Kanban and Constant Work-In-Process (CONWIP) control methods are designed to impose tight controls over inventory, while providing a satisfactory production rate. This paper generalizes systems with kanban or CONWIP control as assembly/disassembly networks with multiple-loop structures. We present a stochastic mathematical model which integrates the information control flows into material flows. Graph theory is used to analyze the multiple-loop structures. An efficient analytical algorithm is developed for evaluating the expected production rate and inventory levels. The performance of the algorithm is reported in terms of accuracy, reliability and speed. / Singapore-MIT Alliance (SMA)

Prolidase has potential applications in cheese debittering, organophosphate detoxification and as an enzyme replacement therapy in prolidase-deficient patients. Recombinant Lactococcus lactis prolidases and their catalytic properties have previously been characterized in Dr. Tanaka's research group. Unlike other prolidases, L. lactis prolidase shows allosteric behaviour, metal-dependent substrate specificity and substrate inhibition. The current project focuses on elucidating the three-dimensional structure of L. lactis prolidase using X-ray crystallography. Hexagonal plate-like crystals of wild-type L. lactis prolidase were grown by the hanging drop vapour diffusion method, allowing the crystals to grow to about 50 µm in their longest dimension. The crystallization cocktail in which they grew contained 0.08 M sodium cacodylate (pH 6.5), 0.16 M calcium acetate, 14 % PEG 8000 and 18 % glycerol. Crystal diffraction data was collected at a wavelength of 0.9795 Å on beamline 08ID-1 of the Canadian Macromolecular Crystallography Facility at the Canadian Light Source and was processed using X-ray Detector Software. The crystals belonged to space group C2 and estimated to contain three molecules in an asymmetric unit. The electron density map of this structure was solved by the molecular replacement method and the structure model was refined against 2.25 Å resolution data. Molecule A forms a dimer with molecule B, while molecule C forms a dimer with molecule C', which is located in the neighbouring crystal asymmetric unit. The electron density of molecule A was well-defined and complete. Therefore, all the 362 amino acid residues of L. lactis prolidase were fitted. The other two molecules were incomplete and less defined. Only 360 and 352 residues could be fitted in molecules B and C, respectively. Molecule C, the worst of the three, compromised the overall quality of the refined structure. However, the functional interpretation of the structure was not compromised since the well-defined molecules form a dimer with each other and the biologically-functional form of L. lactis prolidase is a homodimer. The final Rwork and Rfree are 22.39 and 27.77, respectively. Comparison with other known prolidases revealed that Asp 36 and His 38 are unique to L. lactis prolidase. These residues have been shown to be involved in the allosteric behaviour and substrate inhibition of this enzyme, respectively. Therefore, this crystal structure further supports their suggested contribution in L. lactis prolidase's unique catalytic properties.

Cys-loop receptors are ligand-gated ion channels that are involved in fast synaptic neurotransmission in the central and peripheral nervous system. The Cys-loop receptor RDL ('resistant to dieldrin') is a GABA-gated chloride channel from Drosophila melanogaster and is a major target site for insecticides. The aim of this dissertation was to characterise RDL receptors with particular focus on the agonist binding site. To assess the potency of a range of GABA analogues on RDL receptors, I expressed receptors in Xenopus oocytes and used voltage-clamp electrophysiology to detect receptor responses. I carried out computational modelling of these analogues to determine the dipole separation distances and atomic charges. Computational calculations and functional experiments revealed that agonists require a charged ammonium and an anionic centre, with the most potent agonists having a dipole separation distance of ~5 Å. I made a homology model of the extracellular domain of RDL and docked the active analogues into the putative binding site. I then conducted mutagenesis studies to test the accuracy of this model. Functional data from mutagenesis studies broadly support the location of GABA within this model. This model may be useful for further structure-activity studies and rational drug design. Natural compounds from the traditional Chinese medicine 'Ginkgo biloba' (ginkgolide A, ginkgolide B and bilobalide) have potent insecticidal properties and are similar in structure to picrotoxin. I tested the effect of these compounds on RDL receptor function using voltage-clamp electrophysiology. All compounds were found to inhibit RDL receptor function. I probed the binding site of these compounds using site-directed mutagenesis and electrophysiology. Mutations to the 2'A and 6'T channel-lining (M2) residues greatly reduced the potency of these compounds. I then made a homology model of the transmembrane domain of RDL and docked these compounds into the channel. Compounds docked into the channel pore close to the 2' and 6' channel-lining residues and H-bonding interactions were detected at these locations. Ginkgolides are therefore antagonists of RDL receptors, binding in the channel close to the 2' and 6' residues and this may be the mechanism underlying their potent insecticidal properties. The 5-HT3 receptor is a member of the Cys-loop receptor family and shows homology to RDL receptors. To explore different techniques for studying Cys-loop receptor function I assessed the functionality of two brain derived transcripts of the 5-HT3B subunit (Br1 and Br2) using single-channel electrophysiology and a fluorometric assay. Receptors containing Br1 were found to have a conductance identical to the 5-HT3B subunit whilst Br2 receptors were found not to be expressed. This finding has implications for 5-HT3 brain signalling, in which Br1 may play an important role. In conclusion, work here has described how agonists bind to and activate RDL GABA receptors and I have identified a candidate mechanism for the potent insecticidal properties of Ginkgo biloba extracts. I have also confirmed that 5-HT3 receptor brain transcript Br1 forms functional channels with similar properties to the 5-HT3B subunit.

