Error with Fuzzy Tree Model

Hello everyone,

I create a fuzzy tree (aggregated structure) in Matlab and I have an error which I hope to be supported.

My fuzzy tree has 3 fuzzy models (nd_fis, lk_fis, Aggfis3). nd_fis is the node fis and it has 2 inputs: HC and EC. lk_fis is link_fis and it has 3 input: LQI,Mobility and PFI. Outputs of nd_fis and lk_fis are Inputs of fis3.

However, when I run it, the error is “Error using fistree/set.Connections

Specified connections include unrecognized FIS name "Aggfis3".

Error in fistree (line 293)

obj.Connections = p.Results.Connections;

What's wrong please, following is the code with the line underlined is where the error appears:

// Fuzzy model 1

nd_fis = mamfis('Name',"FNode");

nd_fis = addInput(nd_fis,[0 7],'Name',"HC");

nd_fis = addMF(nd_fis,"HC","trapmf",[0 0 1 2],'Name',"Short");

nd_fis = addMF(nd_fis,"HC","trapmf",[1 2 3 4],'Name',"Medium");

nd_fis = addMF(nd_fis,"HC","trapmf",[3 5 7 7],'Name',"Long");

nd_fis = addInput(nd_fis,[0 250],'Name',"EC");

nd_fis = addMF(nd_fis,"EC","trapmf",[0 0 50 100],'Name',"Low");

nd_fis = addMF(nd_fis,"EC","trapmf",[50 100 150 200],'Name',"Moderate");

nd_fis = addMF(nd_fis,"EC","trapmf",[150 200 250 250],'Name',"High");

nd_fis = addOutput(nd_fis,[0 100],'Name',"NS");

nd_fis = addMF(nd_fis,"NS","trimf",[0 25 45],'Name',"Low");

nd_fis = addMF(nd_fis,"NS","trimf",[25 45 75],'Name',"Medium");

nd_fis = addMF(nd_fis,"NS","trimf",[50 75 100],'Name',"High");

nd_rulelist = ["if HC is Short and EC is Low then NS is High";

"if HC is Short and EC is Moderate then NS is Medium";

"if HC is Short and EC is High then NS is Low";

"if HC is Medium and EC is Low then NS is High";

"if HC is Medium and EC is Moderate then NS is High";

"if HC is Medium and EC is High then NS is Medium";

"if HC is Long and EC is Low then NS is High";

"if HC is Long and EC is Moderate then NS is Medium";

"if HC is Long and EC is High then NS is Low"];

FINALnd_fis = addRule(nd_fis,nd_rulelist);

// Fuzzy model 2

lk_fis = mamfis('Name',"FLink");

lk_fis = addInput(lk_fis,[0 7],'Name',"LQI");

lk_fis = addMF(lk_fis,"LQI","trapmf",[0 0 1 2],'Name',"Fair");

lk_fis = addMF(lk_fis,"LQI","trapmf",[1 2 3 4],'Name',"Good");

lk_fis = addMF(lk_fis,"LQI","trapmf",[3 5 7 7],'Name',"Excellent");

lk_fis = addInput(lk_fis,[0 20],'Name',"Mobility");

lk_fis = addMF(lk_fis,"Mobility","trapmf",[0 0 2 6],'Name',"Slow");

lk_fis = addMF(lk_fis,"Mobility","trapmf",[4 7 11 14],'Name',"Medium");

lk_fis = addMF(lk_fis,"Mobility","trapmf",[12 18 20 20],'Name',"Fast");

lk_fis = addInput(lk_fis,[0 100],'Name',"PFI");

lk_fis = addMF(lk_fis,"PFI","trapmf",[0 0 2 40],'Name',"Low");

lk_fis = addMF(lk_fis,"PFI","trapmf",[30 45 55 70],'Name',"Medium");

lk_fis = addMF(lk_fis,"PFI","trapmf",[60 80 100 100],'Name',"High");

lk_fis = addOutput(lk_fis,[0 100],'Name',"LS");

lk_fis = addMF(lk_fis,"LS","trimf",[0 25 45],'Name',"Low");

lk_fis = addMF(lk_fis,"LS","trimf",[25 45 75],'Name',"Medium");

lk_fis = addMF(lk_fis,"LS","trimf",[50 75 100],'Name',"High");

lk_rulelist = ["if LQI is Excellent and Mobility is Slow and PFI is High then LS is High";

