Pattern classification of energy usage behaviors in an office building

For the view of the different occupants’ energy usage behavior, conducts questionnaire survey and on-site monitoring on the usage habits of air conditioning and lighting equipment in an office building in Tianjin. The results show that different thresholds of occupants to turn on air conditioning and lighting lead to diverse accepted temperature ranges and illuminance levels. By k-means clustering algorithm, the occupants can be divided into three types of sensitive type, common type and insensitive type. Uses Weibull models to classify each type of occupants. Obtains the probability schedule of equipment opening as the inputs of the building simulation, which improves the accuracy of building energy consumption simulation by more than 20%. In addition, occupants with different environment sensitivity have a large difference in the opening probability of air conditioning and lighting equipment, so it is necessary to classify occupants before the establishment of human behavior model.