A Statistical Study on Parameter Selection of Operators in Continuous State Transition Algorithm.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristic Computing

Background:

• The state transition algorithm (STA) is a novel metaheuristic for global optimization.
• Previous continuous STA versions used constant or periodically decreasing parameters.
• Operator parameter selection in continuous STA requires further investigation.

Purpose of the Study:

• To investigate the impact of operator parameters on the search ability of continuous STA.
• To propose a new continuous STA with an optimal parameter strategy.
• To accelerate the search process of the STA.

Main Methods:

• Statistical analysis of four benchmark 2-D functions to understand parameter effects.
• Development of a continuous STA incorporating an optimal parameter selection strategy.
• Application and testing of the proposed STA on 12 benchmarks across 20-D, 30-D, and 50-D spaces.

Main Results:

• The statistical study revealed how parameters influence STA's search performance.
• The proposed STA with optimal parameter strategy demonstrated accelerated search capabilities.
• The method was successfully applied to high-dimensional benchmark problems.

Conclusions:

• Optimal parameter selection is crucial for enhancing STA's efficiency.
• The proposed STA with optimal parameter strategy offers improved performance.
• This approach effectively addresses complex global optimization challenges.