Tree series transformations computed by bottom-up and top-down tree series transducers are called bottom-up and top-down tree series transformations, respectively. (Functional) compositions of such transformations are investigated. It turns out that the class of bottomup tree series transformations over a commutative and complete semiring is closed under left-composition with linear bottom-up tree series transformations and right-composition with boolean deterministic bottom-up tree series transformations. Moreover, it is shown that the class of top-down tree series transformations over a commutative and complete semiring is closed under right-composition with linear, nondeleting top-down tree series transformations. Finally, the composition of a boolean, deterministic, total top-down tree series transformation with a linear top-down tree series transformation is shown to be a top-down tree series transformation.

The field of speech recognition during the last decade has left the research stage and found its way into the public market, and today, speech recognition software is ubiquitous around us. An automatic speech recognizer understands human speech and represents it as text. Most of the current speech recognition software employs variants of deep neural networks. Before the deep learning era, the hybrid of hidden Markov model and Gaussian mixture model (HMM-GMM) was a popular statistical model to solve speech recognition. In this thesis, automatic speech recognition using HMM-GMM was trained on Somali data which consisted of voice recording and its transcription. HMM-GMM is a hybrid system in which the framework is composed of an acoustic model and a language model. The acoustic model represents the time-variant aspect of the speech signal, and the language model determines how probable is the observed sequence of words. This thesis begins with background about speech recognition. Literature survey covers some of the work that has been done in this field. This thesis evaluates how different language models and discounting methods affect the performance of speech recognition systems. Also, log scores were calculated for the top 5 predicted sentences and confidence measures of pre-dicted sentences. The model was trained on 4.5 hrs of voiced data and its corresponding transcription. It was evaluated on 3 mins of testing data. The performance of the trained model on the test set was good, given that the data was devoid of any background noise and lack of variability. The performance of the model is measured using word error rate(WER) and sentence error rate (SER). The performance of the implemented model is also compared with the results of other research work. This thesis also discusses why log and confidence score of the sentence might not be a good way to measure the performance of the resulting model. It also discusses the shortcoming of the HMM-GMM model, how the existing model can be improved, and different alternatives to solve the problem.

Tato disertační práce studuje synchronní formální systémy založené na gramatikách a převodnících a zkoumá jak jejich teoretické vlastnosti, tak i perspektivy praktických aplikací. Práce představuje nové koncepty a definice vycházející ze známých principů řízeného přepisování a synchronizace. Navrhuje alternativní způsob synchronizace bezkontextových gramatik, založený na propojení pravidel. Tento princip rozšiřuje také na řízené gramatiky, konkrétně gramatiky s rozptýleným kontextem a maticové gramatiky. Dále je představen na podobném principu založený nový druh převodníku, tzv. pravidlově omezený převodník. Jedná se o systém složený z konečného automatu a bezkontextové gramatiky. Práce prezentuje nové teoretické výsledky ohledně generativní a přijímajicí síly. Poslední část práce zkoumá možnosti lingvisticky orientovaných aplikací se zameřením na překlad přirozeného jazyka. Diskutuje a srovnává hlavní výhody nových modelů s využitím vybraných případových studií z českého, anglického a japonského jazyka pro ilustraci.