"if LQI is Excellent and Mobility is Slow and PFI is Medium then LS is High";

"if LQI is Excellent and Mobility is Slow and PFI is Low then LS is High";

"if LQI is Excellent and Mobility is Medium and PFI is High then LS is High";

"if LQI is Excellent and Mobility is Medium and PFI is Medium then LS is High";

"if LQI is Excellent and Mobility is Medium and PFI is Low then LS is Medium";

"if LQI is Excellent and Mobility is Fast and PFI is High then LS is Medium";

"if LQI is Excellent and Mobility is Fast and PFI is Medium then LS is Medium";

"if LQI is Excellent and Mobility is Fast and PFI is Low then LS is Low";

"if LQI is Good and Mobility is Slow and PFI is High then LS is High";

"if LQI is Good and Mobility is Slow and PFI is Medium then LS is High";

"if LQI is Good and Mobility is Slow and PFI is Low then LS is Medium";

"if LQI is Good and Mobility is Medium and PFI is High then LS is High";

"if LQI is Good and Mobility is Medium and PFI is Medium then LS is Medium";

"if LQI is Good and Mobility is Medium and PFI is Low then LS is Medium";

"if LQI is Good and Mobility is Fast and PFI is High then LS is Medium";

"if LQI is Good and Mobility is Fast and PFI is Medium then LS is Medium";

"if LQI is Good and Mobility is Fast and PFI is Low then LS is Low";

"if LQI is Fair and Mobility is Slow and PFI is High then LS is High";

"if LQI is Fair and Mobility is Slow and PFI is Medium then LS is Medium";

"if LQI is Fair and Mobility is Slow and PFI is Low then LS is Low";

"if LQI is Fair and Mobility is Medium and PFI is High then LS is Medium";

"if LQI is Fair and Mobility is Medium and PFI is Medium then LS is Medium";

"if LQI is Fair and Mobility is Medium and PFI is Low then LS is Low";

"if LQI is Fair and Mobility is Fast and PFI is High then LS is Medium";

"if LQI is Fair and Mobility is Fast and PFI is Medium then LS is Low";

"if LQI is Fair and Mobility is Fast and PFI is Low then LS is Low"

];

FINALlk_fis = addRule(lk_fis,lk_rulelist);

//Fuzzy model 3

Aggfis3 = mamfis('Name','RoutingDecision');

Aggfis3 = addInput(Aggfis3,[0 100],'Name',"LS");

Aggfis3 = addMF(Aggfis3,"LS","trimf",[0 25 45],'Name',"Low");

Aggfis3 = addMF(Aggfis3,"LS","trimf",[25 45 75],'Name',"Medium");

Aggfis3 = addMF(Aggfis3,"LS","trimf",[50 75 100],'Name',"High");

Aggfis3 = addInput(Aggfis3,[0 100],'Name',"NS");

Aggfis3 = addMF(Aggfis3,"NS","trimf",[0 25 45],'Name',"Low");

Aggfis3 = addMF(Aggfis3,"NS","trimf",[25 45 75],'Name',"Medium");

Aggfis3 = addMF(Aggfis3,"NS","trimf",[50 75 100],'Name',"High");

Aggfis3.Outputs(1).Name = "RD";

Aggfis3 = addMF(Aggfis3,"RD","trimf",[0 12 24],'Name',"VL");

Aggfis3 = addMF(Aggfis3,"RD","trimf",[12 24 26],'Name',"L");

