Multi-strategy collaborative optimization of gravitational search algorithm.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• The standard Gravitational Search Algorithm (GSA) suffers from local optima, slow convergence, and low solution accuracy.
• Addressing these limitations is crucial for effective application in complex problem-solving.

Purpose of the Study:

• To propose a novel multi-strategy cooperative optimization approach for the Gravitational Search Algorithm (GSA).
• To enhance GSA's global exploration and local exploitation capabilities.
• To improve solution accuracy, convergence speed, and stability.

Main Methods:

• In early iterations, original gravitational force updates particles.
• Later stages utilize a globally optimal Lévy random walk for fitter particles and a sparrow algorithm follower strategy for poorer particles.
• Lens-imaging opposition-based learning is employed to increase population diversity and search range.

Main Results:

• The proposed algorithm effectively balances global exploration and local exploitation.
• Performance analysis on 24 benchmark functions shows improved solution accuracy and stability.
• Comparative tests against other GSA variants and advanced algorithms demonstrate superior performance.

Conclusions:

• The multi-strategy cooperative optimization significantly enhances GSA's efficiency and effectiveness.
• The algorithm shows strong potential for real-world engineering design optimization problems.
• This improved GSA offers a robust and efficient solution for complex optimization challenges.