Estimating the hourly electricity profile of Japanese households - Coupling of engineering and statistical methods

Understanding the hourly electricity profile and the electricity consumption by each appliance is essential for encouraging energy-saving measures in the household sector. There are two methods for identifying energy consumption for households in existing studies: the engineering and the statistical methods. Both methods have strengths and limitations. In this study, we developed a hybrid method based on the statistical method by combining following three steps using knowledge of the engineering method; externalizing the electricity consumption for the refrigerator, adding the number of at-home-and-awake members as explanatory variables, and restricting appliance usage hours. The proposed hybrid method could adequately reproduce the total hourly electricity consumption and seasonal variation compared to the engineering method, and could decompose major appliances, some of which that were not disaggregated by the statistical method. For the quantitative analysis of the model improvement, we calculated Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) for each method with direct metering data. For most of appliances, RMSE and MAE of hybrid model were improved from 11% to 71% compared to the existing methods. The collection of more samples to increase the accuracy of the estimation and application to areas of low statistical data availability are future steps.