https://bura.brunel.ac.uk/handle/2438/23 2026-08-22T02:11:24Z 2026-08-22T02:11:24Z Fang, Jingzhong https://bura.brunel.ac.uk/handle/2438/33697 2026-08-14T02:00:47Z 2026-01-01T00:00:00Z

Title: Intelligent outlier detection under imperfect industrial data conditions Authors: Fang, Jingzhong Abstract: With the growing complexity and data intensity of modern industrial systems, intelligent data analysis (IDA) has become essential for reliable data interpretation and efficient operation. By leveraging advanced analytical and computational techniques, IDA can handle complex, largescale data sets, thereby providing valuable insights for industrial operations and significantly improving the stability and efficiency of industrial processes. Nevertheless, data quality remains a fundamental prerequisite for IDA performance, as it directly affects the accuracy and trustworthiness of derived insights. Due to complex operating conditions, high cost, and limited availability of expert annotation, high-quality data is often difficult to acquire. The imperfections in the data can affect training and evaluation, thereby reducing the overall performance in real-world applications. In practice, imperfections in industrial data generally fall into two main categories: 1) data scarcity, such as class imbalance or limited sample size; and 2) label quality issues, such as missing labels or noisy labels. In this thesis, we deal with the above-mentioned data imperfections arising from the data acquisition process and the annotation process to develop innovative approaches for robust and reliable IDA in industrial applications. It should be pointed out that all approaches developed in this thesis have been evaluated and applied to outlier detection on imperfect industrial data collected from real-world wire arc additive manufacturing (WAAM) processes. • To achieve outlier detection on unlabeled data, an improved optimization-based clustering algorithm is proposed, where the initial locations of the cluster centroids in the fuzzy C-means algorithm are optimized by an improved particle swarm optimization (PSO) algorithm. An adaptive switching randomly perturbed particle swarm optimization (ASRPPSO) algorithm is developed to enhance the convergence rate and particle’s search ability of the PSO algorithm, where a distance-based weighting strategy and switching strategy are introduced to update parameters of the optimizer. The particles can conduct a thorough search by accounting for both evolutionary states and the distances to the global and personal best positions, thereby improving the convergence rate and solution accuracy. Via the ASRPPSO-based selection of optimal clustering centroids, the proposed algorithm does not rely on centroid initialization, thereby facilitating a better cluster partition. • With the aim to guarantee the outlier detection performance on imbalanced data under the small sample problem, an optimized deep transfer learning framework is developed, where a novel deep domain adaptation strategy is designed to minimize the cross-domain discrepancies, the weighting factors are designed to handle the data imbalance problem, and the PSO algorithm is utilized for hyper-parameters tuning. By leveraging the domain knowledge and optimal hyper-parameter selection, the developed framework effectively balances performance and efficiency when dealing with outlier detection tasks on imbalanced data under the small sample problem. • To handle the noisy label problem under limited-data conditions in outlier detection, a novel Transformer-embedded learning with noisy labels framework with fuzzyclustering- assisted contrastive learning (TFCCL) is developed, where a fuzzy-clusteringassisted contrastive learning approach, a dynamic two-stage training scheme and a joint learning strategy are introduced to train the outlier detector. The TFCCL framework integrates supervised learning with contrastive learning, thereby reducing reliance on potentially noisy labels and enhancing model robustness. • For the purpose of outlier detection under the noisy label problem when sufficient data are available, a role-differentiated learning with noisy labels (RD-LNL) approach is put forward, where a leader-follower-inspired sample selection (LFSS) strategy is proposed for identifying potential clean samples for model training. A selection metric and an adaptive selection scheme are designed to adjust the number of selected samples. By combining the joint learning and adaptive sample selection, the RD-LNL framework achieves robust outlier detection in the presence of noisy labels. • To fully evaluate the application potential of the developed frameworks. All the developed frameworks are applied to outlier detection tasks on WAAM data sets collected from real-world manufacturing processes to examine their adaptability, robustness, and generalization capability. The experimental results demonstrate the effective and robust performance of the developed outlier detection frameworks on WAAM data sets across various data imperfection conditions. Description: This thesis was submitted for the award of Doctor of Philosophy and was awarded by Brunel University London

2026-01-01T00:00:00Z Horne, Sara Louise https://bura.brunel.ac.uk/handle/2438/33684 2026-08-13T02:01:20Z 2026-01-01T00:00:00Z

