Scholarlink

<p>Multi-objective optimization algorithms are commonly used in engineering, but many face issues like premature convergence, poor diversity, and an unbalanced exploration-exploitation trade-off. Algorithms like Multi-objective Artificial Bee Colony (MOABC) and NSGA-II have limitations in constrained multi-objective optimization problems (CMOPs). This study presents a CMOP framework using a hybrid Modified Artificial Bee Colony–NSGA-II (MOABC–NSGA-II) algorithm to tackle these challenges. The hybrid method combines the adaptive exploration-exploitation of MOABC with the elitist non-dominated sorting and diversity preservation of NSGA-II. The Superiority of Feasible (SF) constraint-handling strategy ensures strict adherence to system limits, while crowding-distance-guided Pareto archiving improves convergence accuracy and maintains solution diversity. A TOPSIS decision module identifies the Best Compromise Solution (BCS) from the final non-dominated archive. The methodology is thoroughly assessed on seventeen benchmark functions, including ZDT, DTLZ, and CEC-2009 test suites, using common performance indicators like Inverted Generational Distance (IGD), Generational Distance (GD), Spacing (S), and Maximum Spread (MS). The comparative analysis with state-of-the-art algorithms—MOPSO, MOABC, NSGA-II, NSGA-III, MOEA/D, and MODE—shows that the proposed hybrid scheme consistently delivers better convergence, solution diversity, and robustness across various problem classes. Statistical tests like the Wilcoxon signed-rank test, Friedman ranking, and critical difference analysis confirm the significance of these improvements. The framework is applied to two engineering case studies for real-world applicability: (1) Renewable-integrated Multi-objective Optimal Power Flow (MOOPF) and (2) Transformer constrained multi-objective characteristic optimization. The results show that the hybrid framework effectively addresses the non-linear, non-convex, and stochastic aspects of power-system optimization problems, proving to be scalable and practical.</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

Performance Assessment of Hybrid Modified Artificial Bee Colony–NSGA-II Algorithm for Constrained Multi-objective Optimization: Benchmark Validation and Real-World Power System Problems

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • Abhishek Bajirao Katkar,
  • Vishal Tukaram Metkari,
  • Himmat Tukaram Jadhav

摘要

Multi-objective optimization algorithms are commonly used in engineering, but many face issues like premature convergence, poor diversity, and an unbalanced exploration-exploitation trade-off. Algorithms like Multi-objective Artificial Bee Colony (MOABC) and NSGA-II have limitations in constrained multi-objective optimization problems (CMOPs). This study presents a CMOP framework using a hybrid Modified Artificial Bee Colony–NSGA-II (MOABC–NSGA-II) algorithm to tackle these challenges. The hybrid method combines the adaptive exploration-exploitation of MOABC with the elitist non-dominated sorting and diversity preservation of NSGA-II. The Superiority of Feasible (SF) constraint-handling strategy ensures strict adherence to system limits, while crowding-distance-guided Pareto archiving improves convergence accuracy and maintains solution diversity. A TOPSIS decision module identifies the Best Compromise Solution (BCS) from the final non-dominated archive. The methodology is thoroughly assessed on seventeen benchmark functions, including ZDT, DTLZ, and CEC-2009 test suites, using common performance indicators like Inverted Generational Distance (IGD), Generational Distance (GD), Spacing (S), and Maximum Spread (MS). The comparative analysis with state-of-the-art algorithms—MOPSO, MOABC, NSGA-II, NSGA-III, MOEA/D, and MODE—shows that the proposed hybrid scheme consistently delivers better convergence, solution diversity, and robustness across various problem classes. Statistical tests like the Wilcoxon signed-rank test, Friedman ranking, and critical difference analysis confirm the significance of these improvements. The framework is applied to two engineering case studies for real-world applicability: (1) Renewable-integrated Multi-objective Optimal Power Flow (MOOPF) and (2) Transformer constrained multi-objective characteristic optimization. The results show that the hybrid framework effectively addresses the non-linear, non-convex, and stochastic aspects of power-system optimization problems, proving to be scalable and practical.