Orthogonally Initiated Particle Swarm Optimization with Advanced Mutation for Real-Parameter Optimization

TL;DR

This paper presents an advanced particle swarm optimizer called OPSO-m, which uses orthogonal array-based learning and adaptive mutation strategies to improve solution accuracy, convergence speed, and stability in high-dimensional real-parameter optimization problems.

Contribution

It introduces a novel orthogonal array-based initialization and self-adaptive learning strategies with mutation, enhancing PSO's performance on complex optimization tasks.

Findings

OPSO-m outperforms state-of-the-art algorithms in solution precision.

It demonstrates faster convergence and higher search efficiency.

The method maintains robust stability across various problem dimensions.

Abstract

This article introduces an enhanced particle swarm optimizer (PSO), termed Orthogonal PSO with Mutation (OPSO-m). Initially, it proposes an orthogonal array-based learning approach to cultivate an improved initial swarm for PSO, significantly boosting the adaptability of swarm-based optimization algorithms. The article further presents archive-based self-adaptive learning strategies, dividing the population into regular and elite subgroups. Each subgroup employs distinct learning mechanisms. The regular group utilizes efficient learning schemes derived from three unique archives, which categorize individuals based on their quality levels. Additionally, a mutation strategy is implemented to update the positions of elite individuals. Comparative studies are conducted to assess the effectiveness of these learning strategies in OPSO-m, evaluating its optimization capacity through exploration-exploitation dynamics and population diversity analysis. The proposed OPSO-m model is tested on real-parameter challenges from the CEC 2017 suite in 10, 30, 50, and 100-dimensional search spaces, with its results compared to contemporary state-of-the-art algorithms using a sensitivity metric. OPSO-m exhibits distinguished performance in the precision of solutions, rapidity of convergence, efficiency in search, and robust stability, thus highlighting its superior aptitude for resolving intricate optimization issues.