Detection of Occupancy Events from Indoor Air Monitoring Data

In recent years, there has been observed an increase of interest in maintaining proper air quality in spaces occupied by people. Various strategies offer to provide the relevant information. In this work we consider attaining it on the basis of indoor air behavior episodes, which are evident from observations or measurements made over a period of time. Initially we focused on the automatic detection of events, which are building blocks of episodes. Events were defined as circumstances when indoor air remained under a fixed influence e.g. from a particular combination of factors affecting it. To reach the objective we applied change point analysis to the time series of a selected indoor air parameter, which was monitored in a continuous manner. There were examined two algorithms of change point detection coupled with the refining criteria, proposed using the domain knowledge. It was demonstrated that change point analysis of CO2 concentration time series allows to distinguish events associated with building use by occupants.