Improved Information Acquisition Optimization via a Functionally Coupled Closed-loop Framework for Complex Engineering Problems

Modern engineering optimization problems pose significant challenges due to high dimensionality, nonlinearity, multimodality, and complex constraints. To address these challenges, this paper proposes an Improved Information Acquisition Optimization (IIAO) algorithm based on a functionally coupled closed-loop optimization framework. Unlike conventional hybrid metaheuristics that combine multiple operators independently, IIAO models the optimization process as a sequential state-transition cycle with cooperative interactions among adaptive global exploration, competitive learning, and statistical refinement. Specifically, an adaptive Lévy-guided exploration strategy enhances population diversity and directional global search capability. A fitness-driven competitive learning mechanism dynamically regulates population evolution to balance exploration and exploitation. In addition, a tangent-guided statistical refinement strategy performs corrective local exploitation using historical population information, while refined elite solutions are fed back to guide subsequent exploration. Through this cooperative feedback mechanism, IIAO effectively alleviates excessive randomness, convergence instability, and cumulative exploitation errors in the original IAO. The proposed algorithm is evaluated on the CEC2017 benchmark suite with dimensions ranging from 10 to 100 and four representative real-world optimization problems. Experimental results demonstrate that IIAO achieves competitive convergence accuracy, robustness, and scalability compared with ten state-of-the-art algorithms, which is further supported by Friedman and Wilcoxon statistical tests. Component-wise analyses further verify the effectiveness of the proposed cooperative mechanisms. Additional studies on image segmentation and engineering applications further verify the effectiveness and generalization capability of the proposed framework.