7103 research outputs found
Sort by
Bioinspired materials for enzyme immobilisation and transport of biomolecules
Get PDFThis thesis was previously restricted to Strathclyde users only from 1st November 2017 until 15 November 2022.Protein and enzyme immobilisation on synthetic material surfaces enables a range of applications from biosensing to industrial biocatalysis. There are several immobilisation techniques, but common methods need multiple preparation steps or are material-dependent, which reduce the effectiveness and success of biosensing/industrial applications.;In this thesis, the possibility of using plant-based polyphenol coatings to immobilise a range of proteins and enzymes (avidin, immunoglobulin G, acid phosphatase, chymotrypsin, lactate dehydrogenase, horseradish peroxidase) on polymeric or oxide materials (cellulose, polyester, silica, alumina and stainless steel) was shown for the first time. Polyphenols are in abundance in nature and less costly than dopamine (common immobilisation agent).;Polyphenols were more effective than physisorption and polydopamine coatings for certain combinations of materials and proteins. Several parameters that can influence the immobilisation procedure were evaluated showing that there is a dependence on the coating and immobilisation pH as well as the type of coating and material.;Polyphenol coatings were also used to functionalise nanoporous anodic aluminium oxide (AAO) membranes in order to measure molecular transport through nanopores. Inspired by the biological nanopores that enable the highly specific and efficient separation of a range of biomolecules, we used AAO membranes with a pore size matching the biological nuclear pore complex for controlling the diffusion of molecules through the pores.;AAO membranes also match the requirements for optical waveguide spectroscopy (OWS) that was used to characterise and differentiate processes that occurred inside and above the nanoporous membranes. In a second approach, a nanoporous membrane was placed between two gaskets to be suspended on the flow cell. This work brings a new concept of how the molecular diffusion can be characterised, which is important for controlling the transport of biomolecules.;It was possible to monitor in situ biocatalytic reactions as well as nanopore transport control by using a responsive polymer that was able to allow and restrict the transport of molecules.Protein and enzyme immobilisation on synthetic material surfaces enables a range of applications from biosensing to industrial biocatalysis. There are several immobilisation techniques, but common methods need multiple preparation steps or are material-dependent, which reduce the effectiveness and success of biosensing/industrial applications.;In this thesis, the possibility of using plant-based polyphenol coatings to immobilise a range of proteins and enzymes (avidin, immunoglobulin G, acid phosphatase, chymotrypsin, lactate dehydrogenase, horseradish peroxidase) on polymeric or oxide materials (cellulose, polyester, silica, alumina and stainless steel) was shown for the first time. Polyphenols are in abundance in nature and less costly than dopamine (common immobilisation agent).;Polyphenols were more effective than physisorption and polydopamine coatings for certain combinations of materials and proteins. Several parameters that can influence the immobilisation procedure were evaluated showing that there is a dependence on the coating and immobilisation pH as well as the type of coating and material.;Polyphenol coatings were also used to functionalise nanoporous anodic aluminium oxide (AAO) membranes in order to measure molecular transport through nanopores. Inspired by the biological nanopores that enable the highly specific and efficient separation of a range of biomolecules, we used AAO membranes with a pore size matching the biological nuclear pore complex for controlling the diffusion of molecules through the pores.;AAO membranes also match the requirements for optical waveguide spectroscopy (OWS) that was used to characterise and differentiate processes that occurred inside and above the nanoporous membranes. In a second approach, a nanoporous membrane was placed between two gaskets to be suspended on the flow cell. This work brings a new concept of how the molecular diffusion can be characterised, which is important for controlling the transport of biomolecules.;It was possible to monitor in situ biocatalytic reactions as well as nanopore transport control by using a responsive polymer that was able to allow and restrict the transport of molecules
Financial sustainability of public Higher Education Institutions in Uzbekistan
Get PDFSince the early years of independence in 1991, a central topic of higher education in Uzbekistan has been how to fill the gap left by the reduced government funding at public Higher Education Institutions (HEIs). The majority of the Uzbek universities, as in many other countries, have responded to the decline in public allocations through charging significantly increased tuition fees. Therefore, the revenue structure of public HEIs has changed from full government funding to mostly tuition funding over the last decade. The main aim of this study is to examine the impact of this shift in the institutional revenue structure on behaviour and efficient resource utilisation of public HEIs in Uzbekistan. For these purposes, this study begins with analysing the contemporary outlook of Uzbek education sector, particularly administrative and financing structures and reforms of the higher education. A comprehensive comparison between the higher education system of Uzbekistan and the rest of the Central Asian Republics (CARs) is also provided in this