published_or_final_version / Mechanical Engineering / Master / Master of Philosophy

M.Ing.

Orientador: José Alfredo Covolan Ulson / Banca: Danilo Hernane Spatti / Banca: Fernando de Souza Campos / Resumo: O uso de técnicas de controle preditivo e de sistemas inteligentes na resolução de problemas de controle de grande complexidade tem crescido significativamente na última década. Parte disso se deve ao fato destas abordagens já terem sido utilizadas com sucesso em uma vasta gama de aplicações, demonstrando sua robustez e eficiência. Porém, o uso destas técnicas ainda é pequeno quando comparado com o controle PID (proporcional integral derivativo) clássico, ainda que apresentem resultados melhores. Assim, a maioria das publicações presentes na literatura que utilizam estas técnicas de controle mais sofisticadas tem como objetivo resolver um problema específico de controle. Para isso, em geral, adota-se uma determinada abordagem sem que necessariamente se faça uma avaliação a respeito de qual dos métodos existentes seria o mais eficiente, uma vez que o foco é apenas a resolução do problema. Nesse contexto, o objetivo deste trabalho foi avaliar o desempenho de três das técnicas de controle inteligente e/ou preditivo mais aceitas para o controle de processos industriais frente ao controlador PID clássico, de uso generalizado na indústria. Mais especificamente, optou-se por utilizar o controlador MPC, o controlador Fuzzy e o controlador Neural Preditivo para o controle de um processo industrial e, então, avaliar o desempenho por meio das métricas ISE e ITSE, além de comparar o sobressinal máximo, erro estacionário e tempo de acomodação. Por fim, os resultados de desempenho de cada ... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: The use of predictive control techniques and intelligent systems in the resolution of complex control problems has increased in the last decade. Part of this is due to the fact that these approaches have already been used successfully in a wide range of applications, demonstrating their robustness and efficiency. However, the use of these techniques is still small when compared to the classic PID (derivative proportional integral) control, although they present better results. Thus, most publications in the literature that use these more sophisticated control techniques aim to solve a specific control problem. For this, in general, a certain approach is adopted without necessarily making an evaluation as to which of the existing methods would be the most efficient, since the focus is only the resolution of the problem. In this context, the objective of this work was to evaluate the performance of three of the most accepted intelligent and / or predictive control techniques for the control of industrial processes against the classical PID controller, widely used in industry. More specifically, it was decided to use the MPC controller, the Fuzzy controller and the Neural Predictive controller to control an industrial process, and then evaluate the performance through the ISE and ITSE metrics, in addition to comparing the maximum overshoot, stationary error and accommodation time. Finally, the performance results of each controller were confronted with the classic PID controller... (Complete abstract click electronic access below) / Mestre

In this thesis, a digital-PID-control single-inductor triple-output (SITO) DC-DC converter is realized in AMS 0.35mum CMOS technology. The size of the chip is about 1600 mum x 1700 mum. To improve load current and reduce cross regulation, a Pre-Sub-Period inductor-current regulation is proposed. Based on the maximum duty cycle limiter, an adaptive inductor current adjustment is realized when the duty cycle of the digital PWM signal is larger than the set maximum duty cycle. By an optimized phase control sequence, the S&H stages of the feedback switching and ADC are controlled to on/off with a minimized delay time. Moreover, the control sequence can virtually remove the setting time. / Multiple voltage supplies are necessary to satisfy the different voltage supply requirements of the different on-chip blocks to reduce power consumption in modem electronic devices, such as the modem embedded systems, the portable devices, personal computing devices and wireless communications and imaging systems. For example, WiMAX transmitter includes different sub-blocks: Baseband processor, IQ modulator and power amplifier. Different blocks should operate with the different power supply voltages to satisfy the different requirements. / Single-input multiple-output DC-DC converter is presented to provide the different voltage supplies and reduce the cost on the elements such as the inductor on PCB and save PCB area. Meanwhile, to remove cross regulation and improve load driving capability, the DC-DC converter should operate in the pseudo-continuous mode/discontinuous mode (P-CCM/DCM). However, in the previous designs, the DC current in the inductor is fixed. When the load becomes heavy enough, cross regulation will significantly affect across the different sub-converters. / Jia, Jingbin. / "December 2009." / Adviser: KaNang Leung Alex. / Source: Dissertation Abstracts International, Volume: 72-01, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2010. / Includes bibliographical references (leaves 121-124). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

