Seasonal heating performance prediction of air-to-water heat pumps based on short-term dynamic monitoring

Air-to-water heat pumps (AWHPs) utilize renewable energy and have found worldwide applications, with the seasonal coefficient of performance (SCOP) as a key index. Considering reliability and costs, short-term dynamic monitoring combined with regression analysis and extrapolation is used for predicting SCOP. Different regression models are being researched. A statistical analysis method is proposed to work out the optimal scheme. A series of prediction models with different independent variables, fitting methods and training dataset acquisition methods are discussed. Two indicators, the qualification rate R +/- 10%and the maximum relative error E-max, are proposed for accuracy evaluation. For analysis a typical AWHP heating system in Beijing was monitored for 78d. Linear fitting performs better than quadratic polynomial fitting. For the consecutive-day method, the prediction deviation decreases with a longer test time and presents diminishing marginal benefits. A critical value of 10-day is identified and unrepresentative days should be avoided. For the typical-meteorological-day method, three days with an outdoor air temperature (T-out) range covering over 50% days of the heating season and including the average Tout of local winter are recommended. Satisfactory prediction results are realized, with R +/- 10%>97% and E-max<14%, while using Tout or the difference between water temperature and T-out presents consistent accuracy. (c) 2021 Elsevier Ltd. All rights reserved.