Modeling and weather-normalization of whole-house metered data for residential end-use load shape estimation

A method is presented for modeling load shapes in the residential sector by using hourly whole-house metered data and temperature. Individual household-level load data are analyzed to achieve data smoothing (noise rejection) and compression (in the ratio of approximately 150:1), and to disaggregate the weather-dependent and weather-independent components of the load. The weather-independent (lifestyle) component is modeled as a weighted sum of orthogonal functions (primarily sinusoids and boxcars), while the weather-dependent component is modeled as a nonlinear dynamic system based on thermodynamic principles. Numerical examples show efficient representation of data and good model fit.<
>