Application of particle filter to parameter identification of space thermal model

Since the consumption of air conditioning devices takes a significant proportion in the electricity demand, implementation of appropriate control strategies will conserve a considerable amount of energy. A proper space thermal model is required for any control strategy, which is built on the accurate identification of unknown parameters. A first-order lumped parameter model is employed to depict space thermal dynamics. The model is discretized using the fourth-order Runge-Kutta method. We use a particle filter algorithm to identify the model parameters, thermal capacitance and conductance. The recursive algorithm follows the steps including initialization, prediction, update, weight normalization, and resampling. An experiment is conducted to collect indoor and outdoor temperature measurements of a test room. The error is very small between the measured data and prediction results derived by the identified parameters, which verify the effectiveness the identification method proposed in this paper.