Data-driven-based aggregate air conditioning loads' external modeling and load tracking control strategy

Abstract

As more and more controllable loads are connected to the power grid, air conditioning loads (ACLs) can be used as flexible loads to participate in load shaving and accommodate the fluctuation caused by renewable energies because of its characteristics of fast response, large potential adjustment power, cold storage and heat storage. In the actual aggregated process of ACLs, the changes of external conditions such as outdoor temperature, temperature of ACLs and number of ACLs will have a certain effect on the aggregated model. However, the time-varying characteristic of the aggregate ACLs′ external model (AAEM) is not be considered in the most of existing control methods. A data-driven-based AAEM and its corresponding load tracking control strategy are proposed. Based on the cost function of individual ACLs considering thermodynamic characteristics, the data-driven-based AAEM is established by using neural network. By this way, computational complexity of calculation AAEM is reduced. The load tracking control strategy considering the data-driven-based AAEM can effectively reduce the number of times of switching and the total cost. The simulation results verify the effectiveness of this method.

Keywords

• air conditioning loads (acls)
• aggregate acls' external model (aaem)
• data-driven
• thermostatically controlled loads
• demand response
• load control