thesis.To the best of my knowledge, this thesis is the first study which scrutinises the higher education systems of the entire CARs. Utilising resource dependence theory (RDT), this study empirically investigates whether or not increased institutional reliance on tuition fees as a main source of revenue has augmented the share of institutional expenditures dedicated to educational activities at public HEIs in Uzbekistan over the period 2000-2013. Drawing on a 14-year panel of university-level data and employing an instrumental variable approach that acknowledges the potential endogeneity of institutional tuition revenue, the author finds that the institutional expenditures for educational expenses are considerably increased as institutions became more dependent on tuition revenue for their financially sustainable operation. This finding is consistent with the predictions of RDT. Robustness of the empirical findings is also tested utilising several diagnostic models.In this study, a stochastic cost frontier analysis is used in order to examine whether the institutional fiscal resources obtained mostly from tuition have been utilised efficiently or inefficiently at public HEIs in Uzbekistan during the period of 2000 to 2013. The Battese and Coelli (1995) method is applied to measure the influences of institution, staff and student characteristics on cost efficiency of the universities. According to mean efficiency scores that the Uzbek universities are not remarkably cost efficient in producing education and research outputs, although the significant improvements in the efficiency followed throughout the sample period. Interestingly, findings also reveal that public HEIs with a greater share of public funding are less cost efficient relative to those institutions with a smaller share of public funding.Since the early years of independence in 1991, a central topic of higher education in Uzbekistan has been how to fill the gap left by the reduced government funding at public Higher Education Institutions (HEIs). The majority of the Uzbek universities, as in many other countries, have responded to the decline in public allocations through charging significantly increased tuition fees. Therefore, the revenue structure of public HEIs has changed from full government funding to mostly tuition funding over the last decade. The main aim of this study is to examine the impact of this shift in the institutional revenue structure on behaviour and efficient resource utilisation of public HEIs in Uzbekistan. For these purposes, this study begins with analysing the contemporary outlook of Uzbek education sector, particularly administrative and financing structures and reforms of the higher education. A comprehensive comparison between the higher education system of Uzbekistan and the rest of the Central Asian Republics (CARs) is also provided in this thesis.To the best of my knowledge, this thesis is the first study which scrutinises the higher education systems of the entire CARs. Utilising resource dependence theory (RDT), this study empirically investigates whether or not increased institutional reliance on tuition fees as a main source of revenue has augmented the share of institutional expenditures dedicated to educational activities at public HEIs in Uzbekistan over the period 2000-2013. Drawing on a 14-year panel of university-level data and employing an instrumental variable approach that acknowledges the potential endogeneity of institutional tuition revenue, the author finds that the institutional expenditures for educational expenses are considerably increased as institutions became more dependent on tuition revenue for their financially sustainable operation. This finding is consistent with the predictions of RDT. Robustness of the empirical findings is also tested utilising several diagnostic models.In this study, a stochastic cost frontier analysis is used in order to examine whether the institutional fiscal resources obtained mostly from tuition have been utilised efficiently or inefficiently at public HEIs in Uzbekistan during the period of 2000 to 2013. The Battese and Coelli (1995) method is applied to measure the influences of institution, staff and student characteristics on cost efficiency of the universities. According to mean efficiency scores that the Uzbek universities are not remarkably cost efficient in producing education and research outputs, although the significant improvements in the efficiency followed throughout the sample period. Interestingly, findings also reveal that public HEIs with a greater share of public funding are less cost efficient relative to those institutions with a smaller share of public funding
Three essays on government intervention in the oil and gas industry
Get PDFFrom resource nationalism to 'softcore' government intervention through regulatory changes, this thesis attempts to comprehend in depth and formally analyse both the economic as well as the non-financial factors influencing the development of hydrocarbon resources and the investment decision-making in the oil and gas industry. The empirical work of Chapter 1 is inspired by the competition between IOCs ad NOCs and the phenomenon of resource nationalism. It provides empirical evidence on how the socio-economic conditions can affect the way a country will choose to develop its natural resources presenting the social determinants which contribute to the rise of nationalisation in the oil and gas industry. Chapter 2 focuses on government intervention on the upstream pipeline transportation networks and the issue of third party access under conditions of natural monopoly. It applies basic regulatory economic principles on oil and gas transportation networks and explores