In the last decade, deep learning models have been successfully applied to a variety of applications and solved many tasks. The ultimate goal of this study is to produce deep learning models to improve the skills of forecasting ocean dynamic events in general and those of the Loop Current (LC) system in particular. A specific forecast target is to predict the geographic location of the (LC) extension and duration, LC eddy shedding events for a long lead time with high accuracy. Also, this study aims to improve the predictability of velocity fields (or more precisely, velocity volumes) of subsurface currents. In this dissertation, several deep learning based prediction models have been proposed. The core of these models is the Long-Short Term Memory (LSTM) network. This type of recurrent neural network is trained with Sea Surface Height (SSH) and LC velocity datasets. The hyperparameters of these models are tuned according to each model's characteristics and data complexity. Prior to training, SSH and velocity data are decomposed into their temporal and spatial counterparts.A model uses the Robust Principle Component Analysis is first proposed, which produces a six-week lead time in forecasting SSH evolution. Next, the Wavelet+EOF+LSTM (WELL) model is proposed to improve the forecasting capability of a prediction model. This model is tested on the prediction of two LC eddies, namely eddy Cameron and Darwin. It is shown that the WELL model can predict the separation of both eddies 10 and 14 weeks ahead respectively, which is two more weeks than the DAC model. Furthermore, the WELL model overcomes the problem due to the partitioning step involved in the DAC model. For subsurface currents forecasting, a layer partitioning method is proposed to predict the subsurface field of the LC system. A weighted average fusion is used to improve the consistency of the predicted layers of the 3D subsurface velocity field. The main challenge of forecasting of the LC and its eddies is the small number of events that have occurred over time, which is only once or twice a year, which makes the training task difficult. Forecasting the velocity of subsurface currents is equally challenging because of the limited insitu measurements. / Includes bibliography. / Dissertation (PhD)--Florida Atlantic University, 2021. / FAU Electronic Theses and Dissertations Collection

This thesis proposes an improved notion of coefficient system for Lagrangian Floer Homology which allows one to produce nontrivial invariants away from characteristic 2, even when coherent orientations of moduli spaces of Floer trajectories do not exist. This explains a suggestion of Witten. The invariant can be computed in examples, and the method explained below should be extensible to other Floer-theoretic invariants. The basic idea is that the moduli spaces of curves admit fundamental classes in homology with coefficients in the orientation lines of the moduli spaces, and the usual construction of coherent orientations actually shows that these fundamental classes naturally map to spaces of paths twisted with appropriate coefficient systems. These twisted path spaces admit enough algebraic structure to make sense of Floer homology with coefficients in these path spaces.

The increase in solar energy installation capacity and the versatility of modern power inverters have enabled widespread penetration of distributed generation in modern power systems. Islanding detection techniques allow for fast identification and corrective action in the face of abnormal events. Current standards specify the operational limits for voltage, frequency, and detection time. Grid codes specify the procedures for disconnection to establish safe network maintenance conditions. Passive, active, and remote techniques require voltage, current, and frequency measurements and the definition of thresholds for detection. Operational parameters such as load mismatch and quality factors influence the detection capabilities. False-positive triggering due to grid transients can lead to unnecessary disconnection of distributed generation resources. Cybersecurity threats pose a critical challenge for power systems and can result in significant operational disruptions and security risks. In particular, when a power system initiates communication links between different nodes or ends, it becomes more vulnerable to various forms of cyber-attacks. As such, it is imperative to address the potential cybersecurity risks associated with communication links. Through a literature review, this work analyzes the performance of several islanding detection techniques and proposes a modified 9-bus benchmark system to verify the robustness of passive and active methods against false-positive detections upon severe grid-side transients. Furthermore, this thesis conducts a detailed analysis of cyber-attacks on the remote islanding detection technique, using a real-time simulator to assess the potential impact of such attacks on the technique's effectiveness by simulating various attack scenarios. The findings of this analysis can help power system operators to better protect their systems from cyber-attacks and ensure the reliable operation of their distributed generation resources. Moreover, it discusses a conceptual implementation of hardware-in-the-loop testing. The modeling of the systems is discussed. Guidelines and international standards are presented. Various setups for experimental work are suggested and implemented.

