A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Smart-valve-assisted model-free predictive control system for chiller plants
Highlights Artificial intelligence application on energy-efficient chiller control. The practical application requires complex pre-work and a lot of training data. This study developed SV-MFPC to solve this application problem. Three buildings with different cooling loads were used as test samples. Up to 34.2% of chiller energy saving at a factory was achieved by SV-MFPC.
Abstract Implementing artificial intelligence (AI) for heating, ventilation and air conditioning (HVAC) energy-saving control requires complex pre-work or considerable training data, which are obstacles to practical application. This study developed smart-valve-assisted model-free predictive control to solve this application problem. Based on reinforce learning, model-free predictive control is recommended for situations in which no data can be provided for learning. To go further, smart valve hardware was used in the chiller plant to perform trial-and-error control of each air-conditioned space. By using smart valves, which can predict flow rates and exert control, control system can regulate and control lower stream for fan coils or air handling units in the chiller plant. The smart valves also record starting parameters to be used in later trial-and-error control and are key hardware for executing the model-free concept. This control engaged in interactive communication through the agent control program developed in this study. The overall control system is referred to simply as SV-MFPC. Energy-saving performances of SV-MFPC were tested in three different kinds of buildings include a hospital with a variable cooling load, an office building with a stable cooling load and a factory with a 24-hour operation load. The energy-saving effect was approximately 30% across these test sites. More important, the pre-work is simple, and the data required are only a thousandth of that required by other artificial intelligence technologies. This overcomes the technology gap and is the key to practical application.
Smart-valve-assisted model-free predictive control system for chiller plants
Highlights Artificial intelligence application on energy-efficient chiller control. The practical application requires complex pre-work and a lot of training data. This study developed SV-MFPC to solve this application problem. Three buildings with different cooling loads were used as test samples. Up to 34.2% of chiller energy saving at a factory was achieved by SV-MFPC.
Abstract Implementing artificial intelligence (AI) for heating, ventilation and air conditioning (HVAC) energy-saving control requires complex pre-work or considerable training data, which are obstacles to practical application. This study developed smart-valve-assisted model-free predictive control to solve this application problem. Based on reinforce learning, model-free predictive control is recommended for situations in which no data can be provided for learning. To go further, smart valve hardware was used in the chiller plant to perform trial-and-error control of each air-conditioned space. By using smart valves, which can predict flow rates and exert control, control system can regulate and control lower stream for fan coils or air handling units in the chiller plant. The smart valves also record starting parameters to be used in later trial-and-error control and are key hardware for executing the model-free concept. This control engaged in interactive communication through the agent control program developed in this study. The overall control system is referred to simply as SV-MFPC. Energy-saving performances of SV-MFPC were tested in three different kinds of buildings include a hospital with a variable cooling load, an office building with a stable cooling load and a factory with a 24-hour operation load. The energy-saving effect was approximately 30% across these test sites. More important, the pre-work is simple, and the data required are only a thousandth of that required by other artificial intelligence technologies. This overcomes the technology gap and is the key to practical application.
Smart-valve-assisted model-free predictive control system for chiller plants
Lee, Dasheng (author) / Lin, Chien-Jung (author) / Lai, Chih-Wei (author) / Huang, Tsai (author)
Energy and Buildings ; 234
2020-12-29
Article (Journal)
Electronic Resource
English
British Library Online Contents | 1994
All-Variable Speed Centrifugal Chiller Plants
| British Library Online Contents | 2014
Chicago gets two district chiller plants
Online Contents | 1995