various regulatory tools and their application in different basins. Special focus is given in the government intervention and the market conditions under which State interference in the market is justified and successful. Finally, Chapter 3, taking into consideration the theory developed in Chapter 2, provides policy recommendations which aim to tackle market inefficiencies in the UKCS for the utilisation of the remaining reserves. The chapter also discusses the role and limitations of government ownership in the UKCS. The three chapters analyse different, but interlinked, issues surrounding the relationship between the government and the oil and gas industry- from hardcore nationalisation of the natural resources to the unique Norwegian model of State ownership.From resource nationalism to 'softcore' government intervention through regulatory changes, this thesis attempts to comprehend in depth and formally analyse both the economic as well as the non-financial factors influencing the development of hydrocarbon resources and the investment decision-making in the oil and gas industry. The empirical work of Chapter 1 is inspired by the competition between IOCs ad NOCs and the phenomenon of resource nationalism. It provides empirical evidence on how the socio-economic conditions can affect the way a country will choose to develop its natural resources presenting the social determinants which contribute to the rise of nationalisation in the oil and gas industry. Chapter 2 focuses on government intervention on the upstream pipeline transportation networks and the issue of third party access under conditions of natural monopoly. It applies basic regulatory economic principles on oil and gas transportation networks and explores various regulatory tools and their application in different basins. Special focus is given in the government intervention and the market conditions under which State interference in the market is justified and successful. Finally, Chapter 3, taking into consideration the theory developed in Chapter 2, provides policy recommendations which aim to tackle market inefficiencies in the UKCS for the utilisation of the remaining reserves. The chapter also discusses the role and limitations of government ownership in the UKCS. The three chapters analyse different, but interlinked, issues surrounding the relationship between the government and the oil and gas industry- from hardcore nationalisation of the natural resources to the unique Norwegian model of State ownership
Formulating test oracles via anomaly detection techniques
Get PDFDevelopments in the automation of test data generation have greatly improved efficiency of the software testing process but the so-called "oracle problem" (deciding the pass or fail outcome of a test execution) is still primarily an expensive and error-prone manual activity. This thesis presents an approach to build an automated test oracle using anomaly detection techniques (based on semi-supervised and unsupervised learning approaches) on dynamic execution data (test input/output pairs and execution traces).;Firstly, anomaly detection techniques based on semi-supervised learning approach were investigated to automatically classify passing and failing executions. A small proportion of the test data is labelled as passing or failing and used in conjunction with the unlabelled data to build a classifier which labels the remaining outputs (classify them as passing or failing tests).;A range of learning algorithms are investigated using several faulty versions of three systems along with varying types of data (inputs/outputs alone, or in combination with execution traces) and different labelling strategies (both failing and passing tests, and passing tests alone). The results show that in many cases labelling just a small proportion of the test cases - as low as 10% - is sufficient to build a classifier that is able to correctly categorise the large majority of the remaining test cases.;This has important practical potential: when checking the test results from a system a developer need only examine a small proportion of these and use this information to train a learning algorithm to automatically classify the remainder.;Secondly, anomaly detection techniques based on unsupervised learning (mainly clustering algorithms) were investigated to automatically detect passing and failing executions. The key hypothesis is that failures will group into small clusters whereas passing executions will group into larger ones. In this investigation, the same dynamic execution data and systems used in previous study were used to evaluate the proposed approach.;The results show that this hypothesis to be valid, and illustrates that the approach has the potential to substantially reduce the numbers of outputs that would need to be manually examined following a test run.;Finally, a comparison study was performed between existing techniques from the specifications mining domain (the data invariant detector Daikon [30]) and anomaly detection techniques (based on semi-supervised and unsupervised learning approaches). In most cases semi-supervised learning techniques (mainly Self-training approach - Naïve Bayes with EM clustering algorithm - and Co-training approach - Naïve Bayes) perform far better under both scenarios (two different labelling strategies) as an automated test classifier than Daikon especially when input/output pairs are used together with execution traces. Furthermore, unsupervised learning techniques performed on a par when compared with Daikon in several cases.Developments in the automation of test data generation have greatly improved efficiency of the software testing process but the so-called "oracle problem" (deciding