Establishment of a personalized occupant behavior identification model for occupant-centric buildings by considering cost sensitivity: Fid-Bau Portal

Establishment of a personalized occupant behavior identification model for occupant-centric buildings by considering cost sensitivity

Li, Zhengrong / Zhu, Han / Ding, Yan / Xu, Xiaofeng / Weng, Binjie

Abstract

Highlights • A personalized occupant behavior model (POBM) was built to realize occupant-centric buildings. • Personalized occupant behaviors were identified by cost sensitivity analysis (CSA). • The evaluation procedure of the POBM–CSA was proposed.

Abstract Modern automation systems play an important role in improving building energy efficiency. However, some evidence shows that occupants are often dissatisfied with automation systems and may intervene, e.g., by manually adjusting automated shading or air conditioning. Therefore, the concept of occupant-centric buildings has been proposed by some scholars to solve the abovementioned problem. As the basis for realizing occupant-centric buildings, personalized occupant behavior should be identified dynamically. Thus, this study proposes a systematic procedure for establishing a personalized occupant behavior model with cost sensitivity analysis (POBM–CSA). The CSA can improve the identification performance of the model while considering the personalized preferences and attitudes of target occupants. By taking shading behavior identification as an example, the total cost and area under the cost curve were used to evaluate this consideration. The results showed that the POBM–random forest with cost-sensitive classification has the best cost sensitivity performance. By establishing a customer rule library, personalized behavior patterns with reduced misclassifications can be output by the POBM–CSA to improve the operation of occupant-centric buildings.