부산대학교 도서관

Air conditioners (ACs) are an important resource for demand-response (DR) services in electricity networks and have motivated much research activity in recent years. Existing approaches to model and control aggregate AC demand largely focus on ensembles of fixed-speed (on/off) compressor technology. However, AC markets are dominated by variable-speed compressor technology, which is rapidly becoming widespread. This paper presents a mean-field, first-principle modelling formalism for the analysis and control of aggregate demand of heterogeneously distributed ensembles of variable-output compressor ACs (VOCACs). This formalism posits that the model structure of an individual VOCAC fitted with the mean values of the VOCAC parameters distributed in the ensemble can accurately represent the aggregate DR of the ensemble, in complete contrast with models for fixed-speed AC ensembles. The proposition is numerically validated in two prevalent DR applications of model predictive control (MPC): shaving peak demand and firming photovoltaic generation capacity. The proposed representative model is embedded in the MPC designs to predict the aggregate demand of a heterogeneously distributed ensemble of 1,000 VOCACs by centralised manipulation of temperature setpoints. The model is extended to enable DR by constraining power consumption, which is illustrated in an MPC implementation of the Australian/New Zealand Standard AS/NZS-4755 DR modes.