SHORT-TERM LOCAL WEATHER FORECAST IN CASE OF SOLVING A PROBLEM OF INCREASING EFFICIENCY OF HEATING SYSTEM

Abstract

The way to increase efficiency of automatic regulation of heating system. For this purpose developed
adaptive model for short-term local weather forecast.The model based on three main factors:
influence of statistics, temperature data for previous years for local area; influence of solar radiation;
temperature changes associated with big displacements of air masses. Also exponential
smoothing methods of numerical time series are mainly used in model. Additionally this model has
a high adaption ability, for each main factor applied special influence coefficient, which are continuously
improved and specified by analysis of model. Described method was used for numerical
experiment in city Chelyabinsk, weather prediction was made for the period of 10 days. Results of
simulation show standard deviation less 0.3 degree, and the maximum prediction error not exceed
1.8 degree.