A method for revising temperature and humidity prediction using additional observations and weather forecasts

Weather prediction is considered to be essential for the predictive control of HVAC systems in which dynamic components, such as a thermal storage tank or heavy building envelope, exist. This paper proposes a method for revising the prior prediction of ambient temperature and humidity by combining two additionally available different data sources, i.e., observations at the building site and forecasts from a weather station. The proposed method applies the theory of orthogonal projection employed in the Kalman filter, and the revised predictions are statistically optimal for determining the minimumvariance linear estimate.