A fully distributed optimal control approach for multi-zone dedicated outdoor air systems to be implemented in IoT-enabled building automation networks. (15th February 2022)

• Record Type: Journal Article
• Authors: Li, Wenzhuo; Wang, Shengwei
• Is Part Of: Applied energy. Volume 308(2022)
• Publication Date: 2022-02-15
• Subjects: Distributed optimal control -- Multi-agent system -- Edge computing -- HVAC system -- IoT -- Communication topology; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2021.118408
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: A fully distributed optimal control approach is proposed for DOASs. Agents coordinate with connected peers directly without a coordinating agent. Impacts of the adopted communication topologies are investigated. It performs the best when adopting the fully connected topology. It is fully compared with centralized and hierarchical distributed approaches. Abstract: For heating, ventilation and air-conditioning (HVAC) systems, centralized optimal control approaches are widely investigated, while hierarchical distributed optimal control approaches are of increasing attention. Using both approaches, the central station and the coordinating agent play critical roles, resulting in low robustness. A novel fully distributed approach may offer higher robustness, but has not yet been investigated for HVAC systems. This paper therefore proposes a fully distributed optimal control approach for multi-zone dedicated outdoor air systems (DOASs) to be implemented in IoT-enabled building automation networks. Without the coordinating agent, information is exchanged directly between connected agents. The optimal solutions are found by coordinating these multiple agents. During iterations, the outdoor air volume of individual rooms and the PAU are optimized locally using the Incremental Cost Consensus (ICC) algorithm, and the outdoor air volume mismatch is estimated locally using the average consensus algorithm. Tests are conducted to validate the proposed approach by comparing it with existing approaches. The impacts of communication topology on the performance of the proposed approach are investigated. Results show that the proposed fully distributed optimal control approach with the fully connected topology performs better than the existing hierarchical distributed approach. Among different communication topologies, the proposed approach with the fully connected topology had the highest robustness, lowest computation complexity and highest optimization efficiency. It also guaranteed the best control performance when deployed over physical platforms (e.g. IoT-based smart sensors of limited capacity), which limit the maximum iteration number.