[pt] Ondas ultrassônicas podem ser usadas para ensaios não destrutivos. Elas são geradas e adquiridas por transdutores. Transdutores acústicos eletromagnéticos (EMATs) possuem algumas vantagens em relação a transdutores tradicionais piezoelétricos, principalmente a capacidade de gerar ondas ultrassônicas sem necessidade de contato físico com o meio em teste. No entanto, sua principal desvantagem é a menor eficiência, que resulta em uma relação sinal-ruído mais baixa. Técnicas de casamento de impedância podem ser utilizadas para combater isso. A rede de casamento de impedância do tipo circuito L é comumente utilizada para garantir a transferência máxima de potência da eletrônica de excitação para o EMAT. Existem diversas variáveis que podem afetar a impedância de um EMAT além do próprio transdutor, como as propriedades e distância do material, a temperatura e a frequência de excitação. Portanto, para garantir a transferência máxima de potência, o circuito de casamento de impedância precisa ser reconfigurado e ter seus valores ajustados sempre que um dos fatores mencionados acima sofrer alteração. O processo de ajuste manual desta rede é trabalhoso e demorado, portanto, sua automação pode trazer grandes benefícios para o uso de transdutores EMAT. Esta dissertação propõe um circuito simplificado, com um único elemento, para casamento de impedância de EMATs. Os valores teóricos ideais para circuitos mono-elemento foram obtidos. Simulações confirmaram sua viabilidade em aumentar a eficiência do EMAT. Circuitos de casamento manual e automáticos foram projetados e construídas. Configurações experimentais foram elaboradas e postas em prática. Experimentos com dois transdutores EMATs distintos foram conduzidos utilizando várias frequências. O sistema automático foi capaz de determinar a melhor configuração para o circuito mono-elemento de casamento de impedância e forneceu um ganho de até 5,6 dB, similar à solução manual. A configuração automática foi mais de duas vezes mais rápida do que a manual. / [en] Ultrasonic waves can be used for nondestructive testing. They are generated and acquired by transducers. Electromagnetic acoustic transducers (EMATs) have some advantages over traditional piezoelectric transducers. Mainly, the ability to generate ultrasonic waves without requiring physical contact with the medium under test. Nevertheless, they present a main drawback of less efficiency, which leads to a lower signal-to-noise ratio. To overcome this, impedance matching techniques can be used. The L-network impedance matching network is often used in order to ensure maximum power transfer to the EMAT from the excitation electronics. There is a wide range of variables that can affect an EMAT s impedance besides the transducer itself, namely, the properties and distance to the specimen material, the temperature, and the excitation frequency. Therefore, to ensure optimal power transfer, the Lnetwork s configuration needs to be tuned whenever one of the factors that affect impedance changes. The process of manually adjusting the impedance matching network is a laborious and time-consuming task, therefore, its automation can be of great benefit to the use of EMAT transducers. In this work, a simplified one-element automatic matching network is proposed. The theoretical optimal values for the one-element matching networks are derived. Simulations confirmed their effectiveness to increase EMAT efficiency. Manual and automatic networks were designed and built. Experiments were performed with two dierent EMATs at several frequencies. The automatic system was able to determine the best conguration for the one-element matching network and provided up to 5.6 dB gain, similar to the manual solution. The automatic setup was more than two-fold faster than the manual one.

Tire pressure monitoring systems (TPMSs) predict over- and underinflated tires, and warn the driver in critical situations. Today, battery powered TPMSs suffer from limited energy. New sensor features such as friction determination or aquaplaning detection require even more energy and would significantly decrease the TPMS lifetime. Harvesting electrical energy inside the tire of a vehicle has been considered as a promising alternative to overcome the limited lifetime of a battery. However, it is a real challenge to design a system, that generates electrical energy at low velocities while being robust at 200 km/h where radial accelerations up to 20000 m/s2 occur. This work focusses on developing different electromechanical energy transducers that meet the high requirements of the automotive sector. Different approaches are addressed on how the change of acceleration and strain within the tire can be used to provide mechanical energy to the energy harvester. The energy harvester converts the mechanical energy into electrical energy. In this thesis, piezoelectric and electromagnetic transducers are discussed in depth, modelled as electromechanical networks. Since the transducers provide energy in the form of an AC voltage, but sensors require a DC voltage, various common interface circuits are compared, using LTspice and applying method of the stochastic signal analysis. Furthermore, a buck-boost converter concept for the electromagnetic energy harvester is optimized and improved. Experiments on a tire test rig validate the theoretically determined output and confirm that well designed energy harvesters in the tire can generate much more energy than required by an TPMS not only at high velocities but also at velocities as low as 20 km/h.

