An Improved DERFF Based on Evolution Process

The utilization and balance of constraint violations and objective function values are important issues in solving constrained optimization problems. In constrained optimization evolutionary algorithms (COEAs), designing related algorithms to achieve optimal performance has been a focus for many researchers. DERFF has made a notable attempt in this direction. By weighting two ranking values, one is based on ε-constraint handling techniques and the other is based on the objective function value, and considering the information from the evolutionary process (i.e., the feasible solution proportion and generation information), good results were achieved. However, in DERFF, when the weight exceeds 1, it is directly set to 1, which leaves room for improvement in the use of evolutionary process information. This paper introduces a new method for determining the weight, i.e., the maximum of the feasible solution proportion and generation information, aiming to reduce the correlation between these two parameters. Different parameters in the formula were also tested in the experiments. The experiments show that the improved algorithm shows a good comparative performance, which lays a certain foundation for the future research.