Study on an adaptive thermal comfort model with K-nearest-neighbors (KNN) algorithm

Compared with the static thermal comfort models like predicted mean vote (PMV) model, adaptive thermal models have a wider range of adaptability. The traditional concept of adaptive thermal comfort is that residents actively adapt to environmental changes. In this paper, a K-nearest neighbors (KNN)-based thermal comfort model is developed to establish a personalized adaptive thermal comfort environment to adapt to the preferences of the occupants. The KNN-based thermal comfort model can adjust the thermal comfort boundary for one specific individual person according to the changing environmental conditions. An artificial intelligent (AI) environmental controller has been built for studying the KNN-based thermal comfort model. 34 volunteers have been invited for testing the effectiveness of the KNN-based thermal comfort model. The test results manifested that the percent accuracy of the KNN model with 1000 sets of training data could reach up to 88.31% and can meet practical demand. The proposed thermal comfort model can help different people establish their personal indoor thermal comfort environment and promote the development of the intelligent and personalized airconditioning systems.

Automated summary

A K-nearest neighbors (KNN)-based thermal comfort model is proposed in this paper to establish a personalized adaptive thermal comfort environment according to individual preferences, and it is studied through an AI environmental controller. Experimental results show that the model can achieve an accuracy of 88.31% with 1000 sets of training data, effectively meeting practical demand.