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

Algorithm 1

Recursive clustering algorithm with drift detection.

Input: ;

Output: ;

Read the first data sample ;

Initialize the first cluster;

for    do

  Read ;

  Compute for all clusters;

  Identify the closest cluster;

  if    then

   Update the closest cluster;

  else

   Update the dissimilarity number ;

   Compute and ;

    if    then

    Update and ;

    end if

    if    then

    Store in the data window ;

    end if

    if    then

    Create new cluster;

    end if

  end if

  Compute and for all clusters;

  if   or   then

   Merge redundant clusters;

  end if

end for