The aim of this thesis is to investigate the potential of replacing the material 1.4418 (R87), currently used in ABB's PFCL201 load cells, with the material 1.402 (R03). Both materials possess desirable properties, including high strength, toughness and magnetoelastic characteristics, making them suitable for force transducer applications in strip tension systems. However, the scarcity and high cost of R87 necessitate exploring the feasibility of utilizing the more affordable and easily obtainable R03. The research methodology involved a combination of mechanical and thermal simulations, as well as the evaluation of prototype measurements made from R03. Mechanical simulations were conducted to identify a new load cell geometry that would facilitate the use of R03, while thermal simulations focused on comparing the thermal properties of R03 with real-life measurements. Furthermore, prototypes made from R03 were tested to investigate the transducer characteristics of the material. Lastly, a cost analysis was performed, comparing the manufacturing costs associated with R87 and R03. The study yielded promising results. R03 improves the manufacturing process and significantly reduces the costs related to it. A new load cell geometry was identified, which could be manufactured using existing resources at the factory. Thermal simulations demonstrated improvements in the thermal properties when employing R03. However, measurements of the PFCL201-20 kN load cell made from R03, indicated that to maintain the same accuracy class and commutation angle as the current R87 load cells, the nominal load would need to be adjusted to 12 kN instead of 20 kN. Nonetheless, with the identified geometric modifications, R03 load cells could still be utilized as 20 kN load cells. Alternatively, by changing either the accuracy class or commutation angle. This research provides valuable insights into the possibility of replacing the expensive and scarce R87 material with the more cost-effective and readily available R03 in ABB's PFCL201 load cells. The findings offer a foundation for future studies and potential business decisions regarding material selection and load cell design optimization.

No description available.

Contactless ultrasonic power transfer (UPT), using piezoelectric transducers, is based on transferring energy using acoustic waves, in which the waves are generated by an acoustic source or transmitter and then transferred through an acoustic medium such as water or human tissue to a sensor or receiver. The receiver then converts the mechanical strain induced by the incident acoustic waves to electricity and delivers to an electrical load, in which the electrical power output of the system can be determined. The execution and efficiency of this technology can be significantly enhanced through patterning, focusing, and localization of the transmitted acoustic energy in space to simultaneously power pre-determined distributed sensors or devices. A passive 3D-printed acoustic hologram plate alongside a single transducer can generate arbitrary and pre-designed ultrasound fields in a particular distance from the hologram mounted on the transmitter, i.e., a target plane. This dissertation presents the use of these simple, cost-effective, and high-fidelity acoustic holograms in UPT systems to selectively enhance and pattern the electrical power output from the receivers. Different holograms are numerically designed to create single and multi-focal pressure patterns in a target plane where an array of receivers are placed. The incident sound wave from a transmitter, after passing through the hologram, is manipulated, hence, the output field is the desired pressure field, which excites the receivers located at the pre-determined focal points more significantly. Furthermore, multi-functional holograms are designed to generate multiple images at different target planes and driving frequencies, called, respectively, multi-image-plane and multi-frequency patterning holograms. The multiple desired pressure distributions are encoded on the single hologram plate and each is reconstructed by changing the axial distance and by switching the frequency. Several proof-of-concept experiments are performed to verify the functionality of the computationally designed holograms, which are fabricated using modern 3D-printers, i.e., the desired wavefronts are encoded in the hologram plates' thickness profile, being input to the 3D-printer. The experiments include measurement of output pressure fields in water using needle hydrophones and acquisition of receivers' voltage output in UPT systems. Another technique investigated in this dissertation is the implementation of acoustic impedance matching layers deposited on the front leading surface of the transmitter and receiver transducers. Current UPT systems suffer from significant acoustic losses through the transmission line from a piezoelectric transmitter to an acoustic medium and then to a piezoelectric receiver. This is due to the unfavorable acoustic impedance mismatch between the transducers and the medium, which causes a narrow transducer bandwidth and a considerable reflection of the acoustic pressure waves at the boundary layers. Using matching layers enhance the acoustic power transmission into the medium and then reinforce the input as an excitation into the receiver. Experiments are performed to identify the input acoustic pressure from a cylindrical transmitter to a receiver disk operating in the 33-mode of piezoelectricity. Significant enhancements are obtained in terms of the receiver's electrical power output when implementing a two-layer matching structure. A design platform is also developed that can facilitate the construction of high-fidelity acoustically matched transducers, that is, the material layers' selection and determination of their thicknesses. Furthermore, this dissertation presents a numerical analysis for the dynamical motions of a high-intensity focused ultrasound (HIFU)-excited microbubble or stable acoustic cavitation, which includes the effects of acoustic nonlinearity, diffraction, and absorption of the medium, and entails the problem of several biomedical ultrasound applications. Finally, the design and use of acoustic holograms in microfluidic channels are addressed which opens the door of acoustic patterning in particle and cell sorting for medical ultrasound systems. / Doctor of Philosophy / This dissertation presents several techniques to enhance the wireless transfer of ultrasonic energy in which the sound wave is generated by an acoustic source or transmitter, transferred through an acoustic medium such as water or human tissue to a sensor or receiver. The receiver transducer then converts the vibrational energy into electricity and delivers to an electrical load in which the electrical power output from the system can be determined. The first enhancement technique presented in this dissertation is using a pre-designed and simple structured plate called an acoustic hologram in conjunction with a transmitter transducer to arbitrarily pattern and shape ultrasound fields at a particular distance from the hologram mounted on the transmitter. The desired wavefront such as single or multi-focal pressure fields or an arbitrary image such as a VT image pattern can simply be encoded in the thickness profile of this hologram plate by removing some of the hologram material based on the desired shape. When the sound wave from the transmitter passes this structured plate, it is locally delayed in proportion to the hologram thickness due to the different speed of sound in the hologram material compared to water. In this dissertation, various hologram types are designed numerically to implement in the ultrasonic power transfer (UPT) systems for powering receivers located at the predetermined focal points more significantly and finally, their functionality and performances are verified in several experiments. Current UPT systems suffer from significant acoustic losses through the transmission from a transmitter to an acoustic medium and then to a receiver due to the different acoustic impedance (defined as the product of density and sound speed) between the medium and transducers material, which reflects most of the incident pressure wave at the boundary layers. The second enhancement technology addressed in this dissertation is using intermediate materials, called acoustic impedance matching layers, bonded to the front side of the transmitter and receiver face to alleviate the acoustic impedance mismatch. Experiments are performed to identify the input acoustic pressure from a transmitter to a receiver. Using a two-layer matching structure, significant enhancements are observed in terms of the receiver's electrical power output. A design platform is also developed that can facilitate the construction of high-fidelity acoustically matched transducers, that is, the material layers' selection and determination of their thicknesses. Furthermore, this dissertation presents a numerical analysis for the dynamical motions of a microbubble exposed to a high-intensity focused ultrasound (HIFU) field, which entails the problem of several biomedical ultrasound applications such as microbubble-mediated ultrasound therapy or targeted drug delivery. Finally, an enhancement technique involving the design and use of acoustic holograms in microfluidic channels is addressed which opens the door of acoustic patterning in particle and cell sorting for medical ultrasound systems.