Title: Methodological advances in injury surveillance and injury inciting event analysis in elite UK netball Authors: Horne, Sara Louise Abstract: Injury remains a persistent challenge in netball across all levels of participation, yet limitations in how injuries are defined, captured and described have constrained injury prevention efforts. To address these limitations, this thesis aimed to develop and apply robust methods to enhance injury surveillance and injury-inciting event analysis in elite UK netball, thereby improving methodological consistency and strengthening the injury prevention evidence base. To meet this aim, four interrelated studies were undertaken that synthesised existing evidence, implemented prospective injury surveillance, developed and evaluated a netball-specific injury-inciting event classification system, and piloted a preliminary biomechanical domain to support the system’s future extension. A scoping review demonstrated substantial methodological inconsistency in netball injury research, particularly in injury definitions, exposure measurement and inciting-event reporting. This informed recommendations to improve standardisation and guided the methodological approach adopted in subsequent studies. Prospective injury surveillance in the Vitality Netball Superleague (VNSL) showed match injury incidence was considerably higher than training, with match injuries predominantly acute lower-limb conditions resulting in high time-loss burden, whereas training injuries were largely overuse-related and frequently non–time-loss. Subsequently, the Netball Injury Inciting Event Classification System (NIIECS) was developed and evaluated to classify situational, behavioural and contextual characteristics of injury events using video-based or non-video (athlete- or clinician-reported) data sources. Acceptable inter- and intra-rater reliability and applied usability were demonstrated for video-based match injury events. Finally, a preliminary biomechanical domain was proposed to extend the NIIECS, with expert feedback indicating strong support for the inclusion of biomechanical descriptors while identifying key priorities for refinement to support future development and integration. Overall, this thesis advances methodological standards for injury surveillance and injury inciting event analysis in netball. It provides an improved foundation for consistent, context informed injury monitoring and interpretation in elite UK netball, supporting evidence informed decision-making to guide injury prevention efforts. This work also establishes a platform for future refinement, including biomechanical integration, to inform targeted injury prevention strategies. Description: This thesis was submitted for the award of Doctor of Philosophy and was awarded by Brunel University London

2026-01-01T00:00:00Z Ekhareafo, Ushemegbe Rita https://bura.brunel.ac.uk/handle/2438/33572 2026-07-13T12:23:29Z 2025-01-01T00:00:00Z

Title: Environmental assessment of chemical contamination from abandoned coal and mineral mines Authors: Ekhareafo, Ushemegbe Rita Abstract: Abandoned mine lands (AMLs) are enduring sources of environmental pollution due to the release of toxic substances like metals, metalloids, organic and organometallic pollutants into the environment. These substances are transported through various processes such as acid mine drainage that aid the movement of pollutants across compartments and are driven by the pollutants' physicochemical properties, environmental conditions and the physical state of the abandoned mine. Chemicals released can be toxic, bioaccumulate, and persist; hence, highly polluted mine lands may pose significant ecological and human health risks. Worldwide, it is estimated that there are over a million abandoned mines. Nigeria is no exception, with over 1500 abandoned mines identified to date across many states, and so far, only a few have been reclaimed due to the high cost of restoration. This research focused on the environmental assessment of chemical contamination from abandoned coal and mineral mines in Nigeria. A systematic evidence map protocol was developed to evaluate current methods for assessing chemical contamination risks from abandoned coal and lead-zinc mines. The protocol was applied to review the evidence, which revealed that current AML risk assessment methods use contaminated land risk indices to evaluate chemical risk. Furthermore, the review identified evolving data-driven and probabilistic risk assessment methods, including logistic and linear regression models, which are emerging risk assessment approaches in AML chemical risk assessment. At the same time, Bayesian Kernel Machine Regression (BKMR), cumulative probability distribution analyses for HI, and Incremental Lifetime Cancer Risk (ILCR) have been reported for mixture risk assessment, and the study showed the relevance of bioindicator-based assessments for validating predictions from chemical-based assessments. A major aim of the project was to assess the chemical contamination and associated ecological and human health risks at coal and lead-zinc mines in Nigeria. In this phase of the research, environmental samples were collected from two abandoned lead-zinc mines at the Gimbi/Rikaya site in Plateau and the Abakaliki site in Ebonyi. The case study on abandoned coal mines was undertaken in Enugu, Nigeria. Site-specific contamination and ecological risk assessment at the Abakaliki and Gimbi/Rikaya sites revealed that both sites were heavily contaminated with Pb, Zn, and Cd, with higher concentrations at Abakaliki. Spatial distributions followed the pattern tailings > sediment > soil, with tailings as the primary contamination source. Sequential extraction confirmed that Pb, Cd, and Zn occur mainly in the exchangeable and reducible fractions, indicating high mobility and bioavailability, particularly in Abakaliki soils, where >34% of Pb occurred in mobile forms. In contrast, Pb mobility was lower at Gimbi, with only about 6% of F1 in the soil. Dissolved metals, Pb, Zn, Ni, Cu, and Mn in surface water at Abakaliki exceeded the environmental quality standards (EQS) and also the Biotic Ligand Model (BLM) bioavailability values. In contrast, only Zn exceeded EQS and bioavailability values at Gimbi. Ecological risk indices (Er, RI) and cumulative risk scores (CRS) indicated very high risk at Abakaliki and considerable risk at Gimbi, with Cd, Pb, and Zn as the dominant risk drivers. Chemical contamination and human health risk assessment (HHRA) at the Onyeama abandoned coal mine revealed the presence of both heavy metals and polycyclic aromatic hydrocarbons in soil, tailings, and water at varying concentrations. Fe, Mn, Zn, and Pb were the dominant metals of concern, reflecting inputs from historical mining residues and secondary weathering. The ΣBaPeq value (0.86 mg TEQ/kg) exceeded the Canadian Soil Quality Guideline (0.6 mg TEQ/kg), with BaP and DBA contributing substantially to the toxic load. The HHRA indicated ingestion and dermal contact as the dominant exposure pathways. Waterborne exposure contributed the highest non-carcinogenic and carcinogenic risk, with cumulative HI and CR values exceeding acceptable thresholds, particularly for children, highlighting their vulnerability. In contrast, soil and tailings contamination presented relatively lower cumulative HI and CCR. Overall, these findings confirm that abandoned Pb/Zn and coal mines pose both ecological and human health risks to inhabitants living on or near abandoned mines, pointing to the need for targeted management measures, including the containment of tailings, remediation of contaminated sediments, and monitoring of surface and groundwater pathways, to mitigate long-term ecological and human health impacts. Description: This thesis was submitted for the award of Doctor of Philosophy and was awarded by Brunel University London

