Identification of CTQs for Complex Products Based on Mutual Information and Improved Gravitational Search Algorithm

Algorithm 1

MIGSA.

Input: data set ; , the number of characteristics;

   , the set of characteristics; ; , the set of classes;

  Values of parameters in GSA: , the size of agents;

   , the maximum iteration; , a constant; , the initial positions

  of agents. , initial velocities of agents; ,

  the best fitness value.

Output: CTQs

Steps:

for   to , do

  Compute

end

Select the characteristic with maximum

while    do

  for each characteristic , do

     Compute

  end

  Select the next characteristic with the maximum .

end

Initialize

while   do

  Evaluate the fitness value for each agent and find

  the best fitness. If the best value is met, then stop.

  Otherwise, update agents’ position by (8). If the

  best global position is no change for consecutive

  generations, use improving strategies to generate

  new agents.

end