Integrated Method for Personal Thermal Comfort Assessment and Optimization through Users’ Feedback, IoT and Machine Learning: A Case Study †

Abstract

Thermal comfort has become a topic issue in building performance assessment as well as energy efficiency. Three methods are mainly recognized for its assessment. Two of them based on standardized methodologies, face the problem by considering the indoor environment in steady-state conditions (PMV and PPD) and users as active subjects whose thermal perception is influenced by outdoor climatic conditions (adaptive approach). The latter method is the starting point to investigate thermal comfort from an overall perspective by considering endogenous variables besides the traditional physical and environmental ones. Following this perspective, the paper describes the results of an in-field investigation of thermal conditions through the use of nearable and wearable solutions, parametric models and machine learning techniques. The aim of the research is the exploration of the reliability of IoT-based solutions combined with advanced algorithms, in order to create a replicable framework for the assessment and improvement of user thermal satisfaction. For this purpose, an experimental test in real offices was carried out involving eight workers. Parametric models are applied for the assessment of thermal comfort; IoT solutions are used to monitor the environmental variables and the users’ parameters; the machine learning CART method allows to predict the users’ profile and the thermal comfort perception respect to the indoor environment.

Personal thermal comfort models with wearable sensors

“…We selected sensors for this study based on three criteria: (1) accuracy, (2) raw data access for research support, and (3) convenience to wear for 24/7. The commercial wrist-bands and smartwatches appeared as the most suitable due to commercial availability and infusion of multiple sensors, as what were employed in previous studies [19,38,48,49]. However, those devices might not measure parameters with acceptable accuracies as described in Appendix A1.…”

“…Each algorithm can be applied to train a personal thermal comfort model based on the data-driven method, leading to 196 personal models in total. Some algorithms have been successfully applied previously to infer thermal comfort using environmental and/or physiological data, such as Classification and Regression Trees [19], Bayesian network [20,58], Logistic regression [23,26], J48 decision tree [59,60], and Random forest [26,38,61], and SVM [29]. In addition, the missing data in the total dataset were imputed using the K-nearest neighbors ("knn") algorithm.…”

“…Kim et al [17] proposed a framework of personal comfort models that can predict an individual's thermal comfort responses by leveraging the Internet of Things (IoT) and machine learning, rather than the responses of an "average person." Such framework has been applied in a few recent studies that aimed to "customize" thermal comfort models for each occupant through users' feedback, IoT and machine learning [18][19][20][21]. The primary advantage of a personal thermal comfort model lies in its capacity of self-learning and updating to suit an individual with a data-driven approach, resulting in higher prediction power.…”

“…Second, most studies evaluated the performance of personal models, which predict categorical responses (e.g., cooler, warmer, no change), using accuracy that is the number of correctly predicted instances divided by the total number of instances in the dataset. Previous studies using such metric reported the prediction accuracy 79% ± 32% (Mean ± SD) for personal comfort models developed from physiological data with wearable sensors [18,19,29,32,39,41]. However, this metric is problematic because it fails to exclude correct prediction purely due to randomness [42,43].…”

Intelligent Thermal Comfort Controlling System for Buildings Based on IoT and AI

“…Their experiment was the indoor environment in steady-state conditions (PMV and PPD) with the use of nearly and wearable solutions. The method was applied by classification and regression tree (CART) algorithm for machine learning with testing in real offices involving eight workers [25]. Luo et al summarized the recent literature during the period of 2016-2019, from perspectives of predicted comfort indices, algorithms applied, input features, data sources, sample size, training proportion, predicting accuracy, etc.…”