TS-SSA: An improved two-stage sparrow search algorithm for large-scale many-objective optimization problems.

Area of Science:

• Optimization algorithms
• Computational intelligence
• Multi-objective optimization

Background:

• Large-scale many-objective optimization problems (LSMaOPs) present significant challenges due to vast search spaces.
• Existing state-of-the-art methods struggle with comprehensive exploration of these complex problems.

Purpose of the Study:

• To propose an improved Sparrow Search Algorithm (SSA) named Two-Stage Sparrow Search Algorithm (TS-SSA) for effectively solving LSMaOPs.
• To enhance both convergence and diversity management in optimization algorithms for LSMaOPs.

Main Methods:

• TS-SSA employs a two-stage approach: a Many-Objective Sparrow Search Algorithm (MaOSSA) for convergence and a dynamic multi-population strategy for diversity.
• MaOSSA utilizes adaptive population dividing and random bootstrap search strategies.
• The diversity stage incorporates dynamic population dividing and multi-population search strategies.

Main Results:

• TS-SSA demonstrated significant performance and efficiency advantages over 10 state-of-the-art algorithms on DTLZ and LSMOP benchmark problems (3-20 objectives, 300-2000 variables).
• In a real-world application (automatic test scenarios generation), TS-SSA outperformed other algorithms in terms of solution diversity.

Conclusions:

• TS-SSA offers a robust and efficient solution for tackling the complexities of Large-scale Many-Objective Optimization Problems.
• The proposed algorithm shows promise for both theoretical benchmarks and practical applications requiring diverse optimization outcomes.