Energy consumption predicting model of VRV (Variable refrigerant volume) system in office buildings based on data mining

Energy consumption prediction plays an important role in building design & retrofit, energy management system. In this paper, ECI (Energy consumption intensity) of VRV (Variable refrigerant volume) system in office buildings located in Shanghai, Nanjing and Changsha of East China are firstly investigated and analyzed statistically. The annual median value of ECI in the three cities is about 36 kWh/m2.Hourly energy consumption prediction model based on data mining (ie. ANN (Artificial Neural Network), SVM (Support vector regression) and ARIMA (Autoregressive integrated moving average) models) are subsequently discussed. During case study, three months' (ie. from July to September in 2013) energy consumption data of office building located in Shanghai are used to set up different predicting models. RMSE (Root mean square error), MSE (Mean square error) & MAPE (Mean absolute percentage error) are used to evaluate corresponding prediction accuracy. Results show that Predicting model based on ANN is better than the other two models'. The RMSE, MSE & MAPE of ANN model in training course are 0.0681, 0.0045, 0.1710, respectively. According to simulated results in paper, ANN and SVM models are recommended to do energy consumption predicting of VRV system in office buildings.