Gepersonaliseerde modellen voor gebruikerstevredenheid bij binnenklimaat regeling in schoolgebouwen

Providing a good indoor environmental quality (IEQ) in classrooms is important for guaranteeing pupils' and students' satisfaction and learning performance. Occupant-centric control (OCC) for HVAC systems has the potential to improve occupants' satisfaction with the IEQ by including occupants' preferences in the control loop. An OCC system that is responsive to occupants' unique preferences requires a personalised modelling approach for predicting these unique preferences. Currently, personalised comfort models have been defined and tested for predicting thermal comfort preferences. However, these approaches often require a large number of occupant feedback votes, which are challenging to collect in reality. In addition, current modelling approaches are often untransparent which decreases their comprehensibility. Moreover, a personalised modelling approach for predicting occupants' indoor air quality (IAQ) satisfaction has not been defined in literature, limiting the possibilities of defining an OCC system that is responsive to occupants' IAQ preferences. Finally, the application of HVAC control systems integrating personalised comfort models in multi-occupancy spaces (e.g., open-plan offices, classrooms, meeting rooms) is challenging since conflicting occupant preferences have to be resolved. Classrooms in educational buildings is a type of multi-occupancy space that is currently understudied. First, the state-of-the-art on occupant-centric HVAC control systems and personalised comfort models was investigated. Afterwards, three data collection campaigns were conducted in different educational buildings to construct a dataset comprising of occupants' satisfaction and the corresponding IEQ conditions in the classroom. A statistical analysis was performed to determine the main influencing variables for occupants' satisfaction with the IEQ in classrooms. Subsequently, the performance of mixed-effects (ME) models as a personalised modelling approach for predicting occupants' thermal and IAQ satisfaction was evaluated. Finally, an OCC framework was defined which integrates the ME models. The performance of the OCC framework was determined via a simulation study. In total, a dataset comprising of 6834 satisfaction assessments was collected from 321 unique occupants in classrooms. The statistical analyses on the dataset showed that room temperature was significantly associated with occupants' thermal and IAQ satisfaction in all of the monitored classrooms. Other environmental (e.g., CO2-concentration and relative humidity) and contextual variables (e.g., occupants' location in classroom) showed statistically significant relationships with satisfaction, but these were not present in all classrooms. In addition, the results in this project showed that a ME modelling approach was capable of predicting occupants' individual thermal and IAQ preferences. The ME model incorporated the individuality of the occupants and was therefore able to achieve higher prediction accuracies than models that did not take this individuality into account. Moreover, only a limited number of datapoints (i.e., 10 occupant feedback votes) was required to make stable predictions. Simple ME models with room temperature as sole predictor achieved similar accuracies for predicting occupants' thermal and IAQ satisfaction as more complex ME models. Finally, an OCC framework for classrooms, which integrated the ME models, was defined. The OCC framework was able to improve occupants' thermal and IAQ satisfaction.