Les contrôleurs du VIH (HICs) sont de rares patients qui contrôlent spontanément la réplication du VIH en absence de thérapie antirétrovirale. Ces patients sont caractérisés par un taux normal de cellules T CD4+, le maintien d’une charge virale indétectable (<50 copies d’ARN viral/mL de plasma), et présentent un très faible risque de progression vers le SIDA. De nombreux travaux suggèrent que le contrôle de la réplication virale chez ces patients est dû à une réponse cellulaire antivirale particulièrement efficace. Notre équipe a récemment montré que les HICs inclus dans la Cohorte ANRS CO21 CODEX maintiennent des réponses T CD4+ très sensibles, associées à l'expression de TCRs ayant une forte avidité pour certains peptides Gag du VIH. Nous avons montré en particulier que les cellules T CD4+ de HICs répondent à de très faibles concentrations de l’épitope immunodominant Gag293. Dans une première étude, nous avons pu mettre en évidence que la réponse de haute avidité observée pour les cellules T CD4+ des HICs semble s'expliquer par une propriété intrinsèque de leur TCR. Pour identifier les déterminants moléculaires qui sous-tendent cette réponse de haute avidité, nous avons caractérisé le répertoire TCR des cellules T CD4+ spécifiques de l’épitope immunodominant Gag293. Les HICs présentent un répertoire TCR hautement biaisé, caractérisé par une expression préférentielle des chaînes variables TRAV24 et TRBV2, la présence de motifs conservés au sein des régions CDR3 de ces deux chaînes, et une prévalence élevée de clonotypes publics. Le clonotype public le plus représenté est capable de conférer une réponse spécifique à Gag293 très sensible et polyfonctionnelle aux cellules T CD4+ primaires, ainsi qu'une reconnaissance de multiples allèles HLA-DR. Nous cherchons à présent à déterminer si les candidats vaccins peuvent induire des réponses T CD4+ similaires à celles observées chez des patients HIC. Pour cela, nous avons analysé les réponses spécifiques de Gag293 chez des volontaires sains ayant reçu le candidat vaccin ADN ADVAX administré par soit par électroporation (EP) ou soit par injection intramusculaire (IM). La comparaison des réponses spécifiques de Gag293 dans des lignées T CD4+ primaires par test ELISpot IFNγ a montré que la sensibilité à l'antigène chez les volontaires vaccinés était proche de celle des patients contrôleurs et significativement plus élevée que celle mesurée chez les patients traités (p<0.05). Par contre, les volontaires du groupe IM n'ont pas montré de réponse détectable contre Gag293. L'analyse du répertoire TCR des cellules spécifiques de Gag293 chez les volontaires vaccinés a révélé une amplification préférentielle de la chaîne variable TRBV2, chaîne qui prédomine également chez les patients HIC. Cependant, l'utilisation des chaînes variables TRAV est apparue plus diverse chez les sujets vaccinés, avec une amplification préférentielle des chaînes TRAV29 et TRAV24. L'analyse des clonotypes TCR amplifiés chez les vaccinés a mis en évidence un recoupement inattendu avec le répertoire des HIC, avec une fréquence élevée de motifs partagés (30%). Les clonotypes publiques TRAV24 et TRBV2 les plus fréquemment retrouvés dans le répertoire des Contrôleurs ont également été induits par la vaccination ADN EP, indiquant le potentiel pour une réponse de forte avidité. Par contre, le clonotype TRAV29 le plus exprimé, chez les vaccinés a induit des réponses de plus faible avidité, même si ces réponses étaient restreintes par de multiples allèles HLA-DR. L'ensemble de ces résultats suggère que les clonotypes TRAV24, plutôt que les clonotypes TRAV29, sont