48080 research outputs found
Sort by
An In-field Method for the Calibration of Instrumentation for Respirable Crystalline Silica Analysis
No full textCrystalline silica is a carcinogen which is associated with many diseases including silicosis, chronic obstructive pulmonary disease (COPD) and lung cancer. These diseases come about because of exposure to and inhalation of respirable crystalline silica (RCS) which is produced during work activities such as tunnelling, blasting, construction, quarrying, cutting, and polishing engineered stone benchtop materials. For dust particle to be deposited in the lungs of humans, it depends significantly on the particle size or aerodynamic diameter. Dust particles are referred to as respirable when they have aerodynamic diameter less than 10 μm and a median cut point (d50) of 4 μm. These particle sizes can penetrate the alveoli. Some of these activities release dust that contains a very high concentration of RCS. When workers are continuously exposed to these size fractions, and they inhale them over time it affects their health and their ability to carry out tasks effectively. These result in them being retired early from their work and battling with the effects of their conditions such as silicosis for the rest of their lives To measure the RCS concentration in dust in a workplace, dust is sampled onto filters, which are normally sent to commercial laboratories that are off-site and in certain countries the testing laboratories are in another country. The results can take up to 6 weeks before they are received by which time the work environment has usually changed due to various factors, such as some workers may have been moved from the work area to another, production in the area sampled may have ceased and moved to a different section of the work site. Workers may also have been exposed to silica from the dust as production has continued while waiting for the results of the samples previously collected. The situation above is common in developing countries and also remote sites in developed countries all of whom would benefit from more timely and if possible, accurate onsite RCS analysis. If such data was available, operations could use it in their risk assessment process thus, implement enhanced health and safety procedures before workers are potentially exposed.A systematic literature review was undertaken to investigate previous research that has attempted to address the issue of timely onsite RCS measurement and identify gaps that this research has addressed. The gaps identified were the paucity of research around standardized method of dust collection for in-field RCS analysis and the need for more research around in-field RCS analyses and the use of portable RCS instrumentation. Also included were the absence of a calibration method for RCS measuring instrument that can be deployed in-field, and the absence of a standardized approach to overcoming the impact of interferences present in host materials.</p
Automated Change Impact Analysis for Deep Learning Programs
No full textDeep learning (DL) is the core technology underlying many Artificial Intelligence (AI) applications. Maintaining deep learning software programs presents unique challenges due to their complex, interdependent structures. Traditional Change Impact Analysis (CIA) methods, while effective for conventional software, fall short when they are applied to deep learning programs due to the intricate interdependencies within neural networks, particularly those related to the unique architectural layers and hyperparameters. As AI deep learning applications continue to expand in many fields such as healthcare, finance and autonomous systems, maintaining these software programs becomes challenging since any modification can have far-reaching impacts on overall performance and stability. This research presents a novel approach to CIA tailored specifically for deep learning programs. We introduce a comprehensive taxonomy that systematically categorizes changes in deep learning architectures, hyperparameters, and regularization techniques.Our approach leverages this taxonomy to map changes automatically, thereby enabling structured and efficient impact analysis in complex deep learning programs. To predict the potential impact of modifications, the framework incorporates co-change pattern mining to identify and generate co-change rules, capturing dependencies based on historical co-change patterns. We also designed domain-expert rules derived from relevant literature, which encode specific co-change patterns pertinent to neural network structures. The framework’s hybrid approach combines these co-change patterns and domain-expert rules to generate accurate predictions, aligning data-driven insights with domain-specific knowledge. Additionally, we incorporate a Large Language Model (LLM) to enhance the semantic understanding of these changes, improving recall by detecting relevant impacts while addressing limitations in conventional CIA methods.To validate this approach, we constructed a dataset from GitHub commits in Keras-based projects, using real-world modifications to test the predictive accuracy of our CIA framework. Our evaluation shows that the hybrid approach significantly enhances recall, achieving an average recall of 0.932 (93.2%) at optimal support and confidence thresholds. The LLM-based component demonstrated an average recall of 0.656 (65.6%) but with limited precision. These results highlight the utility of a combined approach, capturing a broad range of impactful changes while minimizing irrelevant suggestions.The contributions of this thesis are multi-faceted: a deep learning-specific taxonomy for Change Impact Analysis, an automated mapping of changes within this taxonomy, and a recommendation system for impact prediction. The findings suggest that the proposed CIA approach has significant potential to support AI software developers in managing deep learning model evolution by providing actionable insights and recommendations, consequently reducing the risks associated with software updates.</p