the pass or fail outcome of a test execution) is still primarily an expensive and error-prone manual activity. This thesis presents an approach to build an automated test oracle using anomaly detection techniques (based on semi-supervised and unsupervised learning approaches) on dynamic execution data (test input/output pairs and execution traces).;Firstly, anomaly detection techniques based on semi-supervised learning approach were investigated to automatically classify passing and failing executions. A small proportion of the test data is labelled as passing or failing and used in conjunction with the unlabelled data to build a classifier which labels the remaining outputs (classify them as passing or failing tests).;A range of learning algorithms are investigated using several faulty versions of three systems along with varying types of data (inputs/outputs alone, or in combination with execution traces) and different labelling strategies (both failing and passing tests, and passing tests alone). The results show that in many cases labelling just a small proportion of the test cases - as low as 10% - is sufficient to build a classifier that is able to correctly categorise the large majority of the remaining test cases.;This has important practical potential: when checking the test results from a system a developer need only examine a small proportion of these and use this information to train a learning algorithm to automatically classify the remainder.;Secondly, anomaly detection techniques based on unsupervised learning (mainly clustering algorithms) were investigated to automatically detect passing and failing executions. The key hypothesis is that failures will group into small clusters whereas passing executions will group into larger ones. In this investigation, the same dynamic execution data and systems used in previous study were used to evaluate the proposed approach.;The results show that this hypothesis to be valid, and illustrates that the approach has the potential to substantially reduce the numbers of outputs that would need to be manually examined following a test run.;Finally, a comparison study was performed between existing techniques from the specifications mining domain (the data invariant detector Daikon [30]) and anomaly detection techniques (based on semi-supervised and unsupervised learning approaches). In most cases semi-supervised learning techniques (mainly Self-training approach - Naïve Bayes with EM clustering algorithm - and Co-training approach - Naïve Bayes) perform far better under both scenarios (two different labelling strategies) as an automated test classifier than Daikon especially when input/output pairs are used together with execution traces. Furthermore, unsupervised learning techniques performed on a par when compared with Daikon in several cases
Hybrid GaN-based optoelectronics for visible light communications
Get PDFDevelopment of light-emitting diode technology is driven mainly by the need for efficient solid-state lighting, but it is also creating opportunities for new applications such as visible light communications (VLC). Here, the solid-state visible light sources are used to transmit data with the added requirement of a short excited-state lifetime so that sources can be modulated at high speed. This research focuses on hybrid optical sources for visible light communications with an emphasis on novel formats of colour-converters for multi-wavelength photoluminescence as well as white-light generation. Such converters include red and green colloidal quantum dots, the organic semiconductor BBEHPV andII-VI / III-V epitaxial structures.;Solution-processable and environmentally stable polymeric films based on red and green colloidal quantum dots are demonstrated. Modulation bandwidth up to 24 MHz, photoluminescence quantum yields up to 61% and peak emission tunability across the visible spectrum makes these materials interesting as colour-converters for VLC. Free-space data transmission was demonstrated in this case with data rates up to 400 Mbit/s and 500 Mbit/s using 2-PAM modulation scheme for green and red quantum dots, respectively.;Hybrid sources consisting of 450nm InGaN LEDs with capillary-bonded micron-thick ZnCdSe/ZnCdMgSe multi-quantum-well colour-converting membranes with peak emission at 540 nm are reported. After processing, the membrane was capillary bonded onto the sapphire side of the μLED resulting in a maximum converted average power of 37 μW. The -3dB optical modulation bandwidth of the bare LED, hybrid device and II-VI were 79 MHz, 51 MHz and 145 MHz, respectively.;Visible light communication using both InGaN LEDs and a InGaN laser diode, down-converted by a red-emitting AlInGaP multi-quantum-well nanomembrane are also reported. Similarly to the previous devices, the AlInGaP nanomembrane was bonded onto the μLED array. For the down-converted laser diode approach, the nanomembrane can be sandwiched between a sapphire lens and optionally onto a distributed feedback reflector. The down-converter structure is remotely excited by the laser diode. Data transmission up to 870 Mb/s using M-PAM andOFDM modulation schemes is demonstrated for the μLED integrated nanomembrane. ODFM transmission at 1.2Gb/s is achieved for the laser diode pumped sample.Development of light-emitting diode technology is driven mainly by the need for efficient solid-state lighting, but it is also creating opportunities for new applications such as visible light communications (VLC). Here, the solid-state visible light sources are used to transmit data with the added requirement of a short excited-state lifetime so that sources can be modulated at high speed. This research focuses on hybrid optical sources for visible light