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

No / Keratinocyte traction forces play a crucial role in wound healing. The aim of this study was to develop a novel cell traction force (CTF) transducer system based on cholesteryl ester liquid crystals (LC). Keratinocytes cultured on LC induced linear and isolated deformation lines in the LC surface. As suggested by the fluorescence staining, the deformation lines appeared to correlate with the forces generated by the contraction of circumferential actin filaments which were transmitted to the LC surface via the focal adhesions. Due to the linear viscoelastic behavior of the LC, Hooke's equation was used to quantify the CTFs by associating Young's modulus of LC to the cell induced stresses and biaxial strain in forming the LC deformation. Young's modulus of the LC was profiled by using spherical indentation and determined at approximately 87.1+/-17.2kPa. A new technique involving cytochalasin-B treatment was used to disrupt the intracellular force generating actin fibers, and consequently the biaxial strain in the LC induced by the cells was determined. Due to the improved sensitivity and spatial resolution ( approximately 1mum) of the LC based CTF transducer, a wide range of CTFs was determined (10-120nN). These were found to be linearly proportional to the length of the deformations. The linear relationship of CTF-deformations was then applied in a bespoke CTF mapping software to estimate CTFs and to map CTF fields. The generated CTF map highlighted distinct distributions and different magnitude of CTFs were revealed for polarized and non-polarized keratinocytes.