Scholarlink

<p>The Educational Competition Optimizer (ECO) is a novel human-based optimization algorithm inspired by educational competition phenomena in society. Despite its strong performance across various test sets and optimization problems, ECO encounters challenges in high-dimensional and complex problem spaces. This paper introduces the Improved Educational Competition Optimizer (IECO), an advanced variant of ECO designed to address these challenges. IECO incorporates a jumping strategy, an exponential logarithmic adaptation strategy, and an early stopping strategy. The jumping strategy enhances exploration by adjusting the step size during preliminary searches. The exponential logarithmic adaptation strategy fine-tunes parameters such as the school ratio (G), student patience (P), student talent (E), and adaptive operator (w). The early stopping strategy aids in escaping local optima. These enhancements significantly boost IECO’s exploration and exploitation capabilities, leading to superior convergence and efficiency in handling high-dimensional problems. We evaluate IECO on classical functions, CEC 2017 and CEC 2022 test functions, 2 engineering hyperparameter optimization problems, 10 real-world engineering design problems, and 2 multi-UAV path planning problems. Comparative analyses include (1) ablation algorithms: ECO with only the jumping strategy (JECO), ECO with only the exponential logarithmic adaptation strategy (ELAECO), ECO without early stopping strategy (ECO-), and the original ECO; (2) recently developed algorithms: PO, DBO, and AOA; (3) widely cited algorithms: HHO, WOA, and SCA; and (4) CEC competition-winning algorithms: iLSHADEϵ, sCMAgES, COLSHADE, and EnMODE. Experimental results, supported by 95% confidence intervals and five performance metrics, demonstrate IECO’s superior performance in solving complex and high-dimensional engineering optimization problems. These findings position IECO as a new state-of-the-art optimizer in single-objective optimization. The source code for IECO can be accessed at <a href="https://github.com/junbolian/IECO">https://github.com/junbolian/IECO</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

IECO: an improved educational competition optimizer for state-of-the-art engineering optimization

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月19日

查看全文
  • Xiaojie Tang,
  • Junbo Jacob Lian,
  • Ling Ma,
  • Xincan Wu,
  • Rui Zhong,
  • Yujun Zhang,
  • Huiling Chen

摘要

The Educational Competition Optimizer (ECO) is a novel human-based optimization algorithm inspired by educational competition phenomena in society. Despite its strong performance across various test sets and optimization problems, ECO encounters challenges in high-dimensional and complex problem spaces. This paper introduces the Improved Educational Competition Optimizer (IECO), an advanced variant of ECO designed to address these challenges. IECO incorporates a jumping strategy, an exponential logarithmic adaptation strategy, and an early stopping strategy. The jumping strategy enhances exploration by adjusting the step size during preliminary searches. The exponential logarithmic adaptation strategy fine-tunes parameters such as the school ratio (G), student patience (P), student talent (E), and adaptive operator (w). The early stopping strategy aids in escaping local optima. These enhancements significantly boost IECO’s exploration and exploitation capabilities, leading to superior convergence and efficiency in handling high-dimensional problems. We evaluate IECO on classical functions, CEC 2017 and CEC 2022 test functions, 2 engineering hyperparameter optimization problems, 10 real-world engineering design problems, and 2 multi-UAV path planning problems. Comparative analyses include (1) ablation algorithms: ECO with only the jumping strategy (JECO), ECO with only the exponential logarithmic adaptation strategy (ELAECO), ECO without early stopping strategy (ECO-), and the original ECO; (2) recently developed algorithms: PO, DBO, and AOA; (3) widely cited algorithms: HHO, WOA, and SCA; and (4) CEC competition-winning algorithms: iLSHADEϵ, sCMAgES, COLSHADE, and EnMODE. Experimental results, supported by 95% confidence intervals and five performance metrics, demonstrate IECO’s superior performance in solving complex and high-dimensional engineering optimization problems. These findings position IECO as a new state-of-the-art optimizer in single-objective optimization. The source code for IECO can be accessed at https://github.com/junbolian/IECO.