communications with an emphasis on novel formats of colour-converters for multi-wavelength photoluminescence as well as white-light generation. Such converters include red and green colloidal quantum dots, the organic semiconductor BBEHPV andII-VI / III-V epitaxial structures.;Solution-processable and environmentally stable polymeric films based on red and green colloidal quantum dots are demonstrated. Modulation bandwidth up to 24 MHz, photoluminescence quantum yields up to 61% and peak emission tunability across the visible spectrum makes these materials interesting as colour-converters for VLC. Free-space data transmission was demonstrated in this case with data rates up to 400 Mbit/s and 500 Mbit/s using 2-PAM modulation scheme for green and red quantum dots, respectively.;Hybrid sources consisting of 450nm InGaN LEDs with capillary-bonded micron-thick ZnCdSe/ZnCdMgSe multi-quantum-well colour-converting membranes with peak emission at 540 nm are reported. After processing, the membrane was capillary bonded onto the sapphire side of the μLED resulting in a maximum converted average power of 37 μW. The -3dB optical modulation bandwidth of the bare LED, hybrid device and II-VI were 79 MHz, 51 MHz and 145 MHz, respectively.;Visible light communication using both InGaN LEDs and a InGaN laser diode, down-converted by a red-emitting AlInGaP multi-quantum-well nanomembrane are also reported. Similarly to the previous devices, the AlInGaP nanomembrane was bonded onto the μLED array. For the down-converted laser diode approach, the nanomembrane can be sandwiched between a sapphire lens and optionally onto a distributed feedback reflector. The down-converter structure is remotely excited by the laser diode. Data transmission up to 870 Mb/s using M-PAM andOFDM modulation schemes is demonstrated for the μLED integrated nanomembrane. ODFM transmission at 1.2Gb/s is achieved for the laser diode pumped sample
Secure decentralised storage networks
Get PDFIn recent years, cloud-based computing and storage have become increasingly popular,as they remove the need for users and developers to buy or rent expensive dedicated hardware on an ongoing basis. This has led to the increasing centralisation of both services and storage, where users are reliant upon a small number of cloud-based providers to hold their data, and provide them with services they use. Recent events have shown that security breaches of centralised data stores can lead to significant quantities of personal data being revealed. This centralisation can also result in inconvenience in the event of the failure of the service provider, resulting in potential data loss or a loss of utility of the service.In contrast, a decentralised service and storage architecture removes the single point of failure from a network, and allows users to remove their dependency on a single company or service provider. In addition, by preventing storage providers from having access to user data, as is inherently needed in a decentralised network to preserve confidentiality,it is possible for users to protect their data from theft or unauthorised access,giving rise to data security and privacy benefits.This thesis explores the the challenges encountered in implementing a secure decentralised network, based around storage, and presents solutions to some of these problems.A security analysis of the MaidSafe network is firstly given, setting the context of the work, and investigating the state-of-the-art. Potential uses for decentralised services are considered, including for use on mobile devices. The importance of client device security is also considered, and a number of vulnerabilities affecting the security of client-based software are identified and explored. A practical design of decentralised architecture for preserving user privacy when discovering users is also contributed, to illustrate how decentralised service design can be used to enhance privacy of existing systems, and solve otherwise unsolved problems. A review and analysis of the privacy policies of popular web-based services then shows the extent to which user privacy is at risk from centralised web services. Finally, the concepts of identity and authentication within decentralised networks are considered, with a novel smartcard-based approach to securing user credentials within a decentralised network demonstrated.In recent years, cloud-based computing and storage have become increasingly popular,as they remove the need for users and developers to buy or rent expensive dedicated hardware on an ongoing basis. This has led to the increasing centralisation of both services and storage, where users are reliant upon a small number of cloud-based providers to hold their data, and provide them with services they use. Recent events have shown that security breaches of centralised data stores can lead to significant quantities of personal data being revealed. This centralisation can also result in inconvenience in the event of the failure of the service provider, resulting in potential data loss or a loss of utility of the service.In contrast, a decentralised service and storage architecture removes the single point of failure from a network, and allows users to remove their dependency on a single company or service provider. In addition, by preventing storage providers from having access to user data, as is inherently needed in a decentralised network to preserve confidentiality,it is possible for users to protect their data from theft or unauthorised access,giving rise to data security and privacy benefits.This thesis explores the the challenges encountered in implementing a secure decentralised network, based around storage, and presents solutions to some of these problems.A