PREDICTIVE CONTROL BY NEURAL NETWORK OF HOT-WATER FLOOR HEATING SYSTEM WITH SHEET PHASE CHANGE MATERIAL INTO THE FLOOR OF RC APARTMENT HOUSE.

Abstract

Assuming the use of excess daytime electricity, we built and verified a model that can predict, using neural network, how to control a floor heating system with PCM installed during the daytime to reduce nighttime electricity consumption while still providing a comfortable temperature range. The results confirmed the effectiveness of PCM in reducing floor surface temperature and room temperature, and showed that it can be predicted more accurately than multiple regression analysis by ANN. In addition, it was suggested that excess electricity during the daytime could be utilized to reduce CO₂ emissions.