Impedance pumping is a simple valve-less pumping mechanism which offers a low energy, low noise alternative for conventional pumping at micro and macro scales. When a fluid-filled elastic tube is connected to rigid tanks, a net flow in either direction can be induced by periodically compressing the elastic section asymmetrically from the ends. The flow generated is due to the mismatch in acoustic impedance between the elastic and rigid regions. Prior works show a complex non-linear behaviour of the flow in response to the compression frequency, including distinct resonance peaks and reversals in flow direction. The research study explores the physics behind the fluid flow through a tube caused by impedance difference and derives a simple yet useful mathematical model for Two-tank system based on Energy conserving compartment model (ECCM). Parameters investigated include pinch frequency, pinch location, and physical parameters of rigid tank and flexible tube. Material properties such as elastic tube radius and tube thickness have been shown to have role in net flow characteristics under external pumping. With larger tube radius and smaller tube thickness, substantial increase in net flow generated has been achieved at lower external pumping frequency. Phase synchronization between external pumping pressure and fluid flux at the junction between elastic tube and tanks is identified as a key factor in determining the direction of net power. The model derived is extended to Multi-tank system and the flow behaviours under similar conditions have been compared to Two-tank system. Introduction of middle tank has affected the flow characteristics of Open-tank impedance pumping. With respect to terminal tanks, symmetry in net flow characteristics with equal magnitude but opposite direction has been observed for pinching locations of same distance from both ends of the tube. For such case in Multi-stage system also exhibited net flow of same magnitude and same direction with respect to middle tank. .Multi-Stage model could generate higher net flow rate compared to Two-tank system in all equivalent cases due to the capacitance effect of middle tank making it a promising advancement in valve-less pumping. Experimental works were conducted to validate the developed mathematical model. Experiments using Two-tank model achieved highest net flow with tank level differences of 1.10cm and-1.15cm for pumping positions C=15 and C=4 respectively while equal level difference of 1.07cm achieved in both directions using numerical methods. Experiments involving Multi-tank model has shown a maximum net flow rate of 1.81cm and-2.06cm for positions C15|15 and C4|4 respectively while height difference of 2.19cm in both directions has been achieved using numerical simulation. Both methods of investigation demonstrate resonant excitation frequency to be around 7.2-7.3 Hz. It has been established using numerical and experimental data that the net flow direction caused by a particular excitation frequency for a given pinching position, the value of resonant frequency, and the flow behavior of impedance pumping has been well described using a simple, yet reliable numerical model developed. Due to its simplicity; it could be promising to extend to more complicated structures such as in coronary blood flow where branching and varying geometries are present or in designing a more complex controller for valve-less pumping model.

In the modern world, systems such as aircraft systems are becoming increasingly complex, often consisting of a large number of components. As no component is perfectly reliable, they can fail, some in many different ways, leading to a large number of potential component failures on complex systems. Component failures can have detrimental effects on the performance of the system, with some component failures even causing system failure, potentially damaging the system, or more importantly, potentially endangering human life. In order to be able to detect component failures on complex systems, the inclusion of sensors is becoming increasingly common. In addition to being able to detect component failures, the sensors can be used to diagnose component failures, with certain symptoms and the resultant sensor readings being produced by certain component failures. Another benefit is that it may also be possible to prevent system failure by detecting component failures early, activating redundant components and enabling the mission to be completed. However, including sensors in the system increases the cost of the system, can add weight to the system and require space for installation, a factor of particular importance for weight critical systems, such as aircraft systems. Therefore, a balance between being able to detect and diagnose failures in systems and the cost, weight and space requirements of the sensors needs to be achieved. In this thesis, a novel sensor selection methodology is proposed, which is based on a performance metric. Individual sensors, and combinations of sensors are ranked based on their performance of detecting faults and diagnosing failures in the system. In addition to the sensors’ detection and diagnostic performance, the metric also considers the effect that the component failures have on the functionality of the system, where sensors that detect critical failures are favoured over sensors that do not detect such failures. The performance metric is then extended to consider the time taken to detect and diagnose component failures, as the sooner component failures are detected, the more likely system failure can be prevented. This is important in a system that operates in a phased mission. In addition, a proposed two-level Genetic Algorithm is used in order to efficiently determine a suitable combination of sensors for larger systems, where an exhaustive calculation of the performance metric for all combinations of sensors is not feasible. For a simple flow system, a Bayesian Belief Network (BBN) is used to model the effects of component failures, and sensor readings. During the fault diagnostic process, observed sensor readings can be introduced in the BBN, which then can be used to identify the failed components. However, an alternative system modelling and fault diagnostic technique is proposed as a part of this thesis which can be used on larger systems, and can determine sensor readings and component failures more quickly than the BBN method. This method is based on a series of if-then-else statements in order to determine the effect that the component states have on the performance of the system. The work proposed in this thesis is applied to three example systems: a simple flow system, an example aircraft fuel system and the fuel system for an Airbus A380-800 aircraft.

