Research On Multi-Objective Optimization of Complex Systems Based on A Feedback-Driven Structure-Preserving Optimization Architecture and An Improved NSGA-II Algorithm

Ruipeng Tan; Jiabi Chen

doi:10.54097/nbtxqt72

Authors

Ruipeng Tan
Jiabi Chen

DOI:

https://doi.org/10.54097/nbtxqt72

Keywords:

Optimization, Multi-objective Optimization, NSGA-II, Feedback Mechanism, Complex Systems.

Abstract

The complex systems typically exhibit multiple objectives, multiple constraints, and dynamic structural characteristics, which make their multi-objective optimization and structural adaptability problems highly challenging. To address these challenges, this paper proposes a Structure-Preserving Optimization Architecture (SPOA) to systematically describe the nonlinear interaction process between input stimuli, constraint boundaries, and response structures. In terms of algorithm implementation, SPOA employs an improved NSGA-II algorithm to approximate the Pareto optimal solution set in a non-convex and constrained high-dimensional decision space. Finally, simulation experiments are conducted using a tourism system as an example scenario, validating the effectiveness of the proposed model. The results show that the model demonstrates excellent performance in terms of convergence stability, parameter transferability, and robustness to structural perturbations, and exhibits broad applicability in complex systems.

Downloads

Download data is not yet available.

References

[1] Deb K, Pratap A, Agarwal S, et al. A fast and elitist multiobjective genetic algorithm: NSGA-II [J]. IEEE Transactions on Evolutionary Computation, 2002, 6(2): 182-197.

[2] Li X, Zhang Y, Wang J. Deep learning-enhanced multi-objective optimization for large-scale scheduling [J]. IEEE Transactions on Industrial Informatics, 2021, 17(5): 3210-3219.

[3] Wang J, Li H, Zhang Q. Multi-objective optimization for smart grid energy management [J]. Applied Energy, 2022, 305: 117849.

[4] Chen Y, Liu Z, Wang X. Dynamic multi-objective optimization for autonomous vehicle control [J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(3): 2456-2467.

[5] Li X, Xie L, Li X, et al. A Retrofit Hierarchical Architecture for Real-Time Optimization and Control Integration [J]. Processes, 2020, 8(2): 181.

[6] Liu H, Zhang X, Chen Y. Reinforcement learning for bilevel optimization in industrial systems [J]. IEEE Transactions on Automation Science and Engineering, 2020, 17(4): 1987-1998.

[7] Poonia V, Kulshrestha R, Sangwan K S. a Comparative Study of ε-constraint, LP-metric, and Weighted Sum Multi-Objective Optimization Methods in a Circular Economy [J]. Procedia CIRP, 2024, 122: 294-299.

[8] Mesquita-Cunha M, Figueira J R, Barbosa-Póvoa A P. New ε-constraint methods for multi-objective integer linear programming: A Pareto front representation approach [J]. European Journal of Operational Research, 2023, 306(1): 286-307.

[9] Zhang Q, Yang L, Li H. Adaptive multi-objective evolutionary algorithms with reinforcement learning [J]. IEEE Transactions on Evolutionary Computation, 2024, 28(2): 345-357.

[10] Kumar A, Deb K. Multi-objective scheduling using evolutionary algorithms [J]. Journal of the Operational Research Society, 2021, 72(6): 1345-1356.

[11] Xu Z, Li Y, Zhang W. NSGA-II for QoS-aware network resource allocation [J]. Computer Networks, 2023, 220: 109487.

[12] Deb K, Jain H. An evolutionary many-objective optimization algorithm using reference-point-based nondominated sorting approach [J]. IEEE Transactions on Evolutionary Computation, 2014, 18(4): 577-601.

[13] Zhou L, Zhang H, Li X. Transfer learning for dynamic multi-objective optimization [J]. Applied Soft Computing, 2022, 125: 108976.

[14] Wang M, Zhou Y, Li J. Generalized frameworks for multi-objective optimization in dynamic systems [J]. IEEE Transactions on Cybernetics, 2024, 54(3): 1567-1579.