Aggfis3 = addMF(Aggfis3,"RD","trimf",[24 36 48],'Name',"ML");

Aggfis3 = addMF(Aggfis3,"RD","trimf",[36 48 60],'Name',"M");

Aggfis3 = addMF(Aggfis3,"RD","trimf",[48 60 72],'Name',"MH");

Aggfis3 = addMF(Aggfis3,"RD","trimf",[60 72 84],'Name',"H");

Aggfis3 = addMF(Aggfis3,"RD","trimf",[72 84 100],'Name',"VH");

rulefis3d = ["if NS is Low and LS is Low then RD is VL";

"if NS is Low and LS is Medium then RD is L";

"if NS is Low and LS is High then RD is ML";

"if NS is Medium and LS is Low then RD is ML";

"if NS is Medium and LS is Medium,then RD is M";

"if NS is Medium and LS is High then RD is MH";

"if NS is High and LS is Low then RD is MH";

"if NS is High and LS is Medium then RD is H";

"if NS is High and LS is High then RD is VH"];

Aggfis3 = addRule(Aggfis3,rulefis3d);

con1 = ["FNode/NS" "Aggfis3/NS"];

con2 = ["FLink/LS" "Aggfis3/LS"];

aggTree = fistree([FINALnd_fis FINALlk_fis Aggfis3],[con1;con2]);

aggTree.FIS(3) = addRule(aggTree.FIS(3), rulefis3d)

Answer 1

[Sam Chak]

HI @Layla Mohammad

Please check this out and see if these MFs, Surfaces, and the FIS Tree are what you are expecting.

% Fuzzy model 1
nd_fis = mamfis('Name', "FNode");
nd_fis = addInput(nd_fis, [0 7], 'Name', "HC");
nd_fis = addMF(nd_fis, "HC", "trapmf", [0 0 1 2], 'Name', "Short" );
nd_fis = addMF(nd_fis, "HC", "trapmf", [1 2 3 4], 'Name', "Medium");
nd_fis = addMF(nd_fis, "HC", "trapmf", [3 5 7 7], 'Name', "Long"  );
nd_fis = addInput(nd_fis, [0 250], 'Name', "EC");
nd_fis = addMF(nd_fis, "EC", "trapmf", [  0   0  50 100], 'Name', "Low"     );
nd_fis = addMF(nd_fis, "EC", "trapmf", [ 50 100 150 200], 'Name', "Moderate");
nd_fis = addMF(nd_fis, "EC", "trapmf", [150 200 250 250], 'Name', "High"    );
nd_fis = addOutput(nd_fis, [0 100], 'Name', "NS");
nd_fis = addMF(nd_fis, "NS", "trimf", [ 0  25  45], 'Name', "Low"   );
nd_fis = addMF(nd_fis, "NS", "trimf", [25  45  75], 'Name', "Medium");
nd_fis = addMF(nd_fis, "NS", "trimf", [50  75 100], 'Name', "High"  );
nd_rulelist = ["if HC is Short  and EC is Low      then NS is High"; 
               "if HC is Short  and EC is Moderate then NS is Medium"; 
               "if HC is Short  and EC is High     then NS is Low"; 
               "if HC is Medium and EC is Low      then NS is High"; 
               "if HC is Medium and EC is Moderate then NS is High"; 
               "if HC is Medium and EC is High     then NS is Medium"; 
               "if HC is Long   and EC is Low      then NS is High"; 
               "if HC is Long   and EC is Moderate then NS is Medium";
               "if HC is Long   and EC is High     then NS is Low"];
nd_fis = addRule(nd_fis, nd_rulelist);      % <--- change in this line

figure(1)
tiledlayout(2,2);
nexttile
plotmf(nd_fis, 'input', 1)
nexttile
plotmf(nd_fis, 'input', 2)
nexttile
plotmf(nd_fis, 'output', 1)
nexttile
opt = gensurfOptions('NumGridPoints', 21);
gensurf(nd_fis, opt)

