Repository logo
 

Modelling and Optimization of an Intelligent Home Energy Management System (HEMS) in Microgrid

Files

Thesis is embargoed till 27 June 2026(30.93 MB)

Date

2023

Authors

Lin, Xin

Supervisor

Zamora, Ramon
Baguley, Craig

Item type

Degree name

Doctor of Philosophy

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

Recently, carbon emissions and depletion of fossil fuels are exacerbated by the high demand for electrical energy in the residential sector. Renewable energy sources (RES) as an environmentally friendly energy source have been widely installed in residential buildings. However, the intermittent of RES makes it challenging to providing stable power to residential customers. To solve this problem, an energy storage system (ESS) is integrated into homes to improve the quality of electrical energy. This kind of power system integrating RES and ESS in the house can be defined as a home microgrid. The home energy management system (HEMS) plays an important role in the home microgrid, which can collect data information through the home local area network and the Internet to predict the user’s energy usage. Then, the HEMS will use the predicted data to formulate an optimal power scheduling plan, so as to reduce the operating cost of the system and improve the utilization efficiency of RES. Furthermore, the HEMS is beneficial to both utility and home customers. For the utility, the HEMS can moderate the household’s power demand from the utility during peak periods, thereby reducing the supply pressure on the main grid. The HEMS can increase the penetration of RES installations in the residential sector, thereby reducing carbon emissions. For home users, the HEMS can enable users to change from traditional consumers to prosumers that actively participate in the energy trading market. In addition, the HEMS can change the user’s energy usage patterns, thereby reducing household energy consumption. Therefore, it is necessary to study the application of the HEMS in modern smart grids. The purpose of this study is to develop an energy management system for a smart home with photovoltaics (PV) and hybrid energy storage system (HESS), with the aim of reducing electricity costs while meeting occupant comfort needs. The research work carried out by this study is mainly divided into two parts. The first part is the local level energy management system, which can be further divided into two sub-parts presented in separate independent chapters. The two subparts are the decentralized control method for single type energy storage and the multifunctional control method for HESS, respectively. The second part is the system level energy management control, which is also divided into two parts and presented in separate independent chapters. These two subparts are load prediction and power optimization respectively. Chapter 1 provides the background and motivation of this thesis. It also points out the research gaps, research questions and specific contributions of the thesis. Chapter 2 provides a comprehensive review of smart energy management systems in home microgrids. This part first introduces the concept and structure of household microgrid in detail, including HESS composed of high energy density storage (HES) and high power density storage (HPS). Further, this part describes the role and function of HEMS in the home microgrid. These functions can be grouped into load forecasting, optimization and control. Each function is categorized and analyzed in detail. Then, a case study is provided to verify the feasibility of HEMS containing HESS. Finally, this chapter reveals the flaws and gaps in intelligent HEMS research, which constitute a part of the research questions of this thesis. The proposed method starts with a fully decentralized control approach for battery storage systems in the local level energy management system which is presented in Chapter 3. This part of the study develops a state of charge (SoC) balance control method to address the SoC imbalance among batteries regardless of whether the batteries have the same capacity. In this way, a battery with a higher SoC and capacity can deliver more power in discharging mode than ones with a lower SoC and capacity. During charging mode, the battery with higher SoC and lower capacity is controlled to draw less power than the ones with higher capacity and lower SoC. Therefore, SoC balance can be achieved among distributed BSUs. Then, a high-pass filter (HPF) based power droop method is designed to realize voltage regulation and power sharing. In this method, the secondary voltage restoration control is equivalent to an HPF, thereby eliminating the communication connection between the primary control and the secondary control. This part also establishes a processor-in-the-loop (PIL) simulation platform to verify the performance of the method. Chapter 4 proposes a comprehensive multifunctional control method for multiple HESSs composed of batteries and supercapacitors (SCs) in a DC microgrid. The study in this part develops a novel droop-based coordinated control to realize power sharing between battery and SC. At the same time, the method can also realize voltage regulation without an additional voltage recovery control method, thereby reducing the complexity of control. Among them, v-dP is proposed to regulate the battery to provide the average power provided, while the traditional droop control is used to adjust the SC to deliver the instantaneous power. Different line resistances in the system reduce the accuracy of power distribution among the batteries. Therefore, a voltage compensator based on a consensus algorithm is developed to simultaneously achieve accurate power sharing and SoC balancing among batteries. Then, an SoC recovery control of the SCs is adopted to ensure the continuous operation of the SCs. Finally, a power management system is presented to prevent the batteries overuse in different operating mode. After completing the design of the energy management system at the local level, the research work moved into the development of the system level energy management system. Chapter 5 proposes a hybrid approach for improving the accuracy of short-term household load forecasting. The hybrid approach consists of two models. The first model is an ensemble model used to predict home heat and air conditioning (HAC) load usage. The ensemble model includes support vector machine (SVM), back propagation neural network (BPNN), and generalized regression neural network (GRNN). Among them, genetic algorithm is adopted to optimize SVM and BPNN to enhance their performance. Then, a thermal dynamic model is developed to track the indoor temperature which is used as one of the inputs to the ensemble model. Another model is a deep ensemble model used to track the consumption of lighting and other loads. The deep ensemble model consists of three bidirectional long short-term memory (Bi-LSTM) networks. Among them, the Bayesian algorithm is used to optimize the hyperparameters of LSTM. Finally, a trimmed algorithm is used to integrate the results of the ensemble and deep ensemble models to remove undesired outliers. The prediction data obtained according to the method in Chapter 5 will be sent to the optimization layer in the system level energy management system to generate the optimal power scheduling plan. Therefore, the purpose of Chapter 6 is to develop an optimization technique to formulate an optimal power scheduling plan with the objectives of minimizing the operating cost and maximizing the user comforts. Furthermore, this chapter also presents the interaction between system level and local level energy management systems. Therefore, Chapter 6 proposes a multi-level HEMS for a DC home microgrid, which consists of two parts. The first part is the long- and short-time optimization methods based on model predictive control (MPC). This long-term optimization is proposed to minimize the operating cost of the system and maximize the user comfort. In addition, the degradation models of batteries and SCs are designed to evaluate their degradation costs. Hence, the total operating cost covers electricity cost, PV power generation cost, battery and SC wear costs. Then, a predicted mean vote - percentage people dissatisfied (PMV-PPD) model is adopted to evaluate thermal environment comfort. An illuminance model is employed to assess the indoor visual comfort. The purpose of this short-time optimization is to alleviate the power fluctuations caused by the randomness of end-user behaviors and PV generation, as well as to ensure the capacity of the SC within a safe range. The second part of this chapter is the local level energy management system, which method is further improved based on the method in Chapter 4. Furthermore, the local level energy management system is more simplified and has better performance compared to the method presented in Chapter 4.

Description

Keywords

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/16322

Collections

Doctoral Theses