Parameters Estimation and Optimal Scheduling of a Retrofit-Free Refrigerator

Refrigerators, as a type of thermostatically controlled load (TCL), have good capacities to participate in demand response (DR) programs. However, existing methods have paid little attention to refrigerator parameter estimation and have rarely focused on retrofit-free control strategies. These control strategies often take the refrigerator’s temperature set-point or the compressor ON/OFF status as decision variables, requiring retrofitting the thermostatic controller. To address this issue, this paper proposes a 2R1C model for the refrigerator, which facilitates parameters estimation and scheduling strategy design. Based on the 2R1C model, a least-square-estimation (LSE)-based parameters estimation method is proposed to improve the accuracy and efficiency. In addition, a price-based scheduling method is proposed for an actual retrofit-free refrigerator that uses the main-power ON/OFF status as the decision variable, which can be easily implemented through an external power socket. By establishing a temperature-ramp-oriented thermostat control model considering the main-power ON/OFF status, the scheduling model is developed and linearized into a mixed integer linear program (MILP). The optimization, simulation, and practical control of an actual refrigerator verify the effectiveness of the proposed method.
Note to Practitioners
—This paper addresses the optimal scheduling problem of the refrigerator participating in DR, applicable to the majority of practical refrigerators. Existing optimal scheduling strategies for refrigerators suffer from two main issues. Firstly, most of these strategies assume that the thermal parameters of refrigerator loads are given, leading to inaccuracies due to the neglect of parameter estimation. Secondly, the majority of control strategies define the decision variable as the refrigerator’s power consumption rather than the temperature set-point, which deviates from the actual mode in which refrigerators indirectly control power through temperature set-points. This paper proposes a modeling approach based on 2R1C for the mode in which practical refrigerator control power indirectly through temperature set-points. On this basis, a parameter estimation strategy is designed, and an optimization strategy based on MILP is developed with the main-power ON/OFF status of the refrigerator as the decision variable. Compared to traditional methods, the proposed approach can be implemented through an external power socket, making it suitable for the vast majority of refrigerators without the need for retrofitting. In future research, we aim to extend this single refrigerator scheduling strategy to the control of a large number of refrigerators and even clusters of other TCLs, achieving more complex application scenarios.