Electrical and Computer Engineering

Overvoltage Protection in HVDC Systems

Supervisor

Dr. Abhisek Ukil

Discipline

Electrical and Computer Engineering

Project code: ENG041

The advance of voltage source converter-based high-voltage direct current (HVDC, >35kV) transmission systems makes it possible to build a multi-terminal HVDC grid. Compared with HVAC grids, active power conduction losses are relatively low and reactive power conduction losses are zero in an HVDC grid. One of the most important and pivotal topics for such DC grid, is the protection, for safeguarding the grid against faults [1-3]. For the AC grid, the circuit-breakers (CB), relays, and the protection standard (IEC60255) are established, which are not available for the proposed DC grid, the nature of the current and the voltage being different. In this project, theoretical investigations about overvoltage protection, e.g., its requirements, state-of-the-art devices, R&D needs, etc. would be performed. This would be followed by in-depth simulation verification of the most promising methods using software like PSCAD, EMTP.

Pre-requisite: Power systems, Power electronics courses

Time: August/2018 – March/2019

1. N. Geddada, Y.M. Yeap, A. Ukil, "Experimental Validation of Fault Identification in VSC Based DC Grid System," IEEE Transactions on Industrial Electronics, vol. 65, no. 6, pp. 4799-4809, 2018.
2. Y.M. Yeap, N. Geddada, K. Satpathi, A. Ukil, "Time and Frequency Domain Fault Detection in VSC Interfaced Experimental DC Test System," IEEE Transactions Industrial Informatics, 2018.

DC Arc Fault Detection

Supervisor

Dr. Abhisek Ukil

Discipline

Electrical and Computer Engineering

Project code: ENG042

As the number of DC loads are growing in our daily uses, the low voltage DC (LVDC) distribution system is becoming important, especially for EV charging, datacenter applications. Integration of renewable energy sources at LV level, like rooftop photovoltaic (PV), fuel cells, etc. would be relatively easier with the LVDC system than the LVAC.For high-voltage DC (HVDC) transmission, active power conduction losses are relatively low and reactive power conduction losses are zero. One of the most important and pivotal topics for such DC grid, is the protection, for safeguarding the grid against faults [1-2]. For both levels, arcing faults or high impedance faults are difficult to detect, as it involves very little increase in fault current. In this project, theoretical investigations about dc arc fault detection methods, e.g., its requirements, state-of-the-art devices, R&D needs, etc. would be performed. This would be followed by in-depth simulation verification of the most promising methods using software like PSCAD, EMTP.

Pre-requisite: Power systems, Power electronics courses

Time: August/2018 – March/2019

1. N. Geddada, Y.M. Yeap, A. Ukil, "Experimental Validation of Fault Identification in VSC Based DC Grid System," IEEE Transactions on Industrial Electronics, vol. 65, no. 6, pp. 4799-4809, 2018.
2. Y.M. Yeap, N. Geddada, K. Satpathi, A. Ukil, "Time and Frequency Domain Fault Detection in VSC Interfaced Experimental DC Test System," IEEE Transactions Industrial Informatics, 2018.

Software for education

Supervisor

Nasser Giacaman

Discipline

Electrical and Computer Engineering

Project code: ENG043

In Software Engineering, the best way for students to learn is to practice. As such, this project will involve the development of various tools and approaches to assist students and instructors in software engineering and programming courses. Examples include Active Classroom Programmer and InteractiveDS. A range of projects are available for this area that vary from time to time.

The applications are not limited for Software Engineering, but might also target Mathematics Education.

Students are expected to be confident and independent programmers.

Modelling and prediction of cardiac electrophysiology signals

Supervisor

Avinash Malik
Tommy Peng

Discipline

Electrical and Computer Engineering

Project code: ENG044

1 in 3 New Zealanders suffer from cardiovascular diseases. Doctors often use the electrocardiogram (ECG) to diagnose and monitor cardiac diseases. However, ECGs are recorded from electrodes on the body surface and do not represent what is happening on the heart surface.

During this project, students will use machine learning techniques to manipulate and model pre-recorded body surface signals to predict what is happening on the heart surface.

Our research group has developed mathematical techniques to accurately represent ECGs (from the body surface) and EGMs (from the heart surface) as Gaussian series representations. In this project, the student will develop deep learning neural networks to build a model that can map (and eventually predict) the Gaussian time series parameters of the EGM from the Gaussian time series parameters of the ECGs.

The developed technology can potentially enable predicting heart surface potentials without invasive surgery thereby reducing risks to patients and in-patient hospital costs.

