Details

Autor(en) / Beteiligte

• Syed Asad, Hussain
• Wan, Hang
• Kasun, Hewage
• Rehan, Sadiq
• Huang, Gongsheng

Titel

Distributed real-time optimal control of central air-conditioning systems

Ist Teil von

• Energy and buildings, 2022-02, Vol.256, p.111756, Article 111756

Ort / Verlag

Lausanne: Elsevier B.V

Erscheinungsjahr

2022

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• [Display omitted] •Distributed real-time optimal control was developed for typical HVAC systems.•Physical meaning in decomposition was enhanced in the proposed scheme.•Computation load was significantly distributed from BEMS to local computation unit.•Global convergence of the distributed optimal control was investigated.•Distributed and centralized optimal control led to similar energy performance. Real-time optimal control (RtOpt) is an efficient tool to improve the energy efficiency of heating, ventilating and air-conditioning (HVAC) systems. RtOpt is always centralized and solved entirely in building energy management systems (BEMSs). However, centralized optimization is not be an efficient solution because real-time operation data in local loops need to be sent to the central computation unit in BEMS for optimization. This paper, therefore, proposes a distributed RtOpt scheme for typical centralized HVAC systems, which uses a dual decomposition mechanism to break the centralized problem into one mater problem and several sub-problems that can be solved in parallel and independently. The mater problem is converted to a simple mean value calculation using a projected sub-gradient method, thus distributing the computation load from the central computation unit into local computation units. Unlike available decomposition schemes in literature, the proposed method considered the functionality and spatial distribution of the major components of HVAC systems, which could enhance the physical meaning and facilitate the application of the decomposition scheme. The proposed distributed RtOpt scheme was tested in case studies, with numerical analysis being used to demonstrate its potential benefits.