Incorporating Objective Function Information Into the Feasibility Rule for Constrained Evolutionary Optimization.

Area of Science:

• Optimization
• Computational Intelligence
• Algorithm Design

Background:

• Constrained optimization problems (COPs) are prevalent in science and engineering.
• Balancing constraints and objective functions is a critical challenge in evolutionary algorithms (EAs).
• Existing EAs often struggle to effectively manage this trade-off, leading to suboptimal solutions.

Purpose of the Study:

• To propose a new method for evolutionary constrained optimization that effectively balances constraints and objective functions.
• To enhance the performance of differential evolution (DE) for COPs.
• To improve the ability of EAs to escape local optima in infeasible regions.

Main Methods:

• A novel approach integrating differential evolution (DE) with a feasibility rule and an archive mechanism.
• Offspring generation using DE, followed by feasibility-based comparison with parents.
• An archive stores superior infeasible offspring, which are later used for population replacement.
• A mutation strategy is introduced to facilitate escaping local optima in infeasible regions.
• Objective function information is utilized in DE offspring generation and replacement strategies.

Main Results:

• The proposed method demonstrates an effective balance between constraints and objective functions in COPs.
• Experimental results on IEEE CEC2006 and IEEE CEC2010 benchmark test functions show superior or competitive performance.
• The method's advantage is observed to increase with a higher number of decision variables.

Conclusions:

• The developed method offers a robust solution for constrained optimization problems using evolutionary algorithms.
• The integration of feasibility rules, archiving, and a specialized mutation strategy enhances optimization performance.
• The findings suggest the method's scalability and effectiveness, particularly for high-dimensional problems.