Methods Research On Occupant-behavior Recognition From Data Of Environmental Parameters In Bulidings

Energy saving is one of the important issues in the sustainable development stratagies of buildings.Occupant behavior is reported as one of the key factors that determine the building energy comsupiton.The variety of occupant behaviors in buildings have led to a significant mismatch between simulated building energy performance and measured one.In addition,difference in individual behavior patterns is also the main reason for the poor implementation of designed energy-saving measures.Therefore,it is crucial to collect real occupant behaviors in buildings in a large scale to build occupant behavioral models,so as to achieve purposes of accurate simulation and individual energy-saving designing and controlling.However,there exists a great challenge due to the cost of monitoring devices and privacy concerns.This study proposed an inexpensive and minimally intrusive method,to recognize behavior information from environment parameters by data mining approach.This method meets the following requirements: 1)the indoor environment monitoring device should be developed with the up-to-date techniques of sensing,communication and computing,so as to lower the cost of data collection while maintain accept quality.Meanwhile,the device should be suitble for usage under different demands in daily life.2)It could recognize occupant’s behavior information from environment parameters by data mining approaches which have been fully developed.As this method collects only indoor environment parameters,it will not raise the privacy concerns.It is developed on the basis that behaviors of different types,at different locations and from different occupants should have different impacts on indoor environmental parameters.Its key task is to develop appropriate data mining algorithms for behavior recognition from environment data.Based on the requirements above,the team developed a wireless sensing system for monitoring indoor environment of residences,and experimented in ten rooms of seven families from Tianjin and Beijing for over half a year covering a summer,autumn and winter.With the collected data,the study developed a series of recognition algorithms for different behaviors,including operations on light switch,air-conditioner,window,and family composition characteristics,location of behaviors,and room occupancy states.During the development,we applied and compared various data pre-processing methods,data mining algorithms,and pattern interpretation and evaluation methods,and also discussed their advantages and limits.Finally,the paper showed an application of occupant behavioral data to the building ernegy simulation,including the frequent pattern mining from behavioral data,behavior modeling,and behavior and annual building ernegy simulations in DeST.The results show that the proposed approach could be inexpensive and minimally intrusive on large-scale gathering and interpreting information about occupants’ daily behaviors in buildings.The data collection system developed in this study was able to collect data remotely and continuously from different cities and families at the same time.Curve description and frequent-pattern mining algorithms,etc.have achieved good performance in mining the inherent relationships between occupant behaviors and indoor environment,from data of highly time-dependent type.Based on the study,more work could be done to analyze the relationship between building performance,human behaviors,and indoor environment quality(IEQ),so as to provide references for IEQ improvement and energy reduction.