% Fuzzy model 2
lk_fis = mamfis('Name', "FLink");
lk_fis = addInput(lk_fis, [0 7], 'Name', "LQI");
lk_fis = addMF(lk_fis, "LQI", "trapmf", [0 0 1 2], 'Name', "Fair"     );
lk_fis = addMF(lk_fis, "LQI", "trapmf", [1 2 3 4], 'Name', "Good"     );
lk_fis = addMF(lk_fis, "LQI", "trapmf", [3 5 7 7], 'Name', "Excellent");
lk_fis = addInput(lk_fis, [0 20], 'Name',"Mobility");
lk_fis = addMF(lk_fis, "Mobility", "trapmf", [ 0  0  2  6], 'Name', "Slow"  );
lk_fis = addMF(lk_fis, "Mobility", "trapmf", [ 4  7 11 14], 'Name', "Medium");
lk_fis = addMF(lk_fis, "Mobility", "trapmf", [12 18 20 20], 'Name', "Fast"  );
lk_fis = addInput(lk_fis, [0 100], 'Name', "PFI");
lk_fis = addMF(lk_fis, "PFI", "trapmf", [ 0   0   2  40], 'Name', "Low"   );
lk_fis = addMF(lk_fis, "PFI", "trapmf", [30  45  55  70], 'Name', "Medium");
lk_fis = addMF(lk_fis, "PFI", "trapmf", [60  80 100 100], 'Name', "High"  );
lk_fis = addOutput(lk_fis, [0 100], 'Name', "LS");
lk_fis = addMF(lk_fis, "LS", "trimf", [ 0  25  45], 'Name', "Low"   );
lk_fis = addMF(lk_fis, "LS", "trimf", [25  45  75], 'Name', "Medium");
lk_fis = addMF(lk_fis, "LS", "trimf", [50  75 100], 'Name', "High"  );
lk_rulelist = ["if LQI is Excellent and Mobility is Slow   and PFI is High   then LS is High"; 
               "if LQI is Excellent and Mobility is Slow   and PFI is Medium then LS is High"; 
               "if LQI is Excellent and Mobility is Slow   and PFI is Low    then LS is High"; 
               "if LQI is Excellent and Mobility is Medium and PFI is High   then LS is High"; 
               "if LQI is Excellent and Mobility is Medium and PFI is Medium then LS is High"; 
               "if LQI is Excellent and Mobility is Medium and PFI is Low    then LS is Medium"; 
               "if LQI is Excellent and Mobility is Fast   and PFI is High   then LS is Medium"; 
               "if LQI is Excellent and Mobility is Fast   and PFI is Medium then LS is Medium";
               "if LQI is Excellent and Mobility is Fast   and PFI is Low    then LS is Low";
               "if LQI is Good      and Mobility is Slow   and PFI is High   then LS is High";
               "if LQI is Good      and Mobility is Slow   and PFI is Medium then LS is High";
               "if LQI is Good      and Mobility is Slow   and PFI is Low    then LS is Medium";
               "if LQI is Good      and Mobility is Medium and PFI is High   then LS is High";
               "if LQI is Good      and Mobility is Medium and PFI is Medium then LS is Medium";
               "if LQI is Good      and Mobility is Medium and PFI is Low    then LS is Medium";
               "if LQI is Good      and Mobility is Fast   and PFI is High   then LS is Medium";
               "if LQI is Good      and Mobility is Fast   and PFI is Medium then LS is Medium";
               "if LQI is Good      and Mobility is Fast   and PFI is Low    then LS is Low";
               "if LQI is Fair      and Mobility is Slow   and PFI is High   then LS is High";
               "if LQI is Fair      and Mobility is Slow   and PFI is Medium then LS is Medium";
               "if LQI is Fair      and Mobility is Slow   and PFI is Low    then LS is Low";
               "if LQI is Fair      and Mobility is Medium and PFI is High   then LS is Medium";
               "if LQI is Fair      and Mobility is Medium and PFI is Medium then LS is Medium";
               "if LQI is Fair      and Mobility is Medium and PFI is Low    then LS is Low";
               "if LQI is Fair      and Mobility is Fast   and PFI is High   then LS is Medium";
               "if LQI is Fair      and Mobility is Fast   and PFI is Medium then LS is Low";
               "if LQI is Fair      and Mobility is Fast   and PFI is Low    then LS is Low"];
lk_fis = addRule(lk_fis,lk_rulelist);       % <--- change in this line

