This study aims to categorise the existing irradiance-based blind control occupant behaviour models (BC-OBMs) into distinct clusters of the same characteristics using an unsupervised machine learning approach. Fourteen BC-OBMs, including two base case models (blinds always opened – AL1 and always closed – AL2), have been simulated for an office building in Tiruchirappalli, India and categorised into three distinct clusters named: Passive open (C1), Passive close (C2), and Active (C3). C1 and C2 performed similarly to the base case models, AL1 and AL2, respectively. In contrast, C3 performed as an active cluster with a high number of blind movements (NBMs). This study also quantifies the degree of absolute error between the identified clusters and the base case model AL1 (as per the practice followed in Indian building energy codes) and emphasises the need to incorporate the BC-OBMs into the building energy codes to predict the artificial lighting consumption and lighting levels accurately. To the best of the authors' knowledge, this study is a novel attempt to categorise the existing irradiance-based BC-OBMs into distinct clusters. The research methodology adopted for this study can also be used in future research to categorise the window, artificial light, fan, and thermostat use OBMs into distinctive categories.
Read full abstract