Support vector regression modeling of the performance of an R1234yf automotive air conditioning system


HOŞÖZ M., KAPLAN K., Aral M. C., SUHERMANTO M., ERTUNÇ H. M.

5th International Conference on Energy and Environment Research (ICEER), Prague, Czech Republic, 23 - 27 July 2018, vol.153, pp.309-314 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 153
  • Doi Number: 10.1016/j.egypro.2018.10.067
  • City: Prague
  • Country: Czech Republic
  • Page Numbers: pp.309-314
  • Keywords: Air conditioning, R1234yf, support vector regression, SVR, AAC, ARTIFICIAL NEURAL-NETWORKS
  • Kocaeli University Affiliated: Yes

Abstract

This study aims at modelling various performance parameters of an automotive air conditioning (AAC) system using support vector regression (SVR), a novel soft modelling technique. For this purpose, a bench-top AAC system was set up, charged with alternative refrigerant R1234yf, and tested in a wide range of operating conditions. Next, the cooling capacity and coefficient of performance of the AAC system were evaluated. Then, the proposed SVR was trained by using some of the input-output data pairs, and the performance of model predictions was tested using the remaining data. It was determined that the SVR model yielded very accurate predictions. (C) 2018 The Authors. Published by Elsevier Ltd.