vectors.
Also, all Fuzzy Logic Toolbox™ functions that accepted or returned fuzzy inference systems as structures now
accept and return either
To convert existing fuzzy inference system structures to objects, use the