responsables de la réponse de forte avidité contre Gag293. En conclusion, un candidat vaccin ADN administré par EP possède la capacité d'induire des clonotypes TCR associés au contrôle du VIH. Cette étude suggère que les clonotypes TCR publics pourraient servir de nouveaux biomarqueurs permettant d'évaluer la qualité des réponses vaccinales induites contre le VIH / HIV controllers are rare patients who spontaneously control HIV replication to levels below 50 copies viral RNA/mL in the absence of antiretroviral therapy. Converging evidence indicate that HIV controllers mount a cellular antiviral response that is highly efficient at detecting and eliminating infected cells. In particular, we previously reported that HIV controllers from the ANRS CO21 CODEX cohort harbour a pool of memory CD4+ T cells able to respond to minimal amounts of viral antigens, due to the expression of TCRs with high avidity for immunodominant Gag epitopes. In a first study we show that the high functional avidity observed in Controller CD4+ T cells could be explained by the expression of high avidity T cell receptors (TCRs) which efficiently bound Gag peptide-loaded MHC class II tetramers. Repertoire analysis of CD4+ T cells specific for the immunodominant Gag293 epitope showed a highly skewed repertoire in HIV controllers, with a predominance of the TRAV24 and TRBV2 variable gene families, the presence of conserved motifs in both CDR3 regions, and a high prevalence of public clonotypes. When transferred to primary CD4+ T cells, the most frequent public TCR clonotypes could confer properties associated with an efficient T cell response, such as polyfunctional cytokine secretion, high affinity antigen recognition, and broad HLA-DR crossrestriction. To determine whether candidate HIV vaccines could induce the high sensitivity responses seen in controllers, we analysed Gag293-specific responses in healthy volunteers who received the ADVAX DNA vaccine administered by electroporation (EP) or intramuscular injection (IM). Comparison of Gag293-specific responses in primary CD4+ T cell lines via IFNγ ELISpot revealed that the median antigen sensitivity in vaccinees was close to that observed for controllers but significantly higher than that in treated patients (p <0.05). In contrast, volunteers from the IM group did not show a detectable Gag293-specific response. TCR repertoire analysis of Gag293-specific CD4+ T cells from vaccinees revealed a preferential amplification of TCRβ family chain TRBV2, which also predominates in controllers. However, TRAV family gene usage appeared more diverse in vaccinees compared to controllers, with the preferential amplification of TRAV29 and TRAV24. Sequence analysis revealed an unexpected degree of overlap between the specific repertoires of vaccinees and HIV controllers, as evidenced by a high frequency of shared TCR motifs (>30%). The most frequent TRAV24 and TRBV2 public clonotypes found in the controller TCR repertoire were also induced by EP DNA vaccination, suggesting the potential for high avidity responses. In contrast, an abundant TRAV29 clonotype from vaccinees induced Gag293-specific responses of lower avidity, even though these responses were broadly HLA-DR cross-restricted. Thus, our results suggest that TRAV24 rather than TRAV29 clonotypes may be responsible for the high avidity Gag293-response. In conclusion, DNA vaccination by electroporation has the potential to inducer TCR clonotypes associated with HIV control. Monitoring the amplification of public TCR clonotypes could provide a novel approach to evaluate the quality of HIV vaccine responses