The problem with ports: A potted history of ports and port legislation in NSW, the evolution of legislative port charges, the position following the port privatisations in NSW and the case for legislative reform
No full textPorts in NSW have been of vital importance since the establishment of the new colony in Sydney Cove over 230 years ago. However, ports need to be constructed and maintained. Wharves need to be built. Channels need to be dredged. Navigations aids need to be put in place. Constant maintenance and repairs need to be carried out in the harsh marine environment. Pilots need to be available to guide ships. Naturally, all of this has to be paid for somehow.Government expenditure in relation to a port has historically been recouped through the imposition of legislative port charges. These typically apply to the users of ports and initially in NSW often reflected the historical practice in various parts of England (such as at the wharves on the River Thames in London). A number of these legislative port charges are still applied today, with some still having the same name as was used centuries ago in England.Both public and private sector investment in our ports has been commonplace since the early days of the new colony in Sydney Cove. Indeed, without private sector investment the development of early port infrastructure (and indirectly, Sydney itself) would not have been possible. Indeed, public and private sector investment in our ports has continued until the present.A review of the evolution of legislative port charges in NSW is necessary to properly understand how we ended up where we have. Amongst other things, this review shows that certain challenges (which legislative port charges attempt to address) are timeless. This includes how to recoup expenditure on port infrastructure on an equitable basis, particularly given the private ownership of certain port infrastructure (commonly wharves). Historically, Government carried out various “wharf privatisations” where a leasehold interest in a wharf (together with either the legal right or practical ability to impose legislative port charges) was auctioned to the private sector.However, the privatisation of an entire port was something that did not occur until more recent times and in 2013 and 2014 the three major ports in NSW (Port Botany, Port Kembla and Port of Newcastle) were privatised. As will be seen from the review of historical port charges legislation, the relevant port charges legislation in place at the time was not well suited to such port privatisations. In fact, even without the port privatisations this legislation was unclear in various important ways. Unfortunately, the port privatisations (and the amendments to the legislation that occurred in the lead up to the port privatisations) did not fix the existing legislative problems. In fact, the amendments made the problems worse.These legislative problems exist within a certain geographical, logistical and economic context in NSW which in turn exacerbates some of those problems. This thesis examines the problem with ports from a historical perspective and shows why legislative reform is required.</p
Advancing Battery State of Health Estimation and Remaining Useful Life Prediction Through Deep Learning
No full textLithium-ion batteries are crucial for modern energy storage applications, particularly in electric vehicles and renewable energy systems. However, these batteries inevitably degrade with repeated charging and discharging, resulting in capacity loss, reduced power output, and safety concerns. Accurate estimation of battery state of health (SOH) and prediction of remaining useful life (RUL) are essential for reliable battery management. Existing methods, including model-based and data-driven approaches, often face challenges such as limited data availability, high computational costs, and inadequate generalisation across different battery chemistries and operating conditions. This thesis presents a series methods to systematically tackling these challenges by developing advanced deep learning frameworks for improved SOH estimation and RUL prediction.First, a novel battery ageing tag-free training (BATT) method is introduced for SOH estimation using only incomplete charging curves. By eliminating the need for complete charge-discharge cycles, BATT reduces the data collection burden while maintaining high prediction accuracy. The method is validated across multiple datasets, demonstrating its effectiveness in estimating battery health indicators with minimal data requirements.Second, a semi-supervised learning approach is developed for RUL prediction, leveraging both labelled and unlabelled data to enhance model generalisation. The proposed architecture incorporates an encoder-decoder structure with multiple decoder heads, enabling robust feature extraction from partially labelled datasets. Comparative analyses with supervised learning models show that this approach improves prediction performance while addressing data scarcity issues.Further extending the semi-supervised learning paradigm, this thesis proposes a federated learning method to enable multiple battery users to collaboratively train RUL predictive models without sharing raw data. This method preserves data privacy while enhancing model generalisation across diverse battery chemistries and operating conditions by incorporating massive unlabelled data. Experimental results confirm that federated semi-supervised learning outperforms traditional centralised learning approaches in terms of RUL predictive performance.Finally, a physics-informed machine learning approach is proposed to incorporate domain knowledge into the RUL prediction process. By integrating physically meaningful ageing parameters into deep learning models, this approach improves interpretability and prediction accuracy. The incorporation of domain-specific features reduces reliance on purely data-driven techniques, offering a more generalisable and efficient solution for battery health monitoring.Overall, this thesis addresses the challenges in short- and long-term battery degradation prediction, including the scarcity of run-to-failure test data and the need for privacy protection. Domain knowledge is further unlocked to enhance the performance of traditional deep learning methods in degradation prediction. The proposed methodologies contribute to the development of scalable, data-efficient battery health management systems, paving the way for more reliable and sustainable energy storage solutions.</p