The work described in this thesis explored the manufacturing of fully resorbable high strength glass fibres as the reinforcement of poly-ε-caprolactone (PCL) composites and attempted chitosan and 3,4-Dihydroxy-L-phenylalanine (L-DOPA) as coupling agents to improve the interface of PGF/PCL composites. Phosphate based glasses (PBGs) in the system of P2O5-B2O3-Na2O-CaO-MgO-Fe2O3 were studied on their structural, thermal and degradation properties. The results of the experiments conducted revealed that the glass transition temperature, density and chemical durability were increased with increasing iron oxides in PBGs at the expense of CaO/MgO. The analysis of glass stability indicated that higher CaO/MgO ratios would have a positive effect on the fibre drawing from the glasses investigated. Fibres were produced from the glasses investigated using a melt drawn system and the effect of composition on the mechanical and degradation properties of the fibres were evaluated. Addition of Fe2O3 showed a significant improvement on the mechanical properties (tensile strength and modulus) and chemical durability of the fibres. Annealing treatment also enhanced the Young’s modulus and chemical durability of the fibres, while a reduction in tensile strength of the fibres was observed after annealing. Chitosan (CS) dissolved in dilute acetic acid solution was observed to significantly improve the interface of PGF/PCL composites. Coating of CS also showed a significant effect on glass surface protection, resulting in a maintaining of mechanical strength of the fibres in dilute acid solution. TG, SEM, FTIR, Raman and XPS were used to characterize the fibre surface. The results of XPS revealed a threshold value for CS concentration to improve the interface of the composites. L-DOPA dissolved in Tris buffer solution could also improve the interfacial properties of PGF/PCL composites. Meanwhile, a sufficiently high L-DOPA concentration within 0.02-0.04 g/mL could significantly enhance the tensile strength of the fibres. TG, SEM, FTIR and XPS were used to characterize the fibre surface. The results of these analyses suggested the zwitterionic form for L-DOPA powders and the melanin-like polymer layer on the fibres after coating.

Significant strides have been made in computer vision over the past few years due to the recent development in deep learning, especially deep convolutional neural networks (CNNs). Based on the advances in GPU computing, innovative model architectures and large-scale dataset, CNNs have become the workhorse behind the state of the art performance for most computer vision tasks. For instance, the most advanced deep CNNs are able to achieve and even surpass human-level performance in image classification tasks. Deep CNNs have demonstrated the ability to learn very powerful image features or representations in a supervised manner. However, in spite of the impressive performance, it is still very difficult to interpret and understand the learned deep features when compared to traditional human-crafted ones. It is not very clear what has been learned in the deep features and how to apply them to other tasks like traditional image processing problems. In this thesis, we focus on exploring deep features extracted from pretrained deep convolutional neural networks, based on which we develop new techniques to tackle different traditional image processing problems. First we consider the task to quickly filter out irrelevant information in an image. In particular, we develop a method for exploiting object specific channel (OSC) from pretrained deep CNNs in which neurons are activated by the presence of specific objects in the input image. Building on the basic OSC features and use face detection as a specific example, we introduce a multi-scale approach to constructing robust face heatmaps for rapidly filtering out non-face regions thus significantly improving search efficiency for potential face candidates. Finally we develop a simple and compact face detectors in unconstrained settings with state of the art performance. Second we turn to the task to produce visually pleasing images. We investigate two generative models, variational autoencoder (VAE) and generative adversarial network (GAN), and propose to construct objective functions to train generative models by incorporating pretrained deep CNNs. As a result, high quality face images can be generated with realistic facial parts like clear nose, mouth as well as the tiny texture of hair. Moreover, the learned latent vectors demonstrate the capability of capturing conceptual and semantic information of facial images, which can be used to achieve state of the art performance in facial attribute prediction. Third we consider image information augmentation and reduction tasks. We propose a deep feature consistent principle to measure the similarity between two images in feature space. Based on this principle, we investigate several traditional image processing problems for both image information augmentation (companding and inverse halftoning) and reduction (downscaling, decolorization and HDR tone mapping). The experiments demonstrate the effectiveness of deep learning based solutions to solve these traditional low-level image processing problems. These approaches enjoy many advantages of neural network models such as easy to use and deploy, end-to-end training as a single learning problem without hand-crafted features. Last we investigate objective methods for measuring perceptual image quality and propose a new deep feature based image quality assessment (DFB-IQA) index by measuring the inconsistency between the distorted image and the reference image in feature space. The proposed DFB-IQA index performs very well and behave consistently with subjective mean opinion scores when applied to distorted images created from a variety of different types of distortions. Our works contribute to a growing literature that demonstrates the power of deep learning in solving traditional signal processing problems and advance the state of the art on different tasks.

