Statistical Approach to Physically Based Load Modeling. Final Report

摘要

A methodology for synthesizing the electric load for a system starting from the elementary device models is proposed. Since the methodology is physically based and captures the stochastic nature of the decision processes which generate the load, it is more suitable than identification-based approaches in applications such as load management. It generates a hierarchy of models from the customer or building level to the bulk power level. At each level, the methodology consists of four steps, the representation of primitive components, the classification of components, aggregation, and model validation. This research focuses on modeling the load demand of a large number of similar devices at the lowest level. Each elementary device in the system is represented by two models: the functional model which gives the on or off status of the device and the electric-device model which summarizes the effects of voltage and frequency. Only loads with similar functional and electric characteristics can be replaced by an equivalent model. For such a homogeneous group, statistical aggregation yields a model consisting of coupled ordinary and partial differential equations. The methodology is used to model the space heating of a large number of houses. Simulation of cold-load pickup is used to illustrate the main features of the final model. Data requirements for such models are discussed and found to be modest. (ERA citation 08:031904)