A novel performance indicator for the assessment of the learning ability of smart buildings

The rapid development of artificial intelligence (AI) and machine learning (ML) has made it topical to consider learning ability as one of the key performance characteristics of buildings. So far, the buildings' learning ability has not explained or clarified by definitions or in terms of the proposed frameworks of key performance indicators (KPI). In this paper, a novel performance indicator based on the concept of learning gain is developed to quantify the learning ability of buildings by way of a single, dimensionless number between zero and unity. The implementation of the new Learning Ability Index (LAI) is demonstrated by way of three different case studies chosen from the literature. It is concluded that LAI is an easy and illustrative tool to assess the learning ability of buildings. Particularly, it is useful for monitoring the performance of data-driven processes, when pursuing the preferred strategies to reach higher levels of building intelligence. The LAI considers the time invested in learning plus the quality and diversity of learning material. It is flexible with respect to system boundaries or the performance metrics, wherefore it can be implemented as a generic indicator of system evolution, as well.