Filters
Availability
Publication Year
Resource Type
Faculty
School/Institute
Publication Search Results
Sort
Results per page
1 - 10 of 15
-
(2022)ThesisThere has been a global increase in the research and development of military hypersonic technology. Thermal directed-energy systems have been identified as a capability to defend against hypersonic threats. A numerical and experimental methodology for studying the effects of thermal energy deposition on representative hypersonic panels is presented. This thesis contains four sections, (i) theory and implementation of a first-order, fast, transient thermal-structural code: "Rapid Engineering Determination of Heating over a Trajectory'' (REDHOT), (ii) thermal-structural results from two case studies using REDHOT with energy deposition, (iii) development of an experimental technique to create and measure adverse thermal-structural failure caused by energy deposition, (iv) experimental validation of the technique. The first-order thermal structural code uses the reference-enthalpy method and two-dimensional conduction to calculate the thermal state of a representative hypersonic panel. Thermal stresses are calculated analytically with linear plate theory and non-linear finite element analysis simulation. Numerical results using the HyperX and HEXAFLY-INT trajectory as case studies are presented. REDHOT calculated nominal temperatures without energy deposition are within 1-10% of reported results in literature, acceptable for the first-order analysis in this thesis. Energy deposition is observed to have a greater effect on the skin panel when it is already thermally and aerodynamically loaded. The panel is more structurally compromised for energy pulses of long duration, of higher magnitude and/or applied at times of strong aerodynamic loading. The experimental technique builds on existing electro-resistive heating techniques used for wind tunnel testing. Parametric studies were conducted to understand the design space and determine optimal panel thicknesses and direct-current application to maximise thermal-structural effects. A method to measure the induced thermal strain using digital image correlation was developed. To validate the experimental technique, a model with a 120mm by 80mm graphite panel with varying thicknesses was designed and tested on the bench. For the thinnest available plate, and a direct-current power supply of 350A material failure was not observed. Finite element modelling of the experimental conditions was conducted. Recorded temperatures were approximately within 9% of simulated results. Measured thermal strain was within 0.05% of simulated material.
-
(2020)ThesisCyberbullying is a very damaging aspect of social media platforms. This thesis explores how cyberbullying avoidance behaviours can be encouraged using persuasive design techniques in combination with real-time automated cyberbullying detection and design attributes that would be persuasive in deciding not to cyberbully. Close attention is paid to automatic detection using machine learning techniques and issues identified within the literature with machine learning techniques that use balanced versus unbalanced datasets. This thesis finds that neural networks were able to detect cyberbullying posts ac- curately in unbalanced datasets. Additionally, and in combination with more traditional classifier algorithms deployed in a hybrid solution, posts could be triaged faster and de- tection would work in a real-world scenario. This thesis then considers how automated detection of cyberbullying could aid a “messagebot” to give the appearance of monitoring cyberbullying in real-time and then derives a cyberbullying prevention model.
-
(2020)ThesisIn this thesis, a novel framework has been developed to quantify the radiation from tropical cyclones (TCs) in shortwave (SW) and longwave (LW) portions of the optical spectrum. The framework includes two stages: segmentation of TC clouds and calculation of the radiation attenuation due to TC clouds. The segmentation task was accomplished by an algorithm which takes a time series of brightness temperature images of tropical cyclones and uses image processing techniques to acquire segmentation for each image. The radiation was calculated by combining the segmentations with a radiation dataset provided by the Cloud and Earth's Radiant Energy System dataset via a coordinate-matching scheme due to their difference in resolution. The framework was used to investigate some preliminary results as part of a hypothesis that links the radiation due to TC and climate change. The framework was successfully implemented to analyse TCs' radiation in 2016, at the regional and global scales. Results show that TCs contributed a total of 21.25 TW of radiation to the global upwelling radiation, which is attributable by 152.27 TW in reflected shortwave radiative contribution and 131.02 TW in the emitted longwave radiative reduction. Although the radiation contribution from TCs was confirmed to be of the right order of magnitude to affect the Earth's Energy Balance, its impact on the balance would depend on how much the contribution varies, observed throughout a sufficient period of time. While this inquiry would require another comprehensive research, the framework remains the main contribution of the thesis. The framework has laid an important foundation for future work on TC radiation in general and for further insights into the impacts of TCs on climate change in particular.
