Convergence characteristics of a maximum likelihood load model identification scheme

Abstract Physically based electric load modeling is a rapidly growing area of research in view of the new modeling needs created by the introduction of peak load shaving strategies such as direct cycling by the utilities of devices associated with energy storage (air conditioners, electric space heaters, electric water heaters). Indeed, classical regression based models have failed to yield the kind of information needed to construct adequate control policies. In this paper, we tackle the identification problems connected with a previously proposed statistical physically-based load model synthesis methodology for heating-cooling types of electric loads.