The wear comfort of a prosthesis is of great importance for amputee patients. The wear comfort can be affected by changes in the interface between the residual limb and prosthesis socket, which can be caused by time-dependent volume fluctuations of the tissue, leading to unwanted local pressure marks. The basis to ensure time-independent wear comfort of a prosthesis is to identify these changes. Common techniques for identifying these variations have a negative impact on the sensitive interface between the residual limb and prosthesis. The following paper contains a proof of concept for the detection of local pressure marks without affecting the described interface using structural dynamics measurements, exemplarily shown at a prosthetic socket for transfemoral amputees in a test bench scenario. The dynamical behaviour of the investigated system is analysed in the form of frequency response functions acquired for different pressure