This thesis focuses on subspace based data-driven modeling and control techniques for batch and continuous processes. Motivated by the increasing amount of process data, data-driven modeling approaches have become more popular. These approaches are better in comparison to first-principles models due to their ability to capture true process dynamics. However, data-driven models rely solely on mathematical correlations and are subject to overfitting. As such, applying first-principles based constraints to the subspace model can lead to better predictions and subsequently better control. This thesis demonstrates that the addition of process gain constraints leads to a more accurate constrained model. In addition, this thesis also shows that using the constrained model in a model predictive control (MPC) algorithm allows the system to reach desired setpoints faster. The novel MPC algorithm described in this thesis is specially designed as a quadratic program to include a feedthrough matrix. This is traditionally ignored in industry however this thesis portrays that its inclusion leads to more accurate process control. Given the importance of accurate process data during model identification, the missing data problem is another area that needs improvement. There are two main scenarios with missing data: infrequent sampling/ sensor errors and quality variables. In the infrequent sampling case, data points are missing in set intervals and so correlating between different batches is not possible as the data is missing in the same place everywhere. The quality variable case is different in that quality measurements require additional expensive test making them unavailable for over 90\% of the observations at the regular sampling frequency. This thesis presents a novel subspace approach using partial least squares and principal component analysis to identify a subspace model. This algorithm is used to solve each case of missing data in both simulation (polymethyl methacrylate) and industrial (bioreactor) processes with improved performance. / Dissertation / Doctor of Philosophy (PhD) / An important consideration of chemical processes is the maximization of production and product quality. To that end developing an accurate controller is necessary to avoid wasting resources and off-spec products. All advance process control approaches rely on the accuracy of the process model, therefore, it is important to identify the best model. This thesis presents two novel subspace based modeling approaches the first using first principles based constraints and the second handling missing data approaches. These models are then applied to a modified state space model with a predictive control strategy to show that the improved models lead to improved control. The approaches in this work are tested on both simulation (polymethyl methacrylate) and industrial (bioreactor) processes.

No description available.

No description available.

Many processes involving rotating machinery could benefit from the continuous feedback of force applied to the bearings that support the machinery. Such a system could be used to provide diagnostics for process monitoring in a manufacturing application or to provide information for machine health monitoring. Active Magnetic Bearings (AMBs) have the capability to act concurrently as a shaft force sensor and support bearing. This capability stems from the AMB's control system, which is designed to maintain a specific rotor position, regardless of forces acting on the rotor. Researchers have demonstrated the force sensing ability of AMBs; current state of the art methods typically rely on a direct measurement of magnetic flux density as provided by a Hall probe inserted in the magnetic field. In this work, a system identification approach to force measurement is proposed; the proposed approach is applicable to all active magnetic bearings and does not require Hall probes. Recent developments in system identification of bearing forces (Kasarda et al., 2000) indicate that a different approach is feasible. In the work of Marshall (Marshall et al., 2001), a variety of perturbations are applied to an AMB while the AMB controller signals are interrogated, no outside instrumentation such as force transducers or Hall probes are required. The work of Kasarda and Marshall is the starting point for the work presented here. The initial work was expanded to include a general characterization of air gap for any rotor position. Although this characterization relies on static testing to identify system parameters, the identified parameters can then be used in the measurement of dynamic forces. The identification procedure provides a measurement of effective air gap length. Effective gap length is used to infer the effective position of the rotor with respect to the stator. This measurement is made for several specific rotor locations. The relationship between the effective rotor positions provided by the identification and the rotor positions reported by the AMB system sensors establishes a coordinate transformation. The procedure is also applied at different shaft rotation angles. In this way rotor runout can be identified. / Ph. D.

