FluentFuzzy

FluentFuzzy is a package for doing fuzzy logic in .NET with a natural fluent syntax.

FluentFuzzy is very much WIP and should be expected to be updated with breaking changes.

Contents

FluentFuzzy contains the core fuzzy logic functionality.

FluentFuzzy.Visualization has visualization for some FluentFuzzy classes.

Using

After importing the Nuget package, using FluentFuzzy is a 2-step process.

Setting up input and output

FuzzyInputs and FuzzyOutputs can be set up in the following manner:

// test/FluentFuzzy.Test/Example.cs#L10-L12

const int healthValue = 35;
var health = new FuzzyInput(() => healthValue);
var flee = new FuzzyOutput();

FuzzyInput requires a Func<double> to get the current crisp value of the input.

Setting up member functions

The next step is adding member functions to each input:

// test/FluentFuzzy.Test/Example.cs#L14-L20

const int low = 0;
const int medium = 1;
const int high = 2;

health.Set(low, new Trapezoid(0, 0, 25, 50));
health.Set(medium, new Triangle(25, 50, 75));
health.Set(high, new Trapezoid(50, 75, 100, 100));

Currently, the following membership functions have been implemented:

• Triangle defines a triangle.
• Trapezoid defines a trapezoid.
• Line defines a line.
• Value defines a single value.
• AtLeast defines a threshold value.

Custom membership functions can be created by implementing the IMembershipFunction interface.

After the input functions, membership functions can be added to the FuzzyOutput

// test/FluentFuzzy.Test/Example.cs#L22-L24

flee.Set(low, new Triangle(-0.5, 0, 0.5));
flee.Set(medium, new Triangle(0, 0.5, 1));
flee.Set(high, new Triangle(0.5, 1, 1.5));

The only currently supported defuzzification method is a version of the centroid method. As some of the membership functions, namely Line, Value and AtLeast do not have useful centroids, these cannot be used as output member functions.

Custom output membership functions can be created by implementing the IHasCentroid interface.

Setting up fuzzy rules

After setting up the fuzzy inputs and outputs, rules can be created to connect them.

Rules are created with a readable fluent syntax:

// test/FluentFuzzy.Test/Example.cs#L26-L28

FuzzyRule.If(health.Is(high)).Then(flee.Is(low));
FuzzyRule.If(health.Is(medium)).Then(flee.Is(medium));
FuzzyRule.If(health.Is(low)).Then(flee.Is(high));

Evaluating an output

After creating the fuzzy ruleset, the output can be evaluated by calling the Evaluate method on the output:

// test/FluentFuzzy.Test/Example.cs#L30-L30

Console.WriteLine($"flee(health: {health.Value}) = {flee.Evaluate()}"); // flee(health: 35) = 0,8