Evaluation of Thermal Comfort in a Multi-Occupancy Office using Polak-Ribiére Conjugate Gradient Neuro-Algorithm

Abstract

The evaluation of the level of satisfaction in an indoor air-conditioned environment is an important factor in order to determine the optimum settings for efficient and effective performance of Heating, Ventilation and Air Conditioning (HVAC) systems. In this paper, an efficient and user-friendly thermal comfort evaluation technique is presented. An Artificial Neural Network (ANN) model was developed using the Polak-Ribiére Conjugate Gradient (PRCG) algorithm to solve the problem of evaluating thermal comfort in a multi-occupancy office. The network employs mean radiant temperature, indoor air temperature, relative humidity, metabolic rate, air velocity and clothing insulation as inputs and the predicted mean vote value (PMVV) as the output. In order to validate the performance thereof, the PRCG-trained network was retrained and simulated using the Levenberg-Marquardt (LM) algorithm. The results show that the model performed satisfactorily in evaluating the comfort of occupants and stands as a rapid tool for HVAC system designers in analyzing HVAC systems.