Weld repair is a specific application of fusion welding processes adopted to correct defects which arise during the manufacturing process of aeroengine components and, also, to repair damage in order to extend the operative life, if safe and correctly performed. The operation is carried out by removing the anomaly and re-filling the slot with a fusion welding process. The thermal cycles induced by fusion welding processes produce undesirable residual stresses, which significantly affect the fatigue life of the component and, in some cases, they may give a larger contribution to the total stress field than the stresses caused by the service loads themselves. Based on the sequentially coupled thermo-mechanical analysis, two numerical methodologies have been implemented to simulate the residual stress field induced by weld repair operations. Differently from models presented in the literature, the modelling approaches allow taking into account the pre-existing stress in the component subjected to the weld repair. The experimental work to validate the numerical models involved a full thermal characterization of the welding apparatus used to produce the welds. Residual stress predictions were validated by means of neutron diffraction measurements using the most advanced diffractometer, ENGIN-X, available at ISIS neutron source at the Rutherford Appleton Laboratory. The model can be used a priori, in the attempt of mitigating the effects of the repair operations on the final residual stress field. It can also be used a posteriori, with the aim of determining the stress state in the component, producing the input for the analysis of the fatigue life. In order to prove the applicability of the numerical methodology in the case of aeroengine components, the model was used to study a weld repair in a demonstrative case study. The effects of number and direction of passes were investigated in the weld repair of a laser beam weld in a stainless steel pipe. It was concluded that the interaction between the pre-existing stress in a component and the stress induced by weld repair operations cannot be generalized and established a priori. Numerical models that simulate the process by neglecting the history of the components, ignore the contribution of the pre-existing stress to the final residual stress distribution. Predictions from such models approximate the global stress distribution in the component. The entire characterization of the residual stress field in a component that undergoes a weld repair operation can only be achieved by using models which take into account the pre-existing stress.

Balancing rotating systems is a challenging task, which requires (dis)assembly of the system to enable mass adjustments; thus the development of a method to balance rotatives in-situ (i.e. without disassembly) using pulsed laser ablation (PLA) is a key technology enabler. PLA for in-situ balancing offers inherent advantages of an adjustable frequency (to match that of the rotating part) and variable pulse energy (to control the mass removal). This thesis presents a novel methodology for balancing components in-situ using PLA in a controlled and automated manner. The method utilises a sensor to measure the acceleration of the rigid rotor-bearing system. After signal conditioning using an adaptive peak filter (i.e. an inverted notch filter), a developed peak detection algorithm determines the maxima of the signal to find the angular imbalance position. If corrective action is necessary, PLA occurs. The method accounts for the time delays in the laser system and electronic circuit. Validation on a rotating part showed a PLA targeting accuracy of < 50μm and a precision of < 30μm; the feasibility of the method was confirmed using a simulation and by balancing a rotor with an arbitrary added imbalance. A concept, which was devised to optimise the PLA strategy for removing imbalances, bases on a novel combination of an analytical and machine learning approach. It determines the optimum process parameters of an ablated feature with a specified shape and volume. Additionally, an error budget for the method has been developed. The concept has been validated and shown to be accurate to < 4mg. The error budget could account for variations. It has been shown long features in the circumferential direction of the part increase the material removal rate with only minor increases in the error magnitude. To conclude, a concept for the integration of the two developed models is presented.