We combine the evanescent field (EF) sensing mechanism with the fiber loop ringdown (FLRD) sensing scheme to create FLRD-EF sensors. The EF sensor heads are fabricated by etching the cladding of a single-mode fiber (SMF), while monitoring the etching process by the FLRD technique in real-time, on-line with high control precision. The effect of the sensor head dimensions on the sensors' detection sensitivity and response time are investigated. The EF scattering (EFS) sensing mechanism is combined with the FLRD detection scheme to create a new type of fiber optic index sensor. The detection limit for an optical index change is 3.2×10-5. This is the highest sensitivity for a fiber optic index sensor so far, without using any chemical-coating or optical components at the sensor head. A new type of index-based biosensor using high sensitivity FLRDEFS technique to sense deoxyribonucleic acid (DNA) and bacteria (Escherichia coli) is created.

Supply chain management has evolved over the course of history in order to provide faster and efficient service to those companies that follow its principles. As there have been advances in technology and changes in the way business is conducted across the globe, supply chains also have had to change in order to remain effective. With greater attention paid to resource depletion, environmental impact, and waste reduction; the concept of closed-loop supply chains has garnered the attention of managers who look to make their production processes more efficient. Finding ways to judge the performance of these supply chains is critical to managers. By identifying key performance measures, they are able to gauge how their closed-loop process is performing as well as identify areas for improvement.

Page 139 not included in the thesis. / <p> The problem of coherent carrier recovery and the effects of phase error on the performance of an offset quadrature-phase-shift-keying (QPSK) duobinary system have been investigated in the thesis. The system of interest is similar to RD -3 digital system that is being developed and installed as an efficient high data-rate digital radio communication system by Bell Northern Research Laboratory (BNR). </p> <p> Four carrier regeneration loop structures are investigated and analysed in the thesis. These are: (i) estimate-aided suppressed carrier loop (ii) decision-directed feedback loop (iii) shifted decision-directed feedback loop (iv) half-shifted decision-directed feedback loop All of these loop structures employ the technique of data-aided carrier synchronization. The estimate-aided loop structure exhibits steadystate behavior similar to that of a conventional Costas loop. The remaining three loop structures differ from the estimate-aided loop in the sense that they require decisions to make on the noisy received signal. These are then fedback to the carrier recovery circuit in such a way as to create a spectral line at carrier frequency. The loop behavior in the presence of additive noise has been investigated in some detail. For each loop, analytical expressions for the phase detector characteristic (S-curve) and the steady-state phase error probability density function (pdf) are derived, and provide a means of comparing the performance of the different recovery schemes.</p> / Thesis / Master of Engineering (MEngr)

Classification is a pervasive problem in research that aims at grouping items in categories according to established criteria. There are two prevalent ways to classify items of interest: i) to train and exploit machine learning (ML) algorithms or ii) to resort to human classification (via experts or crowdsourcing). Machine Learning algorithms have been rapidly improving with an impressive performance in complex problems such as object recognition and natural language understanding. However, in many cases they cannot yet deliver the required levels of precision and recall, typically due to difficulty of the problem and (lack of) availability of sufficiently large and clean datasets. Research in crowdsourcing has also made impressive progress in the last few years, and the crowd has been shown to perform well even in difficult tasks [Callaghan et al., 2018; Ranard et al., 2014]. However, crowdsourcing remains expensive, especially when aiming at high levels of accuracy, which often implies collecting more votes per item to make classification more robust to workers' errors. Recently, we witness rapidly emerging the third direction of hybrid crowd-machine classification that can achieve superior performance by combining the cost-effectiveness of automatic machine classifiers with the accuracy of human judgment. In this thesis, we focus on designing crowdsourcing strategies and hybrid crowd-machine approaches that optimize the item classification problem in terms of results and budget. We start by investigating crowd-based classification under the budget constraint with different loss implications, i.,e., when false positive and false negative errors carry different harm to the task. Further, we propose and validate a probabilistic crowd classification algorithm that iteratively estimates the statistical parameters of the task and data to efficiently manage the accuracy vs. cost trade-off. We then investigate how the crowd and machines can support each other in tackling classification problems. We present and evaluate a set of hybrid strategies balancing between investing money in building machines and exploiting them jointly with crowd-based classifiers. While analyzing our results of crowd and hybrid classification, we found it is relevant to study the problem of quality of crowd observations and their confusions as well as another promising direction of linking entities from structured and unstructured sources of data. We propose crowd and neural network grounded algorithms to cope with these challenges followed by rich evaluation on synthetic and real-world datasets.