Candidates should have:

  • An interest in machine learning
  • A good understanding of statistics
  • MATLAB experience (or other programming experience)

Analysing Performance Overhead of Encryption/Decryption for Secure Memory Transactions in Multiprocessor Systems on Chip

Supervisor

Dr. Morteza Biglari-Abhari
Benjamin Tan

Discipline

Electrical and Computer Engineering

Project code: ENG045

The number of applications using network connected embedded systems are increasing rapidly to be employed in the so-called Internet of Things (IoT). However, many of such systems may process sensitive data, which require reliable and secure processing and distribution. There have been some architectural extensions to assist processors and storage systems for security improvements but using encryption/decryption algorithms at the chip level transactions in Multi-processing Systems on Chip (MPSoC) is still an effective mechanism to satisfy the system security requirements. However, this approach introduces some overhead on the processing which may not achieve the required level of performance and energy efficiency. The aim of this project is to develop a system level model to investigate performance overhead for encrypted/decrypted memory transactions in an MPSoC.

Prerequisite: COMPSYS 304

Embedded Camera-Based Person Tracking

Supervisor

Dr. Morteza Biglari-Abhari
Dr. Kevin Wang
Mr. Andrew Chen

Discipline

Electrical and Computer Engineering

Project code: ENG046

A lot of computer vision research has been done into using cameras to track people as they move through a space, but the majority of these algorithms involve computationally expensive methodologies that are not suitable for implementation in embedded systems. The student working on this project will collect and annotate video data, develop an understanding of single-view geometry, work on an embedded implementation of a person tracking algorithm, and trial it in a test environment. Knowledge of Python/C++ is required, and experience with computer vision would be helpful but not required.

HealthBots software improvements

Supervisor

Craig Sutherland
Bruce MacDonald

Discipline

Electrical and Computer Engineering

Project code: ENG047

The Centre for Automation and Robotics Engineering Science (CARES) has been developing robots for healthcare (the HealthBots system). The robots are designed to help elderly people live at home for longer, and performs tasks like reminding about medications, exercises and daily tasks. The system consists of software running on a robot and a web-based backend for storing patient and study information. It is currently used in a number of projects in CARES.
We are looking for a student to make improvements to the software, both on the robot and the web backend. The changes are based on feedback from the current projects.
This will require a knowledge of computer programming and a willingness to learn about how robots work. Knowledge of web-based programming will be useful but not essential.

Nao robot demos

Supervisor

Craig Sutherland

Discipline

Electrical and Computer Engineering

Project code: ENG048

The Department of Electrical and Computer Engineering has recently purchased six Nao robots. We want to use these robots to perform an interactive demonstration, especially for events promoting the department.
We are looking for a student to develop the software for the demo. This will involve learning how to program the Nao robot and working out how to get them to synchronise their movements. This will require a knowledge of computer programming, understanding Python code is useful but not required.

Agile Software Development Review

Supervisor

Dr Rashina Hoda

Discipline

Electrical and Computer Engineering

Project code: ENG049

Agile software development has been a prominent part of software engineering for the past 25 years. It is currently the standard method of developing software in the industry worldwide.

This project involves conducting a review of agile literature to summarize the main trends and evolutions over time. Ideal candidates should have a strong interest in reading and analysing research literature and a strong command over the English language. Knowledge and/or experience of conducting a literature review and that of agile methods will be preferred but is not a must. Students will be provided guidance on conducting the review and writing the report. The project will be best suited to students from software engineering and computer systems engineering backgrounds.

Collaborative game design tool

Supervisor

Dr Rashina Hoda

Discipline

Electrical and Computer Engineering

Project code: ENG050

Game design typically involves multi-disciplinary teams collaborating together to brainstorm and design the story, characters, and other game elements. We’re developing a software tool to support this process. This project involves extending the tool to add more features to support collaborative design. The tool is being developed as a Windows application for the Microsoft Surface Hub. Ideal candidates will have a strong programming background, including both back and front-end. Interest in or experience of game design will be preferred.

Near field antenna characterisation

Supervisor

Dr Andrew Austin

Discipline

Electrical and Computer Engineering

Project code: ENG051

Antennas are typically characterised in the ‘far-field’, however, for large, directional antennas (e.g., satellite-radar antennas), far-field measurements are not always possible due to the physical separations required. This project will investigate ‘near-field’ measurement techniques and develop a near-field probe and scanner.

Experience with MATLAB; experience with electronics desirable; training will be provided on the use of RF test equipment.

Automatic differentiation to efficiently compute sensitivities

Supervisor

Dr Andrew Austin

Discipline

Electrical and Computer Engineering

Project code: ENG052

Automatic differentiation (AD) is a recently proposed novel technique to efficiently compute the derivative of any arbitrary function. Information about the derivatives is extremely useful in optimization and sensitivity analysis. The aim of this project is to build an AD toolbox suitable for estimating the sensitivities in numerical electromagnetics.

Good programming skills required in Matlab or C/C++ or Fortran.

Livability of an LDC Controlled Climate

Supervisor

Prof Grant Covic
Dr Jason James

Discipline

Electrical and Computer Engineering

Project code: ENG053

In power networks around the world the peak demand for electricity is growing at twice the average demand. The current solution to this problem is to increase the infrastructure capacity. The problem is the economic return is based on average not peak supply. Add to the demand an expanding electric vehicle industry, and shift from open fires to heat pumps together with the variability of solar and wind generation, and the situation becomes more complicated. Local Demand Control (LDC) provides a solution.

