Abstract:In the process of achieving the goal of "carbon peak and carbon neutrality" in China, the demand response (DR) technology of the power system is particularly important, And the building air conditioning load (ACL) is an important DR resource, which has great potential for regulation and low control cost. In this context, the equivalent thermal parameter (ETP) model, as a simplified model of the building thermodynamic process, is widely used to study and control the building ACL by simulating the multi-order circuit of resistance and capacitance. However, the traditional ETP model parameter identification methods fail to fully consider the inherent errors and observation errors of the models. Therefore, this paper proposes a parameter identification method based on Bayesian calibration, which analyzes the model parameters from the perspective of probability, and combines the prior knowledge of the system to quantify and adjust the uncertainty in the model. Firstly, the ETP model is abstracted from the specific architectural features, and then the Bayesian calibration is used to identify the model parameters. Experimental results show that the proposed method not only improves the credibility of the model parameters, but also makes its DR performance more accurate and reliable. In addition, the identified ETP model is applied to the construction of DR capability confidence interval and capability curve, which provides effective technical support for the construction of smart grid.