Institutional Repository [SANDBOX]
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Adaptive control system using Artificial Neural Networks for the optimization of heating and cooling function of a residential space

Polymenopoulou Alexandra

Full record


URI: http://purl.tuc.gr/dl/dias/2A54B2CA-A090-4DFC-AE7A-5CCCCC5F7430
Year 2018
Type of Item Diploma Work
License
Details
Bibliographic Citation Alexandra Polymenopoulou, "Adaptive control system using Artificial Neural Networks for the optimization of heating and cooling function of a residential space", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2018 https://doi.org/10.26233/heallink.tuc.80111
Appears in Collections

Summary

The subject of the current thesis project focuses on an adaptive control system development using neural networks, implemented to regulate the indoor temperature of a specifically designed for this purpose residential space. The main purpose of this project is to control the activation or deactivation of the heating and cooling system, in order to ensure the annual reduction of energy cost, compared to a simple thermostat, while maintaining thermal comfort. Firstly, specifications of the residence are defined, and the energy model is simulated for one year, providing the energy data and the behavior of internal temperature, when a simple thermostat is used. After investigating literature, the author constructed neural networks, trained them with data provided from the energy simulation of the residence and used them as models in the control system, to predict the future presence of residents in the area, the indoor temperature and the energy consumption due to heating or cooling operation. The developed control system, simulated for one year, seeks on each simulation timestep the optimal decision for the heating and cooling operation, based on future activity of residents, by calculating the energy cost and the cost of thermal comfort loss for a number of potential time sequences of operation. Then, the annual response of the controller is compared with the output of the energy simulation and results in a reduced energy consumption. Furthermore, additional conclusions are made regarding the efficient usage of multiple neural networks in the control system. Finally, overall suggestions, concerning the optimization of the method developed in this control system, are set to be used for further investigation.

Available Files

Services

Statistics