Surrogate-assisted global and distributed local collaborative optimization algorithm for expensive constrained optimization problems | JoVE Visualize

Area of Science:

Optimization algorithms
Computational intelligence
Engineering mathematics

Background:

Expensive constrained optimization problems pose significant challenges due to increasing complexity and solution costs.
Traditional algorithms struggle to balance global exploration and local exploitation, particularly in high-dimensional or complex constraint scenarios.
Efficiently solving these problems with limited computational resources is a critical research area.

Purpose of the Study:

To introduce a novel algorithm, the surrogate-assisted global and distributed local collaborative optimization (SGDLCO), for tackling expensive constrained optimization problems.
To enhance the efficiency and effectiveness of optimization processes by collaboratively integrating global and local search strategies.
To address the limitations of traditional methods in balancing exploration and exploitation for complex optimization tasks.

Main Methods:

The SGDLCO algorithm employs two collaborative surrogate optimization phases per generation.
A global phase uses classification collaborative mutation for candidate set generation, reducing surrogate model pre-screening pressure.
A local phase implements distributed central region exploration, identifying promising areas via affinity propagation clustering and mathematical modeling.
A three-layer adaptive selection strategy balances feasibility, diversity, and convergence to identify optimal solutions.

Main Results:

The SGDLCO algorithm demonstrates an effective balance between global and local search throughout the optimization process.
Experimental evaluations on five classical test suites confirm the algorithm's superior performance.
The method successfully addresses the challenges of expensive constrained optimization problems.

Conclusions:

The proposed SGDLCO algorithm offers a robust and efficient solution for expensive constrained optimization.
The collaborative integration of global and local surrogate-assisted phases significantly improves optimization outcomes.
SGDLCO provides a valuable tool for researchers and practitioners dealing with computationally intensive optimization challenges.