KEAI: An Adaptive Dynamic Programming and Iterative Optimization Model for Event-Triggered Control in Complex Networks

Abstract

Achieving optimal control of Complex Networks is crucial for optimizing network structures and system solutions. However, current methods mainly suffer from high computational costs and poor performance. Therefore, this study proposes an optimal control model based on Adaptive Dynamic Programming and Iterative Algorithm to address these issues. The model uses Global Dual Heuristic Programming as the foundation, value iterationas the core optimization technique, and integrates event-triggered mechanism and K-means clustering algorithm. The results show that, compared to modelsbased on theHarris Eagle, Arithmetic, and Northern Goshawll optimization algorithms, the proposedmethod reduces physical indicators (e.g., floating-point operations) of the target network by 15%, ensuring accuracy, while achievinga lighter networkand faster convergence. In practical tests, the model is used to optimize the YOLOv8 network. The verification using the CIFAR-10 dataset showed that the YOLOv8’s accuracy improved by 10.2%, with response time reducedby 15ms and energy consumption for single-image recognition cut by 31mJ. These results showthat the proposed model effectively achieves optimal control of complex networks, addresses issue like slow speed and high consumption,