Privacy-preserving federated semi-supervised learning for battery life prediction amid data scarcity
No full textAccurate prediction of remaining useful life (RUL) is essential for effective battery management and lifespan optimisation. While recent machine learning approaches offer promising results, their development relies heavily on abundant degradation data with RUL labels, which requires costly run-to-failure tests lasting years. Although massive degradation data are available from millions of batteries in laboratories and in service, access to such data is often restricted due to privacy concerns. Additionally, they usually suffer from quality issues, particularly the absence of RUL labels. To address these issues, we propose a federated-based semi-supervised learning framework enabling collaborative training among diverse battery users that own limited degradation data with RUL labels. This method not only enhances battery RUL prediction by effectively utilising low-cost routine operational data without RUL labels but also protects data privacy across battery users through secure model parameter aggregation. The proposed method is validated on two battery degradation datasets comprising 40 batteries cycled over 24,900 times. Comparative evaluations against federated learning (FL), semi-supervised learning (SSL), and supervised learning (SL) methods are conducted to highlight the effectiveness of our method. Results show that the FL, SSL, and SL methods achieve root mean squared errors (RMSEs) of 27.1, 33.8, and 40.1 cycles, respectively. In contrast, the proposed method achieves an RMSE of 21.3 cycles, resulting in reductions of 21.4 %, 37.0 %, and 46.9 %. This work underscores the potential of federated semi-supervised learning as a practical solution for accurate RUL prediction with reduced battery tests while addressing privacy concerns.</p
Environmental Engagement: An Effective Way in Mitigation of Firms' Litigation Risks in China
No full textThe increasing incidence of environmental disasters has necessitated stricter government environmental legislation and heightened firms' environment-related litigation risks. Firms with environmental engagements receive favourable treatment from the government as their efforts align with government environmental campaigns and comply with government initiatives. In this study, we examine the impact of firms' environmental engagement on their litigation risks in China. Using a sample of listed firms from 2006 to 2019, we find that firms with environmental engagement are less likely to experience lawsuits. This result is robust across alternative estimation methods. The main results are also more pronounced in non-state-owned enterprises (non-SOEs), firms from high-pollution industries, located in high-pollution provinces and low-trust provinces. Overall, this study provides evidence from a new perspective on how firms can mitigate their litigation costs through environmental engagement.</p
Metal hydride hydrogen storage risk assessment: A review
No full textHydrogen is gaining global attention as a green and clean fuel because of its production through multiple sustainable sources. The hydrogen economy consists of production, storage, delivery, and applications. Hydrogen has some associated risk hazards potential for fire and explosion such as the wide flammability range in air, high flame speed, and low minimum ignition energy. Therefore, this paper conducts a review and analysis of related challenges in hydrogen safety management. This review study discusses and then summarizes the current status of hydrogen storage technologies with the main focus on challenges in applications of metal hydrides. This review covers the technologies related to risk assessment of hydrogen storage in metal hydrides. The quantitative risk assessment is briefly discussed along with an analysis of past risk studies in hydrogen release and dispersion. As mentioned in several literature review studies, the scarcity of hydrogen-specific incident data limits the validity of risk analysis findings. With this context, the current status of hydrogen accident data is discussed, and the possible paths in data collection are proposed. Finally, some recommendations and potential areas for future research are given for developing new storage technologies and improving the safety of hydrogen storage systems.</p
Place attachment and residential water conservation: Application of an extended Theory of Planned Behaviour in Australia's Northern Territory
