The fixed description of HVAC behaviours leads to inaccurate prediction of air conditioning energy consumption, which in turn affects the appropriateness and effectiveness of energy conservation strategies. Based on a naturally ventilated research building located in Hangzhou, China, a stochastic prediction model reflecting actual HVAC behaviours is established based on clustering analysis and the Monte Carlo method, and it is integrated into the AC energy consumption simulation through Python programming. Then, important factors influencing AC energy consumption are clarified by importance analysis based on random forest regression, and the integrated strategies based on them are studied based on the simulation and control variable approach. As a result, the error rate between the measured and simulated AC power consumption is −5.24% and 2.56% in the heating and cooling conditions, respectively. And the relative importance and the number of important factors following the actual HVAC behaviours are remarkably different from those based on the fixed behavioural pattern. The implementation of integrated AC energy conservation strategies based on important influencing factors achieves 35.02% energy savings. Consequently, a theoretical basis for the accurate prediction of AC energy consumption and efficient implementation of energy conservation strategies is established.
Read full abstract