Share

Deep learning-based refrigerant charge fault detection method of air-source heat pump system.

Number: No 096

Author(s) : EOM Y. H., HONG S. B., YOO J. W., KIM M. S.

Summary

Energy demands grow every year, and a significant amount of energy consumption is used for buildings cooling and heating. Since heat pump systems have high efficiency and can be utilized for buildings cooling and heating, they are commonly used around the world. Many studies show heat pumps have the best COP at the optimal refrigerant charge. Therefore, it is imperative to monitor the current refrigerant charge of the system and maintain it optimally in view of energy saving. However, some researches show that many heat pumps in the field have refrigerant leakage fault or overcharge fault. The refrigerant charge error can cause energy waste and thermal discomfort. Hence, many researchers have conducted studies for refrigerant charge fault detection (RCFD) method. In recent years, RCFD methods based on deep learning technology have been developed actively. This paper suggests a novel and efficient RCFD strategy using convolutional neural network (CNN). The CNN based multiple outputs regression model shows excellent results for predicting power consumption, cooling capacity (heating capacity), and the refrigerant charge amount simultaneously with a single model. 

Available documents

Format PDF

Pages: 7 p.

Available

Free

Download

Details

  • Original title: Deep learning-based refrigerant charge fault detection method of air-source heat pump system.
  • Record ID : 30029978
  • Languages: English
  • Subject: Technology
  • Source: 13th IEA Heat Pump Conference 2021: Heat Pumps – Mission for the Green World. Conference proceedings [full papers]
  • Publication date: 2021/08/31

Links


See other articles from the proceedings (198)
See the conference proceedings

Indexing