Political optimizer with interpolation strategy for global optimization.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• Political Optimizer (PO) is a meta-heuristic for global optimization but suffers from local optima stagnation and low convergence accuracy.
• This is due to a greedy election strategy and imbalanced exploration/exploitation during party switching.

Purpose of the Study:

• To address the limitations of the standard Political Optimizer (PO).
• To propose novel PO variants with improved global exploration and exploitation balance.
• To enhance convergence accuracy and escape local optima.

Main Methods:

• Integrated various interpolation strategies (Quadratic, Cubic, Lagrange, Newton) with PO.
• Incorporated Refraction Learning (RL) to boost population diversity.
• Developed a logistic model to balance exploration and exploitation during the party switching stage.

Main Results:

• The novel PO variants, particularly those combining interpolation and RL, demonstrated superior performance.
• Experimental results on benchmark and IEEE CEC 2014 functions confirmed enhanced exploration capacity.
• The proposed methods effectively helped algorithms escape local optima and improved convergence accuracy.

Conclusions:

• The integration of interpolation strategies and Refraction Learning significantly improves the Political Optimizer's performance.
• The proposed logistic model effectively balances exploration and exploitation.
• This novel approach offers a promising solution for global optimization problems, outperforming existing methods.