-
(2022)ThesisThis thesis addresses a real power sharing method of electronically interfaced dis tributed generation (DG) units in the context of a multiple-DG micro-grid system where there is no synchronous generator. The emphasis is primarily on electronically interfaced DG (EI-DG) units like DFIG connected wind generator, PV system and battery storage. In this dissertation, the main goal is designing and testing of a new algorithm to distribute the load changes among intermittent distributed generators according to their power ratings. The power sharing by the commercial distributed energy resource (CDER) unit is mainly based on locally measured signals without communications. In the controller, proposed in this thesis, the voltage source inverter of the battery energy storage system (BESS) changes the frequency of the network with the change in load demand. The real power of each CDER unit is controlled based on a frequency-droop characteristic and a complimentary frequency restoration strategy. A systematic approach to develop a small-signal dynamic model of a multiple-DG micro grid, with a real power management method, is also presented. The model is used to investigate the sensitivity of the design to the changes in parameters and operatingpoint and to optimize performance of the micro grid system. Finally, the performance of the proposed algorithm is tested on a benchmark medium voltage network.
-
(2021)ThesisDisasters are hazards that damage the human environment. Disasters have had significant costs in the past and their financial impact is expected to increase in the future. Community Resilience (CR) is the ability of a community to resist, absorb, accommodate and recover from disasters (UNISDR 2015). The methods developed to evaluate CR are broad and varied, however all seek to enhance a community’s understanding of its abilities. By better understanding a CR, decision makers can make better informed decisions on how and where to apply limited resources. Emergencies and natural hazards have seen the extensive application of information communication technologies (ICT). The application of ICT in disasters includes their use to facilitate the exchange of information between emergency management agencies (EMA) and the public. To date, academic research in this field has focused on how social media (SM) platforms and data mining may be used to enhance EMA situational awareness during disasters and their ability to detect and respond to emerging disasters (Stieglitz, Mirbabaie, Fromm, et al. 2018). However, research into using analysis of social media within CR evaluation has been limited to date. This use of social media analysis for community resilience evaluation can be used to inform decision making in pre-event phases. This thesis seeks to explore whether the analysis of SM data may contribute to the evaluation of CR. To do this, it reviews existing CR evaluation frameworks and identifies possible SM variables for employment in CR evaluation. Subsequently, it reports on a case study that applies SM analysis to CR evaluation of three communities in Australia (the City of Lismore, Western Plains Regional Council, and Queanbeyan-Palerang Regional Council). This case study employs two Indicators from SM analysis (EMA influence and engagement within a community, and active participation of users with Institutions) to evaluate the Institutions Domain of CR. The performance of these Indicators are compared to the overall performance of the Institutions Domain of the Baseline Resilience Indicators for Communities (BRIC) (Cutter, Ash, and Emrich 2014) evaluation framework in the same communities and the performance of the communities in the Australian Natural Disaster Resilience Index (ANDRI) CR evaluation framework (Parsons et al. 2016). The results show that SM analysis may provide a method to enhance CR evaluation through the integration of SM Indicators.
