재실자 착의량 산출을 위한 선행 연구 및 기술 분석

Abstract

Purpose: The aim of this study is to verify the feasibility and applicability of a neural network-based model for estimating clothing insulation of building occupants. This is a preliminary study before developing an estimation model for the clothing insulation. Method: The existing researches on the method of estimating the clothing insulation were investigated and the neural network techniques that can be applied to the model were analyzed. Clothing image datasets were collected and convolutional neural networks (CNNs) that is effective for training images were investigated. Various advanced CNN structures were analyzed to confirm their applicability in developing models. Lastly, an application process for the neural network-based model for estimating clothing insulation and the real-time PMV control was proposed as a flowchart. Result: As a result, the possibility of the neural network-based model for estimating occupants clothing insulation was confirmed, and the basis for providing a comfort indoor thermal environment was established.