Patented AI tool and method for evaluating building quality - Analysis of indoor environment and human comfort a case study

Abstract

Ensuring indoor environmental quality and occupant comfort in buildings is critical to enhancing productivity and well-being, yet existing assessment methods often fail to integrate objective measurements (collected by sensors on an autonomous robot, representing measurable environmental parameters such as air quality and temperature) with subjective feedback (gathered via online surveys). This study addresses this gap by developing an autonomous tool for evaluating the quality of buildings and building systems. The objective is to compare the effectiveness of Bayesian Belief Networks, a novel artificial intelligence-based approach, with a classical Linear Additive Method that incorporates Analytic Hierarchy Process and Multi-Attribute Utility Theory. Data collection is achieved using an autonomous robot for objective measurements and Bluetooth-guided occupant surveys for subjective feedback. Two Bayesian Belief network models and one Linear Additive Method model were developed and evaluated using data from an educational building in the Eastern Province of Saudi Arabia. Results show that while the Bayesian Belief Networks algorithm requires more computational time, it effectively handles complexities in hybrid data and provides more reliable predictions for indoor environmental quality. The comparison revealed that one Bayesian Belief Networks closely matches Linear Additive Method results with a correlation coefficient of 0.92, while the other did not converge, highlighting challenges in model optimization. This novel framework offers significant potential for scalable and accurate building quality assessments across diverse building types and climates.