Details

Autor(en) / Beteiligte

• Wang, Zhaojun
• Zhu, Yingxin
• Wang, Fang
• Wang, Peng
• Shen, Chao
• Liu, Jing

Titel

Multivariable Linear Regression Model for Online Predictive Control of a Typical Chiller Plant System

Ist Teil von

• Proceedings of the 11th International Symposium on Heating, Ventilation and Air Conditioning (ISHVAC 2019), 2020, p.241-249

Ort / Verlag

Singapore: Springer Singapore Pte. Limited

Erscheinungsjahr

2020

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• Since much attention has been paid to the accuracy of prediction models in chiller plant, the practicality and feasibility of models are often compromised for superior performance. In this paper, a data-driven model and online predictive control strategy are presented for a typical chiller plant to optimize its overall performance and energy consumption from the perspective of real-time application. The model predictive controller is developed based on the multivariable linear regression (MLR) model. Meanwhile, the ordinary least squares (OLS) estimation technique was adopted to identified and update the model coefficients. The input variables of the developed model are the temperature difference of chilled water and the temperature difference of cooling water, which are much easier to obtain in a practical system compared to variables such as water flow and part load ratio, thereby rendering an excellent model capable of easy implementation and duplication. The MLR model-based online predictive control strategy is tested and evaluated in a real chiller plant system. The results of application indicate that the online predictive control strategy can enhance the global COP values by 6.52% on average and reduce electricity consumption by 2.45% daily compared to the local control strategy.