Fault Diagnosis with Evolving Fuzzy Classifier Based on Clustering Algorithm and Drift Detection

Algorithm 2

Evolving fuzzy classifier.

Input: ;

Output: ;

Initialize the classifier;

for    do

  Read ;

  Execute the recursive clustering algorithm;

  if  new cluster is created  then

   Create new fuzzy rule;

   Define the new class elicited by expert/system operator;

    = label of the new class;

  end if

  if  cluster is updated  then

   Update the corresponding fuzzy rule;

   Find the most active rule;

    = label of the most active rule;

  end if

  if  clusters are merged  then

   Merge the corresponding fuzzy rules;

  end if

end for