The goal of present work is to decrease electric power consumption in a building employing the developed control method that uses a prediction filter. To accomplish this goal, a model of the premises was constructed in the ANSYS Fluent software and a PWM predictive controller was synthesized. Modeling of daily state of the premises with maintenance of assigned temperature using the predictive controller, a two-position controller and a PID-control was performed. Results of modeling demonstrate that the use of predicting controller, taking into account parameters of the building, heating and ventilation systems, outdoor air temperature with maintaining minimal permissible operating air temperature in the premises at night, at weekends and on holidays, makes it possible to save heat resources. Refusal from continuous control and transition to the PWM predictive controller demonstrated a decrease in operating time of heating equipment by 2.3 times from 24 to 10.5 hours. The proposed control method showed the best controlling accuracy equal to 5 %, compared to a two-position control with hysteresis and a PID-control.