2025-01-01T00:00:00Z Vahid Dastjerdi, Leyli https://bura.brunel.ac.uk/handle/2438/33570 2026-07-13T14:06:53Z 2025-01-01T00:00:00Z

Title: Optimizing of bread quality and sustainability: Evaluating the impact of conventional, microwaves and hybrid baking methods Authors: Vahid Dastjerdi, Leyli Abstract: Baking is a final and one of the most important stages in the breadmaking process, shaping product quality, energy performance, and overall sustainability. Conventional baking methods is widely adopted but it is energy-intensive and often results with uneven heat distribution. This study evaluates the effects of three baking methods including conventional oven (CB), industrial solidstate microwave (IM), domestic microwave (DM) baking, and evaluates the sensorial and physicochemical properties of the resulting breads including moisture content, hardness, colour, specific volume, cell structure, and acrylamide levels, as well as energy consumption, cost, and greenhouse gas (GHG) emissions. A hybrid baking approach, combining CB with microwave baking (applied separately for IM and DM), was also investigated, and the resulting breads were analysed for the same quality attributes. In addition, temperature distribution and moisture variation were monitored during the baking processes for all baking modes. Response Surface Methodology (RSM) was employed to optimise the baking parameters and identify the best baking conditions to maximise improved quality and sustainability performance. The results indicated that IM baked bread achieved the highest moisture content, while domestic microwave led to pronounced moisture loss. Colour analysis revealed that microwave- baked samples developed lighter crusts, whereas CB produced darker crusts due to the higher surface temperatures. Texture analysis showed that IM baking generated softer bread, while DM baking resulted in significantly highest hardness. Although conventional baking achieved the highest specific volume, microwavebaked breads exhibited more irregular crumb structures and lower acrylamide levels. Among all baking modes, IM baking proved to be the most energy-efficient and cost-effective, generating the lowest GHG emissions and therefore representing a more sustainable alternative. It also provided rapid and uniform heating with improved moisture preservation, whereas DM baking caused surface drying and poor internal hydration. Hybrid baking, particularly when incorporating IM, improved thermal uniformity and moisture preservation compared to CB. RSM optimisation further identified IM as the optimal method, achieving highest moisture retention, low hardness, and a desirability score of 0.865. Description: This thesis was submitted for the award of Doctor of Philosophy and was awarded by Brunel University London

2025-01-01T00:00:00Z