In order to increase diversity for GA, Mansoornejad et al use the final population in the existing GA as
the initial population in another new GA, while our approach tries to apply different random seeds for the
GA. The hybrid GA – SQP method proposed by Mansoornejad is sound; however, there are still some
drawbacks which can be improved by our approach. First, it seems somewhat undesirable to predetermine
the stall generation. Since improvements can only be assessed in comparisons with other solutions, the
stopping criterion is unclear. Additionally, users without experience may have difficulties in setting a
suitable threshold value. In our approach, we use SQP to produce the starting solution which provides a
reasonable threshold for the following GAs.