A Memetic Algorithm for Global Optimization of Multimodal Nonseparable Problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Swarm Intelligence

Background:

• High-dimensional, nonseparable optimization problems present significant challenges due to unknown interdependencies among variables.
• Avoiding local optima is a critical issue in many computational tasks.
• Existing optimization algorithms may struggle with the complexity of these problems.

Purpose of the Study:

• To propose a novel memetic algorithm (MA) that improves optimization performance.
• To address the challenge of avoiding local optima in high-dimensional nonseparable problems.
• To introduce a hybrid approach combining cooperative particle swarm optimization and modified harmony search.

Main Methods:

• A novel memetic algorithm (MA) named cooperative particle swarm optimizer-modified harmony search (CPSO-MHS) is developed.
• Cooperative Particle Swarm Optimizer (CPSO) is employed for efficient local search, optimizing each dimension separately.
• Modified Harmony Search (MHS) is utilized for global search, focusing on recombining elements and identifying global optima.

Main Results:

• The CPSO-MHS algorithm demonstrated strong performance across 28 standard benchmark problems.
• Experimental results indicate the algorithm's effectiveness in solving multimodal nonseparable problems.
• Comparisons with seven other optimization algorithms and existing MAs highlight CPSO-MHS's competitive advantages.

Conclusions:

• The proposed CPSO-MHS algorithm offers a robust solution for complex optimization tasks.
• The synergistic interaction between CPSO's local search and MHS's global search is key to its success.
• CPSO-MHS effectively navigates search spaces to find global optima, overcoming local optimum entrapment.