security analysis of the MaidSafe network is firstly given, setting the context of the work, and investigating the state-of-the-art. Potential uses for decentralised services are considered, including for use on mobile devices. The importance of client device security is also considered, and a number of vulnerabilities affecting the security of client-based software are identified and explored. A practical design of decentralised architecture for preserving user privacy when discovering users is also contributed, to illustrate how decentralised service design can be used to enhance privacy of existing systems, and solve otherwise unsolved problems. A review and analysis of the privacy policies of popular web-based services then shows the extent to which user privacy is at risk from centralised web services. Finally, the concepts of identity and authentication within decentralised networks are considered, with a novel smartcard-based approach to securing user credentials within a decentralised network demonstrated
Behavioural profiling of the ZDHHC9 knockout mouse
No full textS-palmitoylation is a reversible post translational lipidation event driven by the enzymatic activity of palmitoyl transferases containing a DHHC (aspartate-histidine-histidine-cysteine) motif. There are 24 distinct zDHHC enzymes present in mammals and each has its own key cellular interactions. Mutations in the ZDHHC9 gene in humans have been linked to intellectual disability (ID) in several families. The aim of this study was to behaviourally phenotype, in a range of established behavioural tests, a novel mouse model in which expression of the zDHHC9 gene has been knocked out; phenotyping of these mice is important to determine the role of zDHHC9 in the brain and whether this mouse line can serve as a new model for ID. Results revealed that the ZDHHC9 knockout mice are hyperactive, which is frequently also seen in other mouse models of ID. Furthermore, the mice also displayed a learning and memory deficit in the Morris Water Maze, and also a possible muscular deficit. Overall, the results suggest that the novel zDHHC9 knockout mouse line may be a valuable resource to investigate zDHHC9 function in the brain and its links to ID.S-palmitoylation is a reversible post translational lipidation event driven by the enzymatic activity of palmitoyl transferases containing a DHHC (aspartate-histidine-histidine-cysteine) motif. There are 24 distinct zDHHC enzymes present in mammals and each has its own key cellular interactions. Mutations in the ZDHHC9 gene in humans have been linked to intellectual disability (ID) in several families. The aim of this study was to behaviourally phenotype, in a range of established behavioural tests, a novel mouse model in which expression of the zDHHC9 gene has been knocked out; phenotyping of these mice is important to determine the role of zDHHC9 in the brain and whether this mouse line can serve as a new model for ID. Results revealed that the ZDHHC9 knockout mice are hyperactive, which is frequently also seen in other mouse models of ID. Furthermore, the mice also displayed a learning and memory deficit in the Morris Water Maze, and also a possible muscular deficit. Overall, the results suggest that the novel zDHHC9 knockout mouse line may be a valuable resource to investigate zDHHC9 function in the brain and its links to ID
Advanced algorithms for polynomial matrix eigenvalue decomposition
Get PDFMatrix factorisations such as the eigen- (EVD) or singular value decomposition (SVD) offer optimality in often various senses to many narrowband signal processing algorithms.For broadband problems, where quantities such as MIMO transfer functions or cross spectral density matrices are conveniently described by polynomial matrices,such narrowband factorisations are suboptimal at best. To extend the utility of EVD and SVD to the broadband case, polynomial matrix factorisations have gained momentum over the past decade, and a number of iterative algorithms for particularly the polynomial matrix EVD (PEVD) have emerged. Existing iterative PEVD algorithms produce factorisations that are computationally costly (i) to calculate and (ii) to apply. For the former, iterative algorithms at every step eliminate off-diagonal energy, but this can be a slow process. For the latter,the polynomial order of the resulting factors, directly impacting on the implementation complexity, typically grows with every iteration of a PEVD algorithm. The work presented in this thesis helps to reduce both computational complexities. To address algorithmic complexity and convergence speed, this thesis firstly proposes a multiple shift approach, which can eliminate more off-diagonal energy at every iteration compared to existing methods. Equally applicable to the second order sequential best rotation (SBR2) algorithm, the idea here is applied to the family of sequential matrix diagonalisation (SMD) algorithms, for which a convergence proof is presented. Maximum energy transfer requires a very laborious parameter search; it is demonstrated that a very significant reduction of this search space can be gained while retaining almost all of the transfered energy. In addition a number of techniques have been developed which improve the efficiency of these PEVD algorithms. With each PEVD algorithm iteration the lengths of the polynomial eigenvectors and eigenvalues increase. To lower the order of the polynomial eigenvectors a novel truncation technique was developed which takes advantage of an ambiguity in the eigenvalues. A drawback of the multiple shift algorithms is the faster growth of the polynomial eigenvectors and eigenvalues; to mitigate this the search step has been analysed and modified to reduce