Machine learning algorithms for predicting occupants' behaviour in the manual control of windows for cross-ventilation in homes

Abstract

The manual control of windows is one of the common adaptive behaviours for occupants to adjust their indoor environment in homes. The cross-ventilation by the window opening provides a useful tool to control the thermal comfort and indoor air quality in homes. The objective of this study was to develop a modelling methodology for predicting individual occupant's behaviour relating to the manual control of windows by using machine learning algorithms. The proposed six machine learning algorithms were trained by the field monitoring data of 23 sample homes. The predictive performance of the machine learning algorithms was analysed. The algorithms predicted the occupant's behaviour more precisely compared with the logistic model. Among the algorithms, K-Nearest Neighbours (KNN) shows the best fitness with the monitored data set. The driving parameters of the manual control of windows in each sample home can be clearly drawn by the algorithms. The proposed machine learning algorithms can help to understand the influence of the occupant's behaviour on the indoor environment in buildings.

Driving factors and models of window opening behaviour in Xi’an residential buildings

“…and reinforcement learning methods have also been used for window modelling. Among the proposed six machine learning algorithms, Park et al 25 found that the K-Nearest Neighbours (KNN) is optimal. For office buildings, Markovic et al 26 proposed a deep learning method and incorporated the model into the simulation of the Modelica thermal building.…”

“…22 In order to shield other influencing factors as much as possible, the residential buildings chosen for this paper are located in the same area of the same city, the windows are all southfacing, and members of the family are relatively similar all year round. Generally, people often use natural ventilation in order to reduce indoor temperature and improve the indoor air quality, 25,[30][31][32][33] and mainly open the windows during the transition seasons when the outdoor temperature is suitable, and mechanical adjustments are incorporated in the winter and summer, but at the same time increase the energy consumption of the building. 34 For these reasons, most scholars mainly study the influence of window opening behaviour in transition seasons.…”

Prediction of phthalates concentration in household dust based on back propagation neural network

“…Artificial neural is a computational structure based on the mechanism of biological neural network. 30,31 It is a kind of imitation, simplification and expression of biological neural network. Artificial neuron network is a method to imitate working principle by studying and simulating biological neuron.…”

The differential dynamic model for the implicit safety culture dissemination in nuclear power plants

“…Besides, there are also some studies on the relationship between human behaviours and occupational health, safety and performance. [8][9][10] A wellknown theory, the 'People-based Safety' theory presented by Geller, 11 discussed the interdependency among the safety culture and the human per se, the human behaviours and the environment. It suggested that a transformation from dependence and independence to interdependence was indispensable to achieve an injury-free workplace requires.…”