Regression method based fault detection of the VRF system for energy saving

A refrigerant charge fault detection strategy was proposed using regression method. First, characteristic variable parameters that are sensitive to the refrigerant charge level of variable refrigerant flow (VRF) system were select. Then, regression models of refrigerant charge amount fault detection were established by using multiple linear regression and support vector regression respectively. The training process of the regression model only needs the normal refrigerant charge operation data, which does not need the complete refrigerant charge level experimental data. Refrigerant charge fault experimental data which contains refrigerant undercharge, overcharge and leakage were used to test the developed fault detection models. Results show that the detection performance of the support vector regression model is better than the multiple linear regression model. In addition, for the refrigerant leakage fault, the support vector regression model can detect the leakage fault after 3 minutes of refrigerant leakage.