the growth. The SMD algorithm suffers from a performance bottleneck in its EVD step. Here a cyclic-by-row Jacobi approximation is developed that significantly reduces the computational cost of the SMD algorithm with almost no impact on numerical performance. The impact of these algorithmic advances is studied in a number of applications. Firstly, based on a source model, it is investigated how the conditioning of the input data affects algorithm performance. Secondly, to highlight the benefit of the better converging multiple shift SMD algorithm it is compared to the SBR2 algorithm for broadband angle of arrival estimation, where the proposed multiple shift versions can achieve a more accurate subspace decomposition and hence a more accurate AoA estimation. Lastly, its demonstrated how PEVD algorithms can be utilised to extend other linear algebraic techniques to polynomial matrices. The example used here is the extension of the generalised eigenvalue decomposition (GEVD) to a polynomial matrix GEVD.Matrix factorisations such as the eigen- (EVD) or singular value decomposition (SVD) offer optimality in often various senses to many narrowband signal processing algorithms.For broadband problems, where quantities such as MIMO transfer functions or cross spectral density matrices are conveniently described by polynomial matrices,such narrowband factorisations are suboptimal at best. To extend the utility of EVD and SVD to the broadband case, polynomial matrix factorisations have gained momentum over the past decade, and a number of iterative algorithms for particularly the polynomial matrix EVD (PEVD) have emerged. Existing iterative PEVD algorithms produce factorisations that are computationally costly (i) to calculate and (ii) to apply. For the former, iterative algorithms at every step eliminate off-diagonal energy, but this can be a slow process. For the latter,the polynomial order of the resulting factors, directly impacting on the implementation complexity, typically grows with every iteration of a PEVD algorithm. The work presented in this thesis helps to reduce both computational complexities. To address algorithmic complexity and convergence speed, this thesis firstly proposes a multiple shift approach, which can eliminate more off-diagonal energy at every iteration compared to existing methods. Equally applicable to the second order sequential best rotation (SBR2) algorithm, the idea here is applied to the family of sequential matrix diagonalisation (SMD) algorithms, for which a convergence proof is presented. Maximum energy transfer requires a very laborious parameter search; it is demonstrated that a very significant reduction of this search space can be gained while retaining almost all of the transfered energy. In addition a number of techniques have been developed which improve the efficiency of these PEVD algorithms. With each PEVD algorithm iteration the lengths of the polynomial eigenvectors and eigenvalues increase. To lower the order of the polynomial eigenvectors a novel truncation technique was developed which takes advantage of an ambiguity in the eigenvalues. A drawback of the multiple shift algorithms is the faster growth of the polynomial eigenvectors and eigenvalues; to mitigate this the search step has been analysed and modified to reduce the growth. The SMD algorithm suffers from a performance bottleneck in its EVD step. Here a cyclic-by-row Jacobi approximation is developed that significantly reduces the computational cost of the SMD algorithm with almost no impact on numerical performance. The impact of these algorithmic advances is studied in a number of applications. Firstly, based on a source model, it is investigated how the conditioning of the input data affects algorithm performance. Secondly, to highlight the benefit of the better converging multiple shift SMD algorithm it is compared to the SBR2 algorithm for broadband angle of arrival estimation, where the proposed multiple shift versions can achieve a more accurate subspace decomposition and hence a more accurate AoA estimation. Lastly, its demonstrated how PEVD algorithms can be utilised to extend other linear algebraic techniques to polynomial matrices. The example used here is the extension of the generalised eigenvalue decomposition (GEVD) to a polynomial matrix GEVD
The development of a remotely expandable anastomosis device for use in paediatric cardiothoracic surgery
Get PDFCongenital Heart Disease (CHD) is one of the main causes of infant mortality in the UK and United States. In 2011, 47% of infants born with CHD in the UK did not survive to their first birthday. Often forms of CHD require the implantation of conduits (tubes) to redirect the blood flow around the reconstructed anatomy in the early days of life. In all cases, these children will require review surgery, often several times to replace the implanted conduit as the child grows. Critically every review procedure carries with it an increased mortality risk.;This project represents the first stage in the development of a conduit/ anastomosis technology, which will grow in response to the growing child, matching haemodynamic demand and eliminating the need for revision surgery.;Two prototypes were designed and developed. Each using an active drive mechanism to expand a stainless steel expansion strip with the slow and controlled movement required for such an application. Each of the designs were silicone coated using custom designed injection moulding techniques ensuring biocompatibility. Power and control systems were designed for each iteration taking into account their expansion methods to achieve the desired expansion rate of 12mm to 20mm in diameter of the implantation period.;Three consecutive