Air-conditioning Use Behavior Prediction And Flexible Load Analysis In Office Buildings

The rapid development of economy and society promotes the rapid growth of energy demand,the accompanying ecological and environmental problems are becoming more and more serious.Meanwhile,the energy consumption volume in the construction field is huge,and HVAC is the key part of its energy consumption.Therefore,the energy conservation and emission reduction in the field of HVAC be paid more attention.By continuously monitoring the air-conditioning operation and indoor and outdoor environmental parameters of an office building in Changsha for one year,this paper explores the energy-saving potential in the field of HVAC from the two aspects of air-conditioning use behavior and flexible load.The main work and research results are as follows:(1)Based on the monitoring data,the characteristics of air-conditioning use behavior of personnel in office buildings are mined.The characteristics are found as follows:The length of service time differs among different room types of air-conditioners.So does the distribution of their time.The utilization ratio of air-conditionings in offices is high,and their service time is concentrated and continuous for a long time,while the ratio is low in conference rooms and halls,and their service time is scattered and the characteristic of "on-demand" is obvious.Meanwhile,it is easy to find the waste of air-conditioning use in large offices and halls;The temperature of each room tends to be set at 18～21℃ under the heating condition,and a higher air-conditioning temperature is generally set in small offices than in larger ones;The temperature is mostly set at 24～26℃ under the cooling condition,and the small office is usually set with a lower air-conditioning temperature;The stuff tend to set higher wind speeds under cooling conditions than heating conditions,and the proportion of low-level wind speeds set in large offices is smaller than small offices.(2)Construct the prediction model of air-conditioning opening and closing behavior as well as its temperature regulation behavior,based on the machine learning algorithm.Analyze the importance of each driving factor of its usage behavior.The results are as follows:Based on the XGBoost algorithm,the average value of the evaluation index of the air-conditioning opening and closing behavior prediction model is 97.1%;based on the SVM algorithm,the average value of the kappa coefficient of the air-conditioning temperature regulation behavior prediction model is 0.74;Month and indoor temperature have a greater impact on the prediction of air-conditioning opening and closing behavior in summer,accounting for 0.37 and 0.30 respectively,indoor temperature are the most important factor in winter,accounting for 0.33;The time characteristic has the greatest influence on the prediction of air-conditioning temperature regulation behavior in summer,accounting for 0.28,while it is indoor relative humidity characteristic in winter,accounting for 0.28.(3)This study proposes a flexible cooling and heating load model based on machine learning combined with EnergyPlus,by establishing the day ahead optimal scheduling model of the integrated energy system including energy storage equipment and photovoltaic equipment,it verifies the superiority of the flexible load model in reducing the economic and environmental cost and meeting the thermal comfort of users.It is also found that the optimal scheduling of the integrated energy system considering the flexible load can increase the penetration of renewable energy and reduce the economic and environmental cost of system operation.This study will provide a reference for establishing other air-conditioning use behavior prediction models and flexible load models.