L’habilité de visuellement ré-identifier des objets est une capacité fondamentale des systèmes de vision. Souvent, ces systèmes s’appuient sur une collection de signatures visuelles basées sur des descripteurs comme SIFT ou SURF. Cependant, ces descripteurs traditionnels ont été conçus pour un certain domaine d’aspects et de géométries de surface (relief limité). Par conséquent, les surfaces très texturées telles que l’écorce des arbres leur posent un défi. Alors, cela rend plus difficile l’utilisation des arbres comme points de repère identifiables à des fins de navigation (robotique) ou le suivi du bois abattu le long d’une chaîne logistique (logistique). Nous proposons donc d’utiliser des descripteurs basés sur les données, qui une fois entraîné avec des images d’écorce, permettront la ré-identification de surfaces d’arbres. À cet effet, nous avons collecté un grand ensemble de données contenant 2 400 images d’écorce présentant de forts changements d’éclairage, annotées par surface et avec la possibilité d’être alignées au pixels près. Nous avons utilisé cet ensemble de données pour échantillonner parmis plus de 2 millions de parcelle d’image de 64x64 pixels afin d’entraîner nos nouveaux descripteurs locaux DeepBark et SqueezeBark. Notre méthode DeepBark a montré un net avantage par rapport aux descripteurs fabriqués à la main SIFT et SURF. Par exemple, nous avons démontré que DeepBark peut atteindre une mAP de 87.2% lorsqu’il doit retrouver 11 images d’écorce pertinentes, i.e correspondant à la même surface physique, à une image requête parmis 7,900 images. Notre travail suggère donc qu’il est possible de ré-identifier la surfaces des arbres dans un contexte difficile, tout en rendant public un nouvel ensemble de données. / The ability to visually re-identify objects is a fundamental capability in vision systems. Oftentimes,it relies on collections of visual signatures based on descriptors, such as SIFT orSURF. However, these traditional descriptors were designed for a certain domain of surface appearances and geometries (limited relief). Consequently, highly-textured surfaces such as tree bark pose a challenge to them. In turn, this makes it more difficult to use trees as identifiable landmarks for navigational purposes (robotics) or to track felled lumber along a supply chain (logistics). We thus propose to use data-driven descriptors trained on bark images for tree surface re-identification. To this effect, we collected a large dataset containing 2,400 bark images with strong illumination changes, annotated by surface and with the ability to pixel align them. We used this dataset to sample from more than 2 million 64 64 pixel patches to train our novel local descriptors DeepBark and SqueezeBark. Our DeepBark method has shown a clear advantage against the hand-crafted descriptors SIFT and SURF. For instance, we demonstrated that DeepBark can reach a mAP of 87.2% when retrieving 11 relevant barkimages, i.e. corresponding to the same physical surface, to a bark query against 7,900 images. ur work thus suggests that re-identifying tree surfaces in a challenging illuminations contextis possible. We also make public our dataset, which can be used to benchmark surfacere-identification techniques.

Yes / Identifying and explaining the pathophysiology of sepsis, as well as the importance of monitoring for indicators of patient deterioration in sepsis.

In this thesis, Verizon's discourse was analyzed as a case study for corporate rhetoric. This rhetorical analysis examined Verizon's discourse before, during and after the crisis of 9/11. Press releases, executive speeches, and other internal documents were analyzed while media reports provided a context for the discourse. Renewal discourse coupled with the theories of organizational identification and values advocacy provided the basic framework for this thesis. Verizon's crisis situation during the crisis of 9/11 provided a unique context in which to study organizational discourse. By placing the response within a larger frame of the organization's history, this study was able to provide insight into traditional crisis response scholarship that often only focuses on the immediate response to crisis. / Master of Arts

