An improved polar lights optimization algorithm for global optimization and engineering applications.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• Existing Polar Lights Optimization (PLO) methods suffer from limited population diversity, slow convergence, and imbalanced exploration-exploitation.
• Addressing these limitations is crucial for improving the performance of optimization algorithms in complex problem-solving.

Purpose of the Study:

• To introduce an enhanced, high-caliber Population Evolution Polar Lights Optimization (IPLO) algorithm.
• To overcome the deficiencies of the standard PLO algorithm, focusing on diversity, convergence speed, and search balance.

Main Methods:

• Population initialization using a pseudo-random lens SPM chaos initialization (PRLS-CI) strategy.
• Reinforcement learning with adaptive dynamics and an exploration-focused reward loss function for balancing global and local search.
• Adaptive t-distribution mutation and the simplex method to enhance diversity and search path efficiency.

Main Results:

• IPLO demonstrated superior performance over existing algorithms on benchmark functions (CEC2017, CEC 2019, CEC 2022) and engineering design problems.
• Achieved significant improvements compared to PLO: 66.7% increase in convergence accuracy, 69.6% increase in convergence speed, and 99.9% enhancement in stability.
• Validated the algorithm's efficacy in tackling complex real-world optimization tasks.

Conclusions:

• The proposed IPLO algorithm effectively enhances population diversity, accelerates convergence, and improves the balance between exploration and exploitation.
• IPLO offers a robust and efficient solution for complex optimization challenges, outperforming traditional methods.