Phosphate-based glass fibre reinforced poly-lactic acid (PGF/PLA) composites containing biocompatible and fully biodegradable constituents can provide huge clinical benefits in comparison with bio-inert metallic materials, such that they have a great potential for use of bone fracture fixation devices. The previous lab-scale phosphate glass fibre production can only provide single filament, which limited the use of these fibres as they could only be produced as non-woven preforms such as unidirectional or random fibre mats. The work conducted in this project explored the manufacture of continuous multifilament phosphate glass fibre strands and textiles products with a pilot plant in co-operation with Sinoma Ltd. in Nanjing, China. The aim of the work conducted in this thesis was to scale-up the manufacture of continuous multifilament PGF strands and textile products. Additionally, this work explored the design and manufacturing of novel PGF/PLA commingled yarns, and also highlighted the successful feasibility and proof-of-concept for producing textiles and textile composites for use as bone plates for load bearing applications. Novel zinc-containing glass formulations based on the system P2O5-CaO-MgO-Na2O-ZnO were investigated in terms of thermal, structural, degradation, viscosity and fibre tensile properties, and also the feasibilities for industrial-scale multifilament fibre production. The replacement of monovalent Na+ with the higher field strength Zn2+ increased the glass transition, crystallization, melting, liquidus temperatures, softening temperature, density and viscosity. The initial addition of ZnO in the glass system increased the mechanical properties of fibres and the chemical durability of the glasses investigated. However, once the addition of a particular amount of ZnO was exceeded, the mechanical properties of fibres and chemical durability of the glasses were seen to decrease. These glass compositions may not be suitable for industrial-scale fibre production. Therefore, the glass code P48B12Na1 was used in following research for the production of multifilament yarns, textiles and commingled textiles. The industrial-scale multifilament production of phosphate glass fibre (PGF) strands was achieved successfully. The textile yarns were produced by combining fibre strands using the ring-spinning method. The PGF textiles were prepared using a home-made inkle loom. PGF textile composites were prepared using film stacking method. The crimp of yarns was found to have a significantly negative effect on the flexural properties of the textile composites in comparison with unidirectional (UD) composites. The number-average molecular weight of PLA was also found to reduce after the production of PLA films and PLA plates, in comparison with the original PLA pellets used. The pilot scale production of PGF/PLA commingled yarns and textiles with designed fibre volume fraction was achieved successfully. Commingled textiles showed a great benefit in improving the mechanical properties of textile composites as compared to the ones produced using film stacking method. The effects of compression moulding parameters (factors) on flexural strength and consolidation quality of PGF/PLA commingled textile based composites were investigated using statistical experimental range analysis, including processing temperature, preheating time, compression time and pressure. Processing temperature provided the greatest effect on the flexural strength of all the composites, followed by preheating time. The statistical experimental design methodology was confirmed to be a powerful tool to study and optimise the processing parameters for production of PGF/PLA commingled textile based composites. Unidirectional (UD) composites, textile composites and 0°/90° composites based on PGF/PLA commingled yarns were prepared by compression moulding. Effects of edge sealing, fibre content and orientation on the degradation performance of the composites were investigated during immersion in phosphate buffered saline (PBS) solution at 37 °C for 28 days. With an increase in fibre content, the initial flexural strength was improved, and all were comparable to human cortical bone. The sharp decrease in flexural properties was seen for all composites during the initial 3 days of immersion was attributed to the loss of fibre/matrix interfacial adhesion. The edge sealed composites revealed less decrease in both pH and weight loss in comparison with unsealed composites due to the fast diffusion of water into the PLA matrix. No significant improvement on the retention of mechanical properties for edge sealed composites was found. Additionally, composites with higher fibre volume fraction revealed faster degradation of PGFs and lower retention of mechanical properties over immersion period of 28 days. It was suggested that 0°/90° composites was more superior than textile composites due to the 0°/90° composites with the desired biaxial mechanical properties were much easier to be designed and manufactured of due to their higher flexibility in layout of commingled UD fabric preforms.

There are two main parts in this thesis: (1) to investigate a multi-layer structure that combines Kretschmann configuration and Otto configuration which we call the Kretschmann-Otto configuration. To examine this structure, we develop an objective based nano-controlled system; and (2) develop a transmission line model method to analyse the responsivity of surface plasmon (SP) sensors. In the first part, the Kretschmann-Otto structure is proposed and the performance of surface plasmon and Fabry-Perot modes formed in this structure is investigated. The motivation for this study is twofold, firstly, to look for modes that may be excited at lower incident angles compared to the usual Kretschmann configuration with similar or superior refractive index responsivity and, secondly, to develop a simple and applicable method to study these structures over a wide range of separations without recourse to the construction of ad hoc structures. With this nano-controlled system, we show that the contribution of gap separation to the minimum reflectivity and the width of reflected curve for the Otto configuration at visible wavelengths at a range of separations not reported hitherto. Moreover, the BFP distributions of Kretschmann-Otto configuration are investigated at various gap separations and the layer responsivity is demonstrated experimentally by fabricating a gold coated coverslip with a protein grating on top of it. We show that the zero order Fabry-Perot mode at normal incident angle has superior refractive index responsivity, by more than an order of magnitude, and layer responsivity by around 5 times compared to the Kretschmann configuration. In the second part, we propose a transmission line model based method to give insight into the sensing process and explain the main determinants to layer responsivity. By applying the appropriate resonant condition to the systems, we derive a circuit model which predicts the responsivity of different configurations. Specifically, the model provides a compact explanation for the change in responsivity due to a high index layer placed between the metal and the analyte. From this method, a parameter arises naturally from the model and the response of a generic sensor to binding of an analyte can be predicted. Intuitively, it may be expected that the energy stored in the circuit is related to the responsivity of the sensor, here we show that, under normal operating conditions, while it is predictive of the sharpness of the resonance the responsivity depends only on the variation of circuit reactance with wave number.

