Confusing output

[Unknown]

Is there a degree of error in calculating the output of a fuzzy logic in jfuzzy lite?
I have the below code from matlab

[System]
Name=’congestion’
Type=’mamdani’
Version=2.0
NumInputs=2
NumOutputs=1
NumRules=5
AndMethod=’min’
OrMethod=’max’
ImpMethod=’min’
AggMethod=’max’
DefuzzMethod=’centroid’

[Input1]
Name=’VehicleDensity’
Range=[0 10]
NumMFs=5
MF1=’VeryLow’:’gaussmf’,[1.062 -2.776e-17]
MF2=’Low’:’gaussmf’,[1.062 2.5]
MF3=’Average’:’gaussmf’,[1.062 5]
MF4=’High’:’gaussmf’,[1.062 7.5]
MF5=’VeryHigh’:’gaussmf’,[1.062 10]

[Input2]
Name=’ModalSpeed’
Range=[0 100]
NumMFs=5
MF1=’VeryLow’:’gaussmf’,[10.62 -2.22e-16]
MF2=’Low’:’gaussmf’,[10.62 25]
MF3=’Average’:’gaussmf’,[10.62 50]
MF4=’High’:’gaussmf’,[10.62 75]
MF5=’VeryHigh’:’gaussmf’,[10.62 99.7354497354497]

[Output1]
Name=’LevelOfService’
Range=[0 60]
NumMFs=5
MF1=’VeryHigh’:’gaussmf’,[6.37 -2.22e-16]
MF2=’High’:’gaussmf’,[6.37 15]
MF3=’Average’:’gaussmf’,[6.37 30]
MF4=’Low’:’gaussmf’,[6.37 45]
MF5=’VeryLow’:’gaussmf’,[6.37 60]

[Rules]
1 5, 1 (1) : 1
2 4, 2 (1) : 1
3 3, 3 (1) : 1
4 2, 4 (1) : 1
5 1, 5 (1) : 1

the above code was used to create the below java code

public void defuzzifytraffic()
    {
        Engine engine = new Engine();
        engine.setName("CongestionDefuzzify");
        InputVariable inputVariable1 = new InputVariable();
        inputVariable1.setEnabled(true);
        inputVariable1.setName("VehicleDensityInput");
        inputVariable1.setRange(0,10);
        inputVariable1.addTerm(new Gaussian("VeryLow",1.062,-2.77e-17));
        inputVariable1.addTerm(new Gaussian("Low",1.062,2.5));
        inputVariable1.addTerm(new Gaussian("Average",1.062,5));
        inputVariable1.addTerm(new Gaussian("High",1.062,7.5));
        inputVariable1.addTerm(new Gaussian("VeryHigh",1.062,10));
        inputVariable1.setInputValue(this.getVehicleDensity());
        engine.addInputVariable(inputVariable1);
        
        InputVariable inputVariable2 = new InputVariable();
        inputVariable2.setEnabled(true);
        inputVariable2.setName("ModalSpeedInput");
        inputVariable2.setRange(0,100);
        inputVariable2.addTerm(new Gaussian("VeryLow",10.62,-2.22e-16));
        inputVariable2.addTerm(new Gaussian("Low",10.62,25));
        inputVariable2.addTerm(new Gaussian("Average",10.62,50));
        inputVariable2.addTerm(new Gaussian("High",10.62,75));
        inputVariable2.addTerm(new Gaussian("VeryHigh",10.62,100));
        inputVariable2.setInputValue(this.getModalSpeed());
        engine.addInputVariable(inputVariable2);
        
        OutputVariable OutputVariable = new OutputVariable();
        OutputVariable.setEnabled(true);
        OutputVariable.setName("LevelOfCongestionResult");
        OutputVariable.setRange(0,60);
        OutputVariable.fuzzyOutput().setAccumulation(new Maximum());
        OutputVariable.setDefaultValue(Double.NaN);
        OutputVariable.addTerm(new Gaussian("VeryHigh",6.37, -2.22e-16));
        OutputVariable.addTerm(new Gaussian("High",6.37, 15));
        OutputVariable.addTerm(new Gaussian("Average",6.37, 30));
        OutputVariable.addTerm(new Gaussian("Low",6.37, 45));
        OutputVariable.addTerm(new Gaussian("VeryLow",6.37, 60));
        
        engine.addOutputVariable(OutputVariable);
        RuleBlock ruleblock = new RuleBlock(); 
        ruleblock.setEnabled(true);
        ruleblock.setConjunction(new Minimum());
        ruleblock.setDisjunction(new Maximum());
        ruleblock.setActivation(new Minimum());
        ruleblock.addRule(Rule.parse("if VehicleDensityInput is VeryLow and ModalSpeedInput is VeryHigh then LevelOfCongestionResult is VeryHigh", engine));
        ruleblock.addRule(Rule.parse("if VehicleDensityInput is Low and ModalSpeedInput is High then LevelOfCongestionResult is High", engine));
        ruleblock.addRule(Rule.parse("if VehicleDensityInput is Average and ModalSpeedInput is Average then LevelOfCongestionResult is Average", engine));
        ruleblock.addRule(Rule.parse("if VehicleDensityInput is High and ModalSpeedInput is Low then LevelOfCongestionResult is Low", engine));
        ruleblock.addRule(Rule.parse("if VehicleDensityInput is VeryHigh and ModalSpeedInput is VeryLow then LevelOfCongestionResult is VeryLow", engine));
        engine.addRuleBlock(ruleblock);
        engine.configure("Minimum", "Maximum", "Minimum", "Maximum", "Centroid");
        StringBuilder status = new StringBuilder();
        if(!engine.isReady(status))
        {
            throw new RuntimeException("Engine not ready, the following errors we met:\n" + status.toString());
        }
        engine.setInputValue("VehicleDensityInput", this.getVehicleDensity());
        engine.setInputValue("ModalSpeedInput", this.getModalSpeed());
        engine.process();
        //this.setCongestionCoefficient(OutputVariable.getOutputValue()); 
        FuzzyLite.logger().info(String.format("result= %s", Op.str(OutputVariable.getOutputValue())));
    }

I get confused because i get different result having used to same input.
For instance, in matlab,

evalfis([5,95],b)

gives me

16.1383

while

engine.setInputValue(“VehicleDensityInput”, 5); engine.setInputValue(“ModalSpeedInput”, 95);

gives

28.521

using QTfuzzylite i get results similar to those i get from matlab. this leads me to think the problem is with my java code.

[Juan Rada-Vilela (admin)]

Hi,

I am sorry for the late response (I was on holidays).

The difference in the result is too big for it to be considered a margin of error.

Could you please try using values without scientific notation (i.e., convert -2.776e-17 to 0.0, -2.22e-16 to 0.0, and 99.7354497354497 to 100)? I just want to make sure the problem is not when parsing such numbers. Please compare the results with Matlab and jfuzzylite, and let me know.

Cheers.

[Author]

Posts