implantation studies were carried out throughout the device's development. In the first study, the device was implanted into the right ventricle of a cadaveric pig to determine any potential impediments that would need to be addressed for implantation into live animal models. The second implantation was carried out on 6 piglets, successfully establishing CPB procedures;The third implantation was successfully carried out on 3 piglets, 2 of which survived the duration of the testing, confirming a successful encapsulation method to ensure biocompatibility, however it appeared the device failed to expand due to power line breakdown.Congenital Heart Disease (CHD) is one of the main causes of infant mortality in the UK and United States. In 2011, 47% of infants born with CHD in the UK did not survive to their first birthday. Often forms of CHD require the implantation of conduits (tubes) to redirect the blood flow around the reconstructed anatomy in the early days of life. In all cases, these children will require review surgery, often several times to replace the implanted conduit as the child grows. Critically every review procedure carries with it an increased mortality risk.;This project represents the first stage in the development of a conduit/ anastomosis technology, which will grow in response to the growing child, matching haemodynamic demand and eliminating the need for revision surgery.;Two prototypes were designed and developed. Each using an active drive mechanism to expand a stainless steel expansion strip with the slow and controlled movement required for such an application. Each of the designs were silicone coated using custom designed injection moulding techniques ensuring biocompatibility. Power and control systems were designed for each iteration taking into account their expansion methods to achieve the desired expansion rate of 12mm to 20mm in diameter of the implantation period.;Three consecutive implantation studies were carried out throughout the device's development. In the first study, the device was implanted into the right ventricle of a cadaveric pig to determine any potential impediments that would need to be addressed for implantation into live animal models. The second implantation was carried out on 6 piglets, successfully establishing CPB procedures;The third implantation was successfully carried out on 3 piglets, 2 of which survived the duration of the testing, confirming a successful encapsulation method to ensure biocompatibility, however it appeared the device failed to expand due to power line breakdown
On the evaluation of high temperature creep-fatigue responses of structures
Get PDFThis thesis details the research and development steps undertaken to improve and enhance the Linear Matching Method Framework (LMMF) for high temperature structural integrity assessment. Firstly, a more accurate interpolation method is developed to calculate the revised yield stress, which is necessary to calculate the creep rupture limit. Convergence issues are also investigated to improve the numerical difficulties often suffered by the "uctuating" revised yield stress. Secondly, the extended Direct Steady Cycle Analysis (eDSCA) is used to solve two complex case studies and a comprehensive theoretical study is undertaken on the effect of combined creep and cyclicplasticity. This subroutine is enhanced by adding many new features, including the possibility to calculate the fatigue damage using only few tensile constants.;Multiple creep dwells within the load cycle can be considered and improvements are made to the cyclic loop construction. In order to make the eDSCA more accurate for creep damage assessment, the stress modifed ductility exhaustion method is implemented combining the advantages of both stress and strain based methods. Thirdly, several case studies on creep-fatigue interaction are discussed thanks to the flexibility of the LMMF including the detrimental effect of creep-ratchetting. Furthermore, a new potentially dangerous mechanism involving the sudden accumulation of plastic strain during the creep dwell is identified and studied by numerical means.This thesis details the research and development steps undertaken to improve and enhance the Linear Matching Method Framework (LMMF) for high temperature structural integrity assessment. Firstly, a more accurate interpolation method is developed to calculate the revised yield stress, which is necessary to calculate the creep rupture limit. Convergence issues are also investigated to improve the numerical difficulties often suffered by the "uctuating" revised yield stress. Secondly, the extended Direct Steady Cycle Analysis (eDSCA) is used to solve two complex case studies and a comprehensive theoretical study is undertaken on the effect of combined creep and cyclicplasticity. This subroutine is enhanced by adding many new features, including the possibility to calculate the fatigue damage using only few tensile constants.;Multiple creep dwells within the load cycle can be considered and improvements are made to the cyclic loop construction. In order to make the eDSCA more accurate for creep damage assessment, the stress modifed ductility exhaustion method is implemented combining the advantages of both stress and strain based methods. Thirdly, several case studies on creep-fatigue interaction are discussed thanks to the flexibility of the LMMF including the detrimental effect of creep-ratchetting. Furthermore, a new potentially dangerous mechanism involving the sudden accumulation of plastic strain during the creep dwell is identified and studied by numerical means
Access Repository Dashboard
Do you manage STAX (Strathclyde Repository)? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!