The Great Wall Construction Algorithm Incorporating Marginal Benefit Assessment for Numerical Optimization and Real-World Applications

Complex global optimization problems are widely encountered in wireless sensor network deployment, engineering design, energy scheduling, structural optimization, and intelligent parameter optimization. Due to their gradient-free search mechanism, strong global exploration capability, simple implementation, and adaptability to nonlinear and multimodal optimization problems, swarm intelligence algorithms have become effective tools for solving complex optimization tasks. In this study, an enhanced Great Wall Construction Algorithm based on a Marginal Benefit Assessment mechanism, termed GWCA-MBA, is proposed for numerical optimization and engineering applications. The proposed method incorporates three cooperative enhancement strategies into the original GWCA framework, including a quasi-oppositional chaotic initialization strategy, a Marginal Benefit Assessment-based adaptive role scheduling mechanism, and an elite differential-Lévy refinement strategy. These mechanisms jointly improve population diversity, adaptive search capability, and local exploitation precision. To evaluate its effectiveness, GWCA-MBA is comprehensively tested on the CEC2017, CEC2020, and CEC2022 benchmark suites and compared with several representative metaheuristic algorithms. Experimental results demonstrate that GWCA-MBA achieves the best Friedman mean ranks of 2.70, 3.50, and 2.33 on the CEC2017, CEC2020, and CEC2022 benchmark suites, respectively, indicating superior overall optimization performance and robustness. In addition, the proposed algorithm exhibits highly stable convergence behavior with significantly lower standard deviation values than the compared algorithms on most benchmark functions. To further verify its engineering applicability, GWCA-MBA is applied to a three-dimensional wireless sensor network coverage optimization problem and two constrained engineering design problems. In the wireless sensor network deployment problem, GWCA-MBA achieves a coverage rate of 75.90%, outperforming the compared algorithms and demonstrating strong spatial optimization capability and practical applicability. Moreover, GWCA-MBA also achieves the first Friedman ranking in the engineering application experiments, further demonstrating its strong practical optimization capability and generalization performance in real-world optimization problems. Overall, GWCA-MBA effectively balances exploration and exploitation and provides a competitive optimization framework for complex numerical optimization and practical engineering applications.