This work is aimed at developing a new signal processing technology based on time-frequency analysis to extract the dynamic characteristics of rotating machinery and bridge structure. Therefore, the work can be divided into two parts, the condition monitoring of the gearbox and the structural health monitoring of the bridges. The first part of the work aims: (i) to propose a Joint Time-Frequency Analysis (JTFA) method for gear fault diagnosis by using the combined autoregressive (AR) model-based filtering and Reassigned Smoothed Pseudo Wigner-Ville Distribution (RSPWVD) methods; (ii) to investigate the use of both vibration and acoustic measurements for fault diagnosis of a gear system by using the proposed fault diagnosis method. To the best of the author’s knowledge, such RSPWVD method has not been utilized for gearbox applications due to problems with the complexity of signals generated by the gearbox. For this purpose, experiments on a single-stage spur gearbox were carried out on a gearbox test-rig using a single-defect with two different severity levels and double-defect gear tooth faults, utilizing vibration and non-contact acoustic sensing. It was experimentally demonstrated that the proposed fault diagnosis method performed better compared to the Continuous Wavelet Transform, the Smoothed Pseudo Wigner-Ville Distribution and even with the newly introduced parameterized time-frequency method, the General Linear Chirplet Transform. The proposed method can provide a more accurate indication of faults in a gearbox, even for the case of multiple gear defects using both acoustic and vibration measurements. The results demonstrate the potential of using non-contact acoustic measurement using the proposed signal processing method as an alternative sensing method for gear condition monitoring applications. Conversely, another focus of this research is on the structural health monitoring of a bridge with GNSS (Global Navigation Satellite System) measurements. However, the implementation of the time-frequency analysis methods utilized in the condition monitoring of a gearbox is not possible to apply directly due to the large scale of the data set from GNSS measurements. The restriction from the duration-bandwidth principle does not permit accurate tracking of the variation of the natural frequencies in every single epoch measurement from the lengthy data in GNSS measurements. Therefore, a simple yet efficient algorithm using the Fast Fourier Transform (FFT) method is proposed to capture the shifting phenomena of the natural frequencies of the bridge during three-day measurements. It is shown that a GNSS sensor can provide useful information regarding the shifting of natural frequencies that are affected by the variation of the ambient temperature during the field test.

Resin Transfer Moulding (RTM) is an advanced manufacturing process for composite components, and is widely applied in industries such as aerospace. A full resin impregnation during RTM is essential to obtain a high-quality component with reduced defects content. To reduce the level of experimental work as well as the cost for optimizing processing parameters, numerical simulation tools are widely used nowadays. The work presented in this thesis is based on one work package from a company sponsored project, focused mainly on research in permeability characterisation and defects generation during the RTM process. From another point of view, this research can be considered as a multi-scale study of RTM process modelling, which consists of a micro-scale study of the fluid-porous interface to determine fibre tow boundary conditions for permeability modelling, a meso-scale study of textiles permeability based on unit cell modelling, and a macro-scale study of RTM process modelling to characterize void formation. Furthermore, experimental work is also included for validation against simulation results. As a key input for RTM process modelling, permeability prediction based on meso-scale unit cell is studied in this research. Micro-CT technology is employed to characterize the internal geometry of 3D woven fabric, which is the input for the generation of unit cell. Besides, geometrical variability is studied for its influence on the permeability prediction results, Due to the dual flow characteristic within 3D woven fabric, proper fibre yarn boundary conditions are developed for permeability modelling with unit cell. Fibre yarn with random fibre distribution models are generated. Flow transverse with the fibre yarn is studied and it’s found that no-slip boundary condition can be applied in the present research. Considering the geometric variability inherent with fabrics, permeability variability study is also conducted in this research. The geometric variability characterization methods are introduced and based on that, the influence of geometric variability on the permeability prediction is studied. Regarding with macro-scale RTM process modelling, parametric study of injection conditions are conducted to investigate its effect on voids formation. Furthermore, permeability experiment of 3D woven fabrics with different fibre volume fraction, as well as parametric study of RTM experiments for voids formation are also included in this research. Comparison of permeability measurement value and modelling results are made, with which the unit cell model is optimized.