"October 2002" / Bibliography: p. 219-228. / xi, 229 p. : ill. (some col.), plates (col.) ; 30 cm. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Proposes a robust indirect position sensing technique implemented in real time using a DSP based motor controller. This is based on a rotor position estimation using the measured phase voltages and currents of the motor. An incremental algorithm is used to estimate the flux linkage increments, which eliminates the integration involved in the flux linkage estimation. This feature provides a wider operating speed range and robustness. / Thesis (Ph.D.)--University of Adelaide, Dept. of Electrical and Electronic Engineering, 2002

No description available.

In this thesis, a Reinforcement Learning (RL) method called Sarsa is used to dynamically tune a PI-controller for a Continuous Stirred Tank Heater (CSTH) experimental setup. The proposed approach uses an approximate model to train the RL agent in the simulation environment before implementation on the real plant. This is done in order to help the RL agent initially start from a reasonably stable policy. Learning without any information about the dynamics of the process is not practically feasible due to the great amount of data (time) that the RL algorithm requires and safety issues. The process in this thesis is modeled with a First Order Plus Time Delay (FOPTD) transfer function, because almost all of the chemical processes can be sufficiently represented by this class of transfer functions. The presence of a delay term in this type of transfer functions makes them inherently more complicated models for RL methods. RL methods should be combined with generalization techniques to handle the continuous state space. Here, parameterized quadratic function approximation compounded with k-nearest neighborhood function approximation is used for the regions close and far from the origin, respectively. Applying each of these generalization methods separately has some disadvantages, hence their combination is used to overcome these flaws. The proposed RL-based PI-controller is initially trained in the simulation environment. Thereafter, the policy of the simulation-based RL agent is used as the starting policy of the RL agent during implementation on the experimental setup. As a result of the existing plant-model mismatch, the performance of the RL-based PI-controller using this primary policy is not as good as the simulationresults; however, training on the real plant results in a significant improvement in this performance. On the other hand, the IMC-tuned PI-controllers, which are the most commonly used feedback controllers are also compared and they also degrade because of the inevitable plant-model mismatch. To improve the performance of these IMC-tuned PI-controllers, re-tuning of these controllers based on a more precise model of the process is necessary. The experimental tests are carried out for the cases of set-point tracking and disturbance rejection. In both cases, the successful adaptability of the RL-based PI-controller is clearly evident. Finally, in the case of a disturbance entering the process, the performance of the proposed model-free self-tuning PI-controller degrades more, when compared to the existing IMC controllers. However, the adaptability of the RL-based PI- controller provides a good solution to this problem. After being trained to handle disturbances in the process, an improved control policy is obtained, which is able to successfully return the output to the set-point. / Process Control

The aim of this thesis is to gather insights on the underlying social factors behind the current state of controller roles in a multinational enterprise. Additionally, the thesis explores the possibility of managing perceptions and expectations of controller roles, in accordance to corporate business objectives. In order to do this, theoretical concepts of Role Theory, Boundary Spanning and Identity Control were utilized. This paper has a qualitative research design and is based on an embedded single-case study. Due to the exploratory nature of the research, the study is based on the method of „systematic combining‟. The objectives of the company were to establish both local and regional controllers as business partners. Findings indicate that the regional controller could focus on business advisory tasks to a large extent due to a formally appointed role, whereas the local controllers were restricted by their role and expectations to focus more on scorekeeping and managerial duties. The term „business partner‟ was spread throughout the organization. By ascribing this label to controllers, other expectations on the role were induced, which also infers a higher hierarchical positioning. Future research could go more in depth on the process of internalizing values.

A new and improved stabilizer was developed for the variable speed generation (VSG) system. The VSG system exhibits periodic oscillations which sometimes leads to a loss of synchronism. After careful study, a simple but effective strategy to stabilize the system was implemented with real time digital feedback control. The VSG system consists of an engine, which is the prime mover, driving a doubly fed machine (DFM), which is the generator. The stator of the DFM is directly connected to the grid while the rotor is connected to the grid through a power electronic converter. The converter used in this study is a series resonance converter (SRC), but the proposed method may also be applied to other kinds of converters. The stabilizer senses the RPM of the engine, the feedback signal, and controls the rotor current amplitude and frequency of the doubly fed machine. Control was implemented using the 80C196KB microcontroller. The software consists of a mix of "C" and assembly language. Speed being an important factor in the implementation, care was taken to minimize the control loop times. The important features of the hardware and software developed for the stabilizer are: (1) 12 MHz controller board (2) Real time digital band pass filter (3) Instantaneous rotor speed measurement (4) Interrupt driven measurement and control loops (5) User defined setup parameters (6) IBM PC based real time serial communication The performance of the VSG system was studied with and without the stabilizer. A significant improvement in the stability of the system was noticed over the entire region of operation. / Graduation date: 1991