A hybrid model for residential loads in a distribution system with high PV penetration

Abstract

This paper presents a hybrid model for residential electricity demand. Consumer load is modeled in two parts. A low frequency model uses a compact Fourier series to represent slowly changing diurnal loads. Cluster analysis is applied to identify particular load classes and a load classifier based upon quadratic discriminant functions is developed. Once the low frequency load components are removed a high frequency residual load remains. This component is modeled using a power spectrum representation. The load modeling approach was used to model consumption on high demand days for households within the Perth Solar City high photovoltaic penetration feeder trials. The load models were successfully validated against physical data sets with respect to load aggregation behaviors and load cumulative probability functions.