figure(2)
tiledlayout(2,2);
nexttile
plotmf(lk_fis, 'output', 1)
nexttile
opt = gensurfOptions('InputIndex', [1 2], 'NumGridPoints', 21);
gensurf(lk_fis, opt)
nexttile
opt = gensurfOptions('InputIndex', [1 3], 'NumGridPoints', 21);
gensurf(lk_fis, opt)
nexttile
opt = gensurfOptions('InputIndex', [2 3], 'NumGridPoints', 21);
gensurf(lk_fis, opt)

% Fuzzy model 3
Aggfis3 = mamfis('Name', 'RoutingDecision');
Aggfis3 = addInput(Aggfis3, [0 100], 'Name', "LS");
Aggfis3 = addMF(Aggfis3, "LS", "trimf", [ 0  25  45], 'Name', "Low"   );
Aggfis3 = addMF(Aggfis3, "LS", "trimf", [25  45  75], 'Name', "Medium");
Aggfis3 = addMF(Aggfis3, "LS", "trimf", [50  75 100], 'Name', "High"  );
Aggfis3 = addInput(Aggfis3, [0 100], 'Name', "NS");
Aggfis3 = addMF(Aggfis3, "NS", "trimf", [ 0  25  45], 'Name', "Low"   );
Aggfis3 = addMF(Aggfis3, "NS", "trimf", [25  45  75], 'Name', "Medium");
Aggfis3 = addMF(Aggfis3, "NS", "trimf", [50  75 100], 'Name', "High"  );
% Aggfis3.Outputs(1).Name = "RD";
Aggfis3 = addOutput(Aggfis3, [0 100], 'Name', "RD");
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [ 0  12  24], 'Name',"VL");
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [12  24  26], 'Name',"L" );
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [24  36  48], 'Name',"ML");
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [36  48  60], 'Name',"M" );
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [48  60  72], 'Name',"MH");
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [60  72  84], 'Name',"H" );
Aggfis3 = addMF(Aggfis3, "RD", "trimf", [72  84 100], 'Name',"VH");
rulefis3d = ["if NS is Low    and LS is Low    then RD is VL"; 
             "if NS is Low    and LS is Medium then RD is L"; 
             "if NS is Low    and LS is High   then RD is ML";
             "if NS is Medium and LS is Low    then RD is ML"; 
             "if NS is Medium and LS is Medium then RD is M"; 
             "if NS is Medium and LS is High   then RD is MH";
             "if NS is High   and LS is Low    then RD is MH"; 
             "if NS is High   and LS is Medium then RD is H"; 
             "if NS is High   and LS is High   then RD is VH"];
Aggfis3 = addRule(Aggfis3, rulefis3d);

figure(3)
tiledlayout(2,2);
nexttile
plotmf(Aggfis3, 'input', 1)
nexttile
plotmf(Aggfis3, 'input', 2)
nexttile
plotmf(Aggfis3, 'output', 1)
nexttile
opt = gensurfOptions('NumGridPoints', 21);
gensurf(Aggfis3, opt)

aggTree = fistree([nd_fis lk_fis Aggfis3], []);     % <--- change in this line
aggTree.Connections = [
    "FNode/NS" "RoutingDecision/NS";                % <--- change in this line
    "FLink/LS" "RoutingDecision/LS"];               % <--- change in this line
figure(4)
plotfis(aggTree)

[Layla Mohammad]

@Sam Chak

Yes thank you

one more question, how to save the result as fis?

