STATISTICAL TIME-SERIES MODELS OF SOLAR-RADIATION AND OUTDOOR TEMPERATURE - IDENTIFICATION OF SEASONAL MODELS BY KALMAN FILTER

To successfully design comfortable and energy-saving office buildings, detailed information concerning the thermal characteristics of the building is necessary. For that purpose, a method to calculate room temperatures and heating load which takes account of the randomness of the external climate is required. Recently, a method of calculation has been presented that will produce the maximum load and distribution of the load of the intermittently heated buildings, and which takes into account the stochastic nature of external climate. The calculation method requires statistical time series models of the external climate. In this paper, as an alternative to the use of hourly historical weather data, a statistical method is proposed to generate synthetic weather data. We have developed adequate statistical models for the solar radiation (ARMA model) and outdoor air temperature (ARMAX model) time series, and their correlations. On identification of the statistical models, solar radiation during nighttime, which is deterministically zero, is treated as missing. Since the usual methods for parameter estimation are not available for time series with missing data, a method of parameter estimation is developed based on the Kalman filter.