No full textUnderstanding the drivers of residential water conservation is crucial amid rising climate change, drought, and water shortage concerns. While the literature suggests that environmental psychology can help identify the factors influencing water-saving behaviour, the critical role of place attachment in water conservation decision-making remains underexplored. Previous research has primarily focused on identifying key determinants through statistical significance and correlations, yet there is still limited understanding of the extent to which these determinants are necessary conditions for achieving water conservation. Thus, this study draws on necessity and sufficiency logic to investigate place attachment as a critical antecedent to an extended Theory of Planned Behaviour (TPB) to predict residential water conservation intention and behaviour. By integrating partial least square structural equation modeling (PLS-SEM) and necessary condition analysis (NCA) to analyse 346 responses from residents in the Northern Territory of Australia (NT), this study identifies the sufficient and necessary conditions that need to exist for residential water conservation intentions and behaviours to occur. The PLS-SEM results show that place attachment does not significantly predict water conservation intentions or behaviours, rather it influences attitude, injunctive norm, descriptive norm, response efficacy, and self-efficacy. While attitude, injunctive norm, response efficacy, and self-efficacy were significant predictors of intention, descriptive norm was not. Further, intention and self-efficacy were also significant predictors of water conservation behaviour. The NCA results revealed that place attachment, attitude, self-efficacy, and response cost were necessary conditions for water conservation intention, while place attachment, self-efficacy, and intention were necessary conditions for achieving water conservation behaviours. Overall, the implications of this study for theory and practice are proposed to guide scholars, policymakers, and water authorities in enhancing residential water conservation.</p
A qualitative exploration of Australian patients’ and parents’ perspectives in optimizing the management of eosinophilic esophagitis
No full textNo description supplied</p
The Impact of Exogenous Stimuli on hiPSC-Derived 3D Neural Models
No full textReplication of biomimetic and clinically relevant neural in vitro models is necessary for physiological and pathophysiological understanding of the human brain. Despite recent advances in human pluripotent stem cell (hiPSC)-derived three-dimensional neural tissue modelling, particularly through organoids and bioprinted constructs, limitations around translation remain. While current models effectively incorporate biochemical stimuli, the specific influences of electrical and mechanical parameters on neural subtype development remain poorly understood. This thesis investigates how electrical stimulation parameters and mechanical cues influence striatal neural subtype development, with implications for both enhanced in vitro modelling and therapeutic applications.To address this, a working model of a human iPSC-derived striatal organoid model was optimised for heterogeneity, containing medium spiny neurons (MSN) and interneurons of the lateral (LGE) and medial ganglionic eminences (MGE). This revealed that initial seeding density significantly impacts region-specific differentiation, with higher densities preferentially generating LGE-originating neurons. Notably, Activin A supplementation induced the first reported instance of morphogen-driven organoid gyrification, characterised by the expression of surface folds with increased cellular density.This striatal model was deployed for experimental work in validating a high-throughput electrical stimulation platform and subsequent investigation of frequency-dependant impact on neural subtype development. Hereby, the influence of high and low frequencies (1 – 210 Hz) was assessed, in line with intrinsic neural oscillations and clinically applicable stimulation parameters. Distinct effects on neural organisation and subtype-specific signalling were observed at later developmental stages. Hereby, GABAergic signalling of the striatal interneurons was significantly enhanced after low-frequency (1 Hz) stimulation, combined with an increase in NMDA receptor expression. Higher frequency stimulation enhanced cellular organisation with distinct striatal marker clustering. These frequency-specific responses provide a molecular basis for understanding how varying stimulation parameters differentially affect neural circuits of neural subtypes with distinct electrophysiological profiles.As a further aspect influencing the suitability and translatability of 3D neural models, the impact of mechanical modulus and biomaterial degradation on neural behaviour was investigated. This was achieved through optimisation of a GelMA/gelatin (4%/4%) hydrogel, and established a platform for examining how substrate properties affect neural development. The modifiable hydrogel composition demonstrated soft mechanical properties conducive to neural tissue development and network formation, while maintaining rheological characteristics compatible with 3D printing parameters. These results enable future studies of mechanical gradients on neural subtype specification.Taken together, these findings advance the understanding of the biophysical and bioelectric influences on neural development and behaviour in vitro. The identification of frequency-dependent responses, particularly the enhancement of GABAergic signalling through low-frequency stimulation, demonstrates the need to adjust stimulation parameters to the unique cellular profiles of the target tissue. The behavioural impact from mechanical forces exerted on a cellular and molecular level, through the choice of modelling technique, further supports the integration of multiple parameters for enhanced in vitro modelling.Overall, the integration of select electrical, mechanical, and biochemical stimuli guides the development of more clinically applicable and reproducible 3D neural tissue models. Combined with the evaluation of neural-subtype specific responses to frequency-dependent stimulation in patient-derived models, this offers a foundation for optimising clinical stimulation parameters towards a targeted therapeutic approach that is particularly valuable in neuropsychiatric disorders with high heterogeneity.</p