För att känna igen eller identifiera egna förband och enheter har man sedan långt tillbaka använt sig av olika metoder. Oftast har dessa utgjorts av olika symboler och tecken på fanor, flaggor, uniformer med mera. I nutid har allt fler olika tekniska system för igenkänning och identifiering, eller så kallad Combat Identification (CID), av egna förband, börjat användas. Dessa system uppvisar mer eller mindre komplexitet, har olika för- respektive nackdelar och bidrar i olika utsträckning till förmågan att framgångsrikt genomföra CID. Syftet med uppsatsen är att jämföra olika CID-system konstruerade för avsutten trupp för att öka kunskapen avseende den militära nytta systemen har med avseende på minskad risk för vådabekämpning när ett förband sätts in i urban miljö. Vid jämförelsen av de olika systemen kan konstateras att CID-systemen endast ger militär nytta när ett förband har en uppgift där stridsförloppen är långsamma, till exempel övervakning. De CID-system som särskilt utmärker sig vid övervakning är TIR-systemen. / Since long ago different methods have been used in order to recognize or identify own units. Most often the methods consisted of different symbols and signs placed on banners, flags, uniforms and so on. Today several different technical systems for recognition and identification of own units, so called Combat Identification (CID), are entering service. The systems show more or less complexity in their construction, hold advantages as well as disadvantages and thus contribute in varying degree to the ability to successfully carry out CID.  The purpose of this essay is to compare different CID-systems designed for ground soldiers, in order to increase the knowledge of what military utility the different systems may have in terms of reducing the risk of friendly fire when a unit is deployed in an urban environment. When comparing the different systems it is clearly that CID-systems only give military utility when a unit has a task where the battle progress is slow, such as surveillance. The CID-system that especially excels when used in surveillance is the TIR-systems.

Yu, Jing. / "November 2011." / Thesis (M.Phil.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 161-172). / Abstracts in English and Chinese. / Abstract --- p.I / 摘要 --- p.III / Acknowledgements --- p.V / Contents --- p.VI / List of Figures --- p.VIII / List of Tables --- p.X / Abbreviations and Symbols --- p.XII / Chapter CHAPTER 1 --- Introduction --- p.1 / Chapter 1.1 --- Phylogenetic study of Hedyotis --- p.2 / Chapter 1.1.1 --- Rubiaceae --- p.2 / Chapter 1.1.2 --- Controversial taxonomic issues --- p.9 / Chapter 1.2 --- Traditional Chinese medicine (TCM) --- p.19 / Chapter 1.2.1 --- Introduction --- p.19 / Chapter 1.2.2 --- Baihuasheshecao --- p.20 / Chapter 1.2.3 --- Authentication of Baihuasheshecao --- p.23 / Chapter 1.3 --- Icefishes in the Hong Kong market --- p.23 / Chapter 1.3.1 --- Introduction --- p.23 / Chapter 1.4 --- Molecular approach --- p.25 / Chapter 1.4.1 --- Introduction to molecular phylogeny --- p.25 / Chapter 1.4.2 --- FINS (Forensically Informative Nucleotide Sequencing) for species identification --- p.27 / Chapter 1.4.3 --- DNA sequence markers --- p.28 / Chapter 1.5 --- Objectives --- p.38 / Chapter CHAPTER 2 --- MATERIALS AND METHODOLOGY --- p.41 / Chapter 2.1 --- Materials --- p.42 / Chapter 2.2 --- DNA extraction --- p.50 / Chapter 2.3 --- Polymerase chain reaction (PCR) method --- p.51 / Chapter 2.4 --- Gel electrophoresis --- p.54 / Chapter 2.5 --- PCR production purification --- p.54 / Chapter 2.6 --- Ligation and transformation and transformation of PCR product --- p.56 / Chapter 2.7 --- DNA sequencing and sequence analyses --- p.58 / Chapter CHAPTER 3 --- USING FORENSICALLY INFORMATIVE NUCLEOTIDE SEQUENCING (FINS) TECHNOLOGY FOR SPECIES IDENTIFICATION --- p.64 / Chapter 3.1 --- Authentication of Baihuasheshecao by FINS Analysis --- p.65 / Chapter 3.1.1 --- Authentication using FINS technology --- p.65 / Chapter 3.1.2 --- Relative effectiveness of DNA regions for FINS analysis --- p.70 / Chapter 3.1.3 --- Phylogenetic interpretation --- p.72 / Chapter 3.2 --- Authentication of Salangids (Icefishes) by FINS Analysis --- p.74 / Chapter 3.2.1 --- Analysis based on mitochondrial ribosome DNA region --- p.74 / Chapter 3.2.2 --- Analysis based on mitochondrial 16S rRNA --- p.77 / Chapter 3.2.3 --- Analysis based on combined regions --- p.79 / Chapter 3.2.4 --- Phylogenetic analysis --- p.81 / Chapter 3.2.5 --- Discussion --- p.85 / Chapter 3.3 --- Conclusions --- p.88 / Chapter CHAPTER 4 --- PHYLOGENTIC STUDY OF HEDYOTIS IN CHINA AND THEIR POSITION IN SPERMACOCEAE --- p.89 / Chapter 4.1 --- Phylogentic study of Hedyotis species in Chinese --- p.90 / Chapter 4.1.1 --- Nuclear ITS region --- p.90 / Chapter 4.1.2 --- Plastid trnL intron and trnL-V intergenic spacer region --- p.94 / Chapter 4.1.3 --- Plastid trnH-psbA intergenic spacer region --- p.98 / Chapter 4.1.4 --- Plastid rbcL region --- p.102 / Chapter 4.1.5 --- Plastid matK region --- p.106 / Chapter 4.1.7 --- Combined analysis --- p.114 / Chapter 4.2 --- The phylogenetic position of Hedyotis (species in China) in the tribe of Spermacoceae s.1 --- p.121 / Chapter 4.2.1 --- Plastid trnL - F intergenic spacer region --- p.121 / Chapter 4.2.2 --- Plastid rbcL region --- p.133 / Chapter 4.2.3 --- Plastid rps16 region --- p.141 / Chapter 4.3 --- Discussion --- p.153 / Chapter 4.3.1 --- Comparison of phylogenetic utility of the six DNA regions --- p.153 / Chapter 4.3.2 --- Diplophragma section --- p.154 / Chapter 4.3.3 --- "Hedyotis, Dimetia, Euoldendandia and Gonotheca sections" --- p.156 / Chapter 4.3.4 --- The position of Hedyotis (species in China) in Spermacoceae --- p.158 / Chapter 4.4 --- Conclusions --- p.160 / REFERENCES --- p.161 / APPENDIX --- p.173

