An adaptive differential evolution algorithm for global optimization in dynamic environments.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Dynamic optimization problems (DOPs) present significant challenges for traditional algorithms.
• Maintaining population diversity and exploration is crucial for effective DOP solutions.
• Existing evolutionary algorithms often struggle with premature convergence and local optima stagnation in dynamic environments.

Purpose of the Study:

• To propose an efficient multipopulation-based adaptive differential evolution (DE) algorithm for solving DOPs.
• To enhance the diversity, exploration, and optima tracking abilities of DE for dynamic environments.
• To prevent premature convergence and stagnation at local optima.

Main Methods:

• Introduced dynamic DE with Brownian and quantum individuals (DDEBQ) algorithm.
• Incorporated Brownian and adaptive quantum individuals to maintain population diversity.
• Employed a neighborhood-driven double mutation strategy to control perturbation.
• Utilized an exclusion rule for better optima tracking.
• Integrated an aging mechanism to prevent stagnation.

Main Results:

• DDEBQ demonstrated superior performance compared to state-of-the-art evolutionary algorithms.
• The algorithm showed statistically significant improvements across various benchmark DOP instances.
• Effectiveness validated using benchmarks from the generalized dynamic benchmark generator (GDBG) system.

Conclusions:

• DDEBQ is an efficient and effective algorithm for solving dynamic optimization problems.
• The proposed mechanisms significantly enhance performance in dynamic and uncertain environments.
• DDEBQ offers a robust solution for complex optimization challenges in dynamic settings.