Extended Kalman filter based extremum seeking algorithm for model-free control of building thermal environment under uncertainty

Abstract

Online optimization of building thermal environment is no trivial task due to the complex and uncertain indoor thermodynamics. In this study, an extended Kalman filter-based extremum seeking control (EKF-ESC) strategy for model-free control of thermal environment under uncertainty is proposed. The EKF-ESC uses extended Kalman filter for estimating gradient of the objective function of the extremum seeking controller, which enables real-time optimization of thermal environment with fast convergence time. In addition, a co-simulation platform is developed by combining EnergyPlus and Simulink/Matlab for validating effectiveness of the EKF-ESC strategy for controlling thermal environment. Our study demonstrates that the EKF-ESC strategy has good adaptivity to uncertainties resulting from changes in weather, electrical equipment and occupancy. Moreover, it is shown that the EKF-ESC is robust to measurement noise. In addition, comparison of convergence performance among EKF-ESC, adaptive ESC, and classical ESC is conducted. It is shown that the EKF-ESC based control strategy has the fastest convergence time towards indoor temperature setpoints. Compared with the classical ESC and adaptive ESC schemes, about 77% and 55% reduction of the convergence time are realized through EKF-ESC. According to the research results, it could be inferred that EKF-ESC strategy has the potential for effective building thermal management under uncertainty with fast convergence time.