Objective The widespread application of complex systems in natural and engineering domains has led to increasingly prominent management challenges due to their inherent characteristics, such as nonlinearity and emergence. Traditional simplified analytical methods struggle to address dynamic uncertainties while existing data-driven algorithms face dual challenges: a lack of explainability and insufficient robustness.

Methods To address these challenges, this paper proposes the development of dual-brain synergistic digital-intelligent agents based on the concepts defined respectively for the logical brain and associative brain, aiming to provide solutions that integrate intelligence and reliability for the construction of intelligent complex industrial systems. The essence of agents, which are rooted in digitalization and designed for intelligent decision-making and control, was introduced. Then the paper elaborates on the logical brain and associative brain from a conceptual perspective: the logical brain operates based on logical rules, while the associative brain is designed for association extraction. Together, the core modules of digital-intelligent agents were formed. Finally, the advantages of dual-brain synergy were emphasized and three fusion methods for modeling were proposed , aiming to enhance the effectiveness of solutions for constructing intelligent complex systems.

Results Driven by the global energy transition and the “dual carbon” goals, oil-gas gathering and transportation systems are evolving toward intelligence at a faster pace. However, technical challenges exists such as complex pipeline networks, multi-source interference, and full-lifecycle management. Currently, the construction of intelligent systems, both in China and abroad, remains in the initial stages of digital transformation. There is an urgent need to overcome bottlenecks in areas such as algorithmic fusion, dynamic data sharing, and deep AI integration to enable a leap from localized optimization to system-wide intelligent decision-making. The effectiveness of the logical brain and associative brain synergy mechanism in the construction of intelligent oil-gas gathering and transportation systems was validated through three case studies: virtual metering of oil wells, production optimization for an oil-gas field, and fault attribution analysis for a transfer station.

Conclusion The dual-brain synergy mechanism is based on the integration of mechanisms and data, which ensures model accuracy while enhancing prediction speed, computational efficiency, and fault diagnosis accuracy. Case studies demonstrate significant improvements in real-time monitoring, decision-making optimization, and fault response capabilities, effectively validating the proposed approach for complex systems. Future research should focus on overcoming bottlenecks in multimodal data fusion with small sample sizes and addressing challenges in cross-scale modeling and intelligent collaboration techniques to support low-carbon transitions. Additionally, studies are expected to develop a mechanism-driven dynamic reinforcement learning framework, advancing system evolution toward full-lifecycle management and multi-energy collaborative optimization. These efforts aim to provide core support for green energy transformation.