As part of an MBIE project 2 ex state houses located at the University of Auckland Ardmore field site have been extensively re-configured to solely support LDC research. One house is focused on determining the livability of an LDC controlled climate. A full array of sensing devices together with the ability to simulate, emulate or control real power loads in any combination and automatically operate blinds, windows and humidifiers (the human effect) provides a unique opportunity to run experiments at real power levels in a completely reconfigurable setup.

To understand the impact of LDC on the household environment a base line needs to be established from which the livability boundary conditions of LDC can be understood. This summer project will focus the recipient on how best to configure the existing system to replicate both the power usage and internal environment for several typical kiwi households.

This project will suit someone interested in a range of engineering specialties and how the system can replicate real life based on collaborative research with other University of Auckland departments. You will further your skills in literature review, writing, collaboration, design and operating a wider system based on real data. It would suit a BE and MB student with a strong interest in electronics within the home and the integration of life with engineering.

Smart daylight sensor for commercial lighting systems

Supervisor

Dariusz Kacprzak

Discipline

Electrical and Computer Engineering

Project code: ENG054

The purpose of the project is to design, build and tests a smart daylight sensor. The sensors must maintain its dimming function regardless the changes to the objects placed in the room. The existing and commercially available sensors are strongly affected by the surface reflectance of the furniture in the rooms. However the proposed sensor will not be affected by the highly reflective furniture. A concept of this sensor was already presented at 11th International Conference on Sensing Technology 2017 (Kacprzak D. “Self-adjustable daylight sensor for lighting systems”). At that time the sensor was tested in a lab on a test desk. No field tests were done. In this project we will design a sensor for commercial applications and the sensor will be tested in actual rooms and office spaces.

Generating a Māori pronunciation Dictionary via Natural language processing

Supervisor

Dr Catherine Watson
Dr Hywel Stoakes
Dr Peter Keegan (Te Puna Wānanga in the Faculty of Education
and Social Work)

Discipline

Electrical and Computer Engineering

Project code: ENG055

To create world class speech technology tools for Māori we need to first develop a suite of language processing tools. This projects aims to develop the first Māori pronunciation dictionary ever to be developed, and this will be used in Speech Synthesis for Te Reo, Computer–based Māori Pronunciation tools, and automatic correction of Historical Māori land records incorrectly transcribed, to name but a few applications. Successful participants will need to have a strong grounding in software development, and an interest in language processing will help. They will also need to have good communication skills because they will be working with a multi-disciplinary team of researchers. Knowledge of Māori language is advantage but not a necessity.

Designing IPT pads for in-motion charging

Supervisor

Professor Grant Covic

Dr Seho Kim

Discipline

Electrical and Computer Engineering

Project code: ENG056

In-motion charging of electric vehicles (EV) using inductive power transfer (IPT) has become a main stream research topic recently. The goal of this project is to design and develop a scaled down in-motion IPT pad, which will be tested for thermal and mechanical stress.

Applicant should preferably have some experience/knowledge in both electrical and mechanical engineering.

Linking heart-lung synchronization with controlled breathing and associated digital pacing devices

Supervisor

Partha Roop and Vinod Suresh

Discipline

Electrical and Computer Engineering

Project code: ENG057

Recent research findings indicate that respiratory sinus arrhythmia (RSA) is an essential component of a healthy heart. This achieves heart-lung synchronisation and may achieve cardiac efficiency (Ben Tal et al. 2014, Norgaret et al. 2014). During this project, we will examine (1) digital pacing protocols that achieves RSA in patients who have lost RSA due to disease, and (2) study the impact of pranayama i.e. specialised breathing techniques, associated with yoga and the link to RSA, if any.

Students with good background in signal processing, embedded systems are preferred. Students from biomedical engineering can also apply for this project. Software engineering students with interest in human health may also apply.

Synchronous Machine Learning for Autonomous Vehicles

Supervisor

Partha Roop

Discipline

Electrical and Computer Engineering

Project code: ENG058

This project will investigate deep neural networks based on the concept of synchronous artificial neural networks -- SANNs (Roop et al. 2018). SANNs have been designed specifically to ensure that safety critical applications can be implemented in a time predictable fashion. This project will examine the design of some autonomous vehicle algorithms, such as automatic obstacle avoidance and lane changing, using SANNs and networks of SANNs.

Students should have good background in embedded systems and C-programming. Background in image / video processing and AI techniques will be useful though not mandatory. Software Engineering students with interest in AI / Machine learning but no embedded systems skills can also apply. Please discuss with Partha if you are in this category.

Design, Build and Testing of Multi-purpose Wireless Sensors for Monitoring Composite Manufacturing Processes