-
Multi-vessel sea state estimation using artificial neural networks and vessel response spectral data(2021)ThesisThe use of ships as wave buoy analogies (SAWB) provides a novel means to estimate sea states, namely wave heights, periods, and directions. While single vessels have shown reasonable results, a previous study has shown improved estimations when using multiple vessels. However, no investigation into the effects of vessel trajectory on multi-vessel sea state estimation (SSE) performance has previously been undertaken. It was hypothesised that the guidance of multiple vessels acting as wave buoys for SSE can improve upon single vessel estimation. Further, it was hypothesised that variation of vessel trajectories will reduce SSE error as it changes the filtering types of the vessels, when compared with common trajectories. While several SAWB methodologies have been developed, the use of machine learning (ML) for SAWB has shown great promise in terms of improved estimation accuracy. Moreover, ML approaches to SAWB-based SSE are model-free, establishing relationships between causal wave properties and vessel motion responses using the input and output data alone. Therefore, neural networks have been designed to intake spectral vessel response data for heave, pitch, and roll, and output causal wave properties, with a high degree of accuracy when compared with similar approaches. As spectra can represent time-series information statistically, it was hypothesised that the proposed SSE methodology could be used to update SSEs online. However, although the spectra had similar shapes over time, the density of spectral ordinates was lower, leading to poor real-time estimation accuracy. Two uncertainty metrics were adopted to test their effectiveness as assessing SSEs, a time-based metric aimed at evaluating random errors, and a spectral agreement metric to assess bias errors. The results showed a combination of the metrics were able to associate uncertainty with SSE error distribution. To investigate the difference in SSE performance when multiple vessels move along different trajectories, three guidance strategies were implemented, and the resulting estimation errors were compared. Further, each multi-vessel strategy was compared with single vessel estimation. The multi-vessel guidance approaches were all found to reduce errors in the estimations over the single vessel SSE, with the strategies involving varied trajectories decreasing estimation error more than the common heading approach.
-
(2020)ThesisReal-world engineering design optimization problems involve constraints that must be satisfied for the design to be viable. The constraints are a manifestation of statutory limitations such as allowable strength, geometric compatibility or other practical considerations such as cost and time required for manufacturing. Constraint handling is therefore an integral part of design optimization. Population based algorithms are often used to solve complex problems with non-linear/black-box functions as objectives/constraints. However, they typically require large number of design evaluations to obtain near-optimum solutions. This may render them unpractical for problems where each evaluation involves a significant computational cost. In this context, this thesis targets to propose new strategies to solve constrained problems within fewer design evaluations. The proposed algorithms are designed for two different categories of problem formulations. In the first category, the study operates on conventional paradigm, where the objective and constraints are evaluated simultaneously and counted as one evaluation. For such problems, this thesis expedites the convergence of EAs by identifying smaller sub-spaces that potentially envelope the optimum using `bump hunting', followed by intensification of offspring generation in these regions. In the second category, which the thesis has larger emphasis on, the constrained problems are formulated differently from the majority of the existing literature. Here, each objective/constraint is assumed to be evaluated independently incurring an individual unit cost. Such scenarios reflect practical problems where each objective/constraint might be a result of a different, independently run, simulation. For this paradigm, the thesis aims to (a) develop novel strategies to determine the sequence for constraint evaluation, and (b) determine how best to use the available limited information through partial evaluation. Two strategies are proposed for constraint sequencing. The first is based on feasibility ratio computed using the already evaluated constraints. The second is a surrogate-based approach, where the likelihood of a solution violating a constraint is estimated based on predicted values. In both cases, the constraints most likely to be violated are evaluated first to establish feasibility in the least possible number of constraint evaluations. A comprehensive benchmarking is conducted to quantitatively demonstrate the computational benefits.
-
(2021)ThesisA Phishing attack is a type of cybercrime where individuals' personal and sensitive information is stolen by perpetrators for their financial gain. In this type of cyber-attack, the perpetrator pretends to be a genuine person or organisation by contacting them through email or other communication mediums. These malicious email links and attachments contain various malicious functions, which can capture victims' usernames, passwords, and online banking details. This is also known as an online identity theft as it harms individuals by not only stealing money but also the identity of victims. Previous research demonstrates that numerous anti-phishing tools have been developed to protect individuals' from being a victim of this kind of cybercrime. However, there is a very minimal amount of research on how to educate people. As phishing attacks are more central to humans, it is very important to educate them about anti-phishing as well as phishing. According to previous research, users' self-efficacy plays a vital role in phishing threat avoidance behaviour by motivating them. This self-efficacy has a co-relation with knowledge. This means one can enhance self-efficacy by enhancing knowledge. The study reported in this thesis focuses on user's self-efficacy to enhance computer users' phishing threat avoidance behaviour. The proposed research work is accomplished by first identifying knowledge elements that enhance IT users' self-efficacy. Then a theoretical model is developed that incorporates knowledge attributes such as observational knowledge, heuristic knowledge, and structural knowledge along with procedural and conceptual knowledge. The theoretical model depicts a mechanism that links knowledge attributes, user self-efficacy, threat avoidance motivation, and threat avoidance behaviour. A game design prototype based on a scenario was designed to demonstrate how investigated knowledge elements can be incorporated into an anti-phishing learning gaming tool. In addition, this also demonstrates how phishing education can be given and learned in a knowledge-based way by using an anti-phishing educational gaming tool. Finally, the research work reported in this thesis identified knowledge attributes that positively influence user's self-efficacy through their phishing threat avoidance behaviour. Therefore, it can be argued that anti-phishing education and educational tools should consider these knowledge attributes as well as IT users' self-efficacy.