Speech signal is basically meant to carry the information about the linguistic message. But, it also contains the speaker-specific information. It is generated by acoustically exciting the cavities of the mouth and nose, and can be used to recognize (identify/verify) a person. This thesis deals with the speaker identification task; i.e., to find the identity of a person using his/her speech from a group of persons already enrolled during the training phase. Listeners use many audible cues in identifying speakers. These cues range from high level cues such as semantics and linguistics of the speech, to low level cues relating to the speaker's vocal tract and voice source characteristics. Generally, the vocal tract characteristics are modeled in modern day speaker identification systems by cepstral coefficients. Although, these coeficients are good at representing vocal tract information, they can be supplemented by using both pitch and voicing information. Pitch provides very important and useful information for identifying speakers. In the current speaker recognition systems, it is very rarely used as it cannot be reliably extracted, and is not always present in the speech signal. In this thesis, an attempt is made to utilize this pitch and voicing information for speaker identification. This thesis illustrates, through the use of a text-independent speaker identification system, the reasonable performance of the cepstral coefficients, achieving an identification error of 6%. Using pitch as a feature in a straight forward manner results in identification errors in the range of 86% to 94%, and this is not very helpful. The two main reasons why the direct use of pitch as a feature does not work for speaker recognition are listed below. First, the speech is not always periodic; only about half of the frames are voiced. Thus, pitch can not be estimated for half of the frames (i.e. for unvoiced frames). The problem is how to account for pitch information for the unvoiced frames during recognition phase. Second, the pitch estimation methods are not very reliable. They classify some of the frames unvoiced when they are really voiced. Also, they make pitch estimation errors (such as doubling or halving of pitch value depending on the method). In order to use pitch information for speaker recognition, we have to overcome these problems. We need a method which does not use the pitch value directly as feature and which should work for voiced as well as unvoiced frames in a reliable manner. We propose here a method which uses the autocorrelation function of the given frame to derive pitch-related features. We call these features the maximum autocorrelation value (MACV) features. These features can be extracted for voiced as well as unvoiced frames and do not suffer from the pitch doubling or halving type of pitch estimation errors. Using these MACV features along with the cepstral features, the speaker identification performance is improved by 45%.