A Data Acquisition Method Based Collaborative Information Model for Heterogeneous Terminal Device Groups in Converter Stations

To address the challenges of diverse protocols for multiple heterogeneous devices and high requirements for real-time performance and reliability in converter station terminals, this paper proposes a dynamic data acquisition method based on multi-agent collaboration and model-driven approaches. Within a three-tier architecture consisting of terminal agents, edge coordination gateway agents, and a central dispatch platform, we deploy lightweight performance prediction models and global optimization modules. The method employs distributed convex optimization (ADMM) and multi-agent deep reinforcement learning algorithms to collaboratively adjust sampling frequencies and compression ratios in real-time. Enhanced by IEEE 1588 PTP time synchronization, TLS mutual authentication, and multi-level fault tolerance mechanisms, the solution ensures communication security and high availability. Simulation results demonstrate that compared with traditional fixed-frequency acquisition, the proposed method improves bandwidth utilization by approximately 25%, reduces end-to-end latency by 30%, and maintains ≥99.8% data integrity rate under network jitter and node failure scenarios. These findings validate its efficiency, reliability, and scalability, providing robust technical support for converter station monitoring and big data analytics.