A Penalty-Based Differential Evolution for Multimodal Optimization.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Locating multiple global optimal solutions (GOSs) for multimodal optimization problems (MMOPs) presents significant computational challenges.
• Existing algorithms often struggle to efficiently identify diverse sets of optimal solutions in complex search spaces.

Purpose of the Study:

• To develop a novel algorithm, Penalty-based Multimodal Optimization Differential Evolution (PMODE), for effectively addressing MMOPs.
• To enhance the capability of finding multiple GOSs in challenging optimization landscapes.

Main Methods:

• PMODE employs a penalty strategy with a dynamic penalty radius to guide the search towards distinct optimal regions.
• An elite selection mechanism is incorporated to identify promising solutions and focus exploration.
• The algorithm utilizes the JADE (Jumping Arithmetic Differential Evolution) variant as its core search engine.

Main Results:

• PMODE demonstrated superior performance compared to several state-of-the-art multimodal optimization algorithms.
• Experiments were conducted on 20 MMOPs from the IEEE CEC2013 special session, validating PMODE's effectiveness.
• The proposed method successfully located multiple GOSs across various benchmark problems.

Conclusions:

• PMODE offers a robust and effective approach for solving multimodal optimization problems.
• The dynamic penalty strategy and elite selection mechanism contribute significantly to finding multiple global optimal solutions.
• PMODE represents a valuable advancement in the field of evolutionary computation for MMOPs.