.png){kind=icon}
.png){kind=icon}
Return
Multi-strategy improved grey wolf optimizer algorithm and its application in engineering optimization problems
.png){kind=icon}
PRE
AI
DOI：10.1007/s10586-026-06009-9.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Effective magnetic target recognition is challenged by the high dimensionality of features and limited sample data. This study aims to develop a robust optimizer to address this contradiction. Multi-Strategy Improved Grey Wolf Optimizer (MSIGWO) was proposed, which integrates an adaptive dimension learning strategy for intelligent feature assessment, a multi-population fusion evolution mechanism for enhanced global search, and an adaptive Lévy flight strategy to balance exploration and exploitation. The algorithm was validated through a three-tier experimental framework. (1) On CEC 2020/2022 benchmarks, MSIGWO achieved first rank on average across 10D to 100D problems, with statistically significant superiority (Friedman test p < 0.05). (2) On a 300-sample simulation dataset, it attained 96.47% average classification accuracy, reduced feature dimensions by 78.4%, and maintained stability under varying noise. (3) On a 44-sample real-world dataset, it improved recognition accuracy from 64.41% to 73.28% via ten-fold cross-validation, reduced features from 116 to 25, and outperformed 11 comparative algorithms (e.g., PSO, GWO). The MSIGWO algorithm demonstrates superior convergence accuracy, feature selection capability, and stability. It provides an effective innovative solution for high-dimensional small-sample problems in magnetic target recognition and complex signal processing.
Keywords:
Multi-strategy grey wolf optimization
High-dimensional small-sample feature selection
Magnetic target field recognition
Hybrid feature selection framework
Multi-level experimental validation