Multiscale Cooperative Differential Evolution Algorithm.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Traditional differential evolution algorithms suffer from limited search range in early stages and slow convergence in later stages.
• These limitations hinder their effectiveness in solving complex global optimization problems.

Purpose of the Study:

• To propose a multiscale cooperative differential evolution algorithm that addresses the limitations of traditional methods.
• To enhance both the search range and convergence speed of differential evolution algorithms.

Main Methods:

• Utilized a multipopulation mechanism with distinct mutation strategies for each subpopulation to maintain diversity.
• Incorporated covariance learning among populations to create an optimal rotating coordinate system for cross-operations.
• Implemented an adaptive parameter adjustment strategy to balance population exploration and convergence.

Main Results:

• The proposed algorithm demonstrated superior performance on the CEC 2005 benchmark functions.
• Experimental results indicated better global optimization capabilities compared to state-of-the-art evolutionary algorithms.

Conclusions:

• The multiscale cooperative differential evolution algorithm effectively overcomes the narrow search range and slow convergence issues.
• The developed algorithm offers a robust and efficient solution for global optimization problems.