Evolutionary Optimization Of Air-Conditioning Schedule Robust For Temperature Forecast Errors

This work proposes an evolutionary optimization method of the air-conditioning temperature schedule robust for the forecast temperature errors. To optimize the air-conditioning temperature schedule in office buildings, we need to conduct a building simulation with the air-temperature forecast since the outside air-temperature strongly affects the optimal operation of the air-conditioning system. However, since the forecasted temperatures often involve errors, the air-conditioning schedule optimized for the forecast may not be optimal practically. To acquire practically feasible air-conditioning temperature schedule robust for the air-temperature forecast errors, in this work, we propose an evolutionary multi-objective air-conditioning schedule optimization method for office buildings. In the proposed method, we estimate the temperature forecast errors by using the normal distribution model and apply an improved multi-objective particle swarm optimization algorithm, OMOPSO, to simultaneously optimize the human thermal comfort, the power consumption, and their differences when the air-temperature forecast involves errors. Experimental results show that the proposed method can acquire air-conditioning schedules robust for the uncertainty on the air-temperature forecast.