Date of Award

May 2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering (Holcomb Dept. of)

Committee Member

Rajendra Singh

Committee Member

Richard Brooks

Committee Member

Amy Apon

Abstract

Internet of Things (IoT) is attempting to transform modern buildings into energy efficient, smart, and connected buildings, by imparting capabilities such as real-time monitoring, situational awareness and intelligence, and intelligent control. Digitizing the modern day building environment using IoT improves asset visibility and generates energy savings. This dissertation provides a survey of the role, impact, and challenges and recommended solutions of IoT for smart buildings. It also presents an IoT-based solution to overcome the challenge of inefficient energy management in a smart building environment. The proposed solution consists of developing an Intelligent Computational Engine (ICE), composed of various IoT devices and technologies for efficient energy management in an IoT driven building environment.

ICE’s capabilities viz. energy consumption prediction and optimized control of electric loads have been developed, deployed, and dispatched in the Real-Time Power and Intelligent Systems (RTPIS) laboratory, which serves as the IoT-driven building case study environment. Two energy consumption prediction models viz. exponential model and Elman recurrent neural network (RNN) model were developed and compared to determine the most accurate model for use in the development of ICE’s energy consumption prediction capability. ICE’s prediction model was developed in MATLAB using cellular computational network (CCN) technique, whereas the optimized control model was developed jointly in MATLAB and Metasys Building Automation System (BAS) using particle swarm optimization (PSO) algorithm and logic connector tool (LCT), respectively. It was demonstrated that the developed CCN-based energy consumption prediction model was highly accurate with low error % by comparing the predicted and the measured energy consumption data over a period of one week. The predicted energy consumption values generated from the CCN model served as a reference for the PSO algorithm to generate control parameters for the optimized control of the electric loads. The LCT model used these control parameters to regulate the electric loads to save energy (increase energy efficiency) without violating any operational constraints.

Having ICE’s energy consumption prediction and optimized control of electric loads capabilities is extremely useful for efficient energy management as they ensure that sufficient energy is generated to meet the demands of the electric loads optimally at any time thereby reducing wasted energy due to excess generation. This, in turn, reduces carbon emissions and generates energy and cost savings. While the ICE was tested in a small case-study environment, it could be scaled to any smart building environment.

Share

COinS