As i have to convert the whole code as c code for simulation integration

[Sam Chak]

You are welcome, @Layla Mohammad. If you find the solution on the FIS Tree issue helpful, please consider accepting ✔ and voting 👍 on the answer. Try the following for converting FIS tree into the homogeneous structure that is intended for code generation.

LaylaTreeData = getFISCodeGenerationData(aggTree)

https://www.mathworks.com/help/fuzzy/getfiscodegenerationdata.html

https://www.mathworks.com/help/coder/ug/homogeneous-vs-heterogeneous-cell-arrays.html

Also check MATLAB® Coder™: https://www.mathworks.com/help/coder/index.html

% Construct FIS Tree
aggTree = fistree([lk_fis nd_fis Aggfis3], []);     % <--- change in this line
aggTree.Connections = [
    "FLink/LS" "RoutingDecision/LS";                % <--- change in this line
    "FNode/NS" "RoutingDecision/NS"];               % <--- change in this line
figure(4)
plotfis(aggTree)

aggTree

aggTree = 
  fistree with properties:

                       Name: "fistreemodel"
                        FIS: [1×3 mamfis]
                Connections: [2×2 string]
                     Inputs: [5×1 string]
                    Outputs: "RoutingDecision/RD"
    DisableStructuralChecks: 0

	See 'getTunableSettings' method for parameter optimization.

% Converting FIS tree into the homogeneous structure
LaylaTreeData = getFISCodeGenerationData(aggTree)

LaylaTreeData = struct with fields:
              FIS: {[1×1 struct]  [1×1 struct]  [1×1 struct]}
      Connections: {2×2 cell}
          Outputs: {'RoutingDecision/RD'}
    EvaluationFcn: [1×1 struct]

[Layla Mohammad]

@Sam Chak Dear Sir

Would you please just send me an example on how to deal with the resulted structure from the last point!

I spent two days reading and trying, nothing was going well.

All i need is to convert the fistree to a file.FIS, Can i do that?

If not, then how to deal with the tree as FIS?

Thanks in advance

[Sam Chak]

Hi @Layla Mohammad

The only way to save a FIS object to a .fis file is by using the writeFIS() command.

plotfis(aggTree)

aggTree

aggTree = 
  fistree with properties:

                       Name: "fistreemodel"
                        FIS: [1×3 mamfis]
                Connections: [2×2 string]
                     Inputs: [5×1 string]
                    Outputs: "RoutingDecision/RD"
    DisableStructuralChecks: 0

	See 'getTunableSettings' method for parameter optimization.

writeFIS(aggTree, aggTree.Name)

Incorrect number or types of inputs or outputs for function 'writeFIS'.

LaylaTreeData = getFISCodeGenerationData(aggTree);

As you can see, it throws an error message. It appears that the Fuzzy Logic Toolbox does not currently have a direct built-in function to save the "FIS tree model" into a .fis file.

However, since the "FIS tree model" is already in the Workspace, it is possible to save "aggTree" or "LaylaTreeData" as a .MAT file. This way, the FIS tree model can be reloaded later for conversion to C code.

[Layla Mohammad]

@Sam Chak Thank you very much for your efforts and reply, I already saved it as mat file.

The thing is i'm struggling in converting to c

[Sam Chak]

Hi @Layla Mohammad

I recommend that you create a new post and tag it with 'MATLAB Coder' as the issue is related to generating C code from MATLAB code for the hardware platform.

Consider using the following title: "Generating C Code from FIS Tree Model using MATLAB Coder."

In your post, you can explain that you currently have a MATLAB algorithm (link to this thread) that generates the FIS Tree Model within the Workspace.

[Layla Mohammad]

@Sam Chak Yes sure

Thank you very much for your reply and efforts