-
(2022)ThesisDesign in the 21st century is a collaborative process that requires contributions from several specialty engineering disciplines across several Original Equipment Manufacturers (OEMs). Due to the competitive nature of design, it is in the best interest of the OEM to cast a wide net and harness expertise from across the world. The immediate trade-off seen by recruiting from such diverse elements is the decentralization of the workforce and resultant challenges for data sharing and platform interoperability. To capitalize on the advantages offered by the development of industry 4.0 and the enhanced ability to draw from a worldwide workforce it is essential that the interfacing of all design technologies is as seamless as possible. A single, but important aspect of such an interface is the ability to validate designs from a generic and globally updated model, at any time. Multidirectional translation between CAD packages and behavioural modelling software with the capability for integration of designs with optimisation packages allows for partial and complete designs to be validated, tested, and improved. Typical CAD to behavioural import methods utilise STEP / STL representations of the original CAD model. Whilst these provide accurate graphical representations of the original model, they can only be interpreted as rigid bodies and the ability to make modifications to the geometry outside of the native CAD domain is limited. In this work, a demonstration of a generic method for the integration of CAD and behavioural modelling software through the import of CATIA V5 models into Simulink is highlighted. The models are imported using a method that enables modification of component geometry within the simulation environment, enabling quick and meaningful decision support and opening a gateway to design engineering across unlinked CAD and Simulation software. The process presented enables standalone design evaluation and modification that can be used as an integral part of design development, supporting each decision with available and accurate feedback.
-
(2022)ThesisTo define a holistic and systematic approach to municipal waste management, an integrated municipal solid waste management (IMSWM) system is proposed. This system includes functional elements of waste generation, source handling, and processing, waste collection, waste processing at facilities, transfer, and disposal. Multi-objective optimization algorithms are used to develop an optimum IMSWM that can satisfy all main pillars of sustainable development, aiming to minimize the total cost of the system (economic), and minimize the total greenhouse gas emissions (environmental), while maximizing the total social suitability of the system (social). For the social objective, the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method is used to identify the main parameters that affect the social suitability of the system. This research focuses on developing an optimized holistic model that considers all four main components of a modern IMSWM namely transfer, recycling, treatment, and disposal. The model is formulated as a mixed-integer linear programming (MILP) problem and solved using the epsilon constraint handling method. A metaheuristic method is developed using non dominated sorting genetic algorithm (NSGA) to deal with larger problems. A solution repair function is developed to handle several equality constraints included in the proposed IMSWM model. Sensitivity analyses are conducted to identify the effect of changes in parameters on the objective functions. Based on the results, the proposed metaheuristic algorithm based on NSGA-II performed better than other algorithms. The interval-parameter programming (IPP) methods are used to consider various uncertainties that exist in the system. The model is applied to the case study of the Australian capital territory (ACT). The data is gathered from several resources including Australian national waste reports, and ACT government transport Canberra and city services (TCCS). Based on the waste characteristic and city map several feasible scenarios are recommended. Several non-dominated solutions are identified for the model that the decision-maker can choose the most desirable solution based on the preferences. Based on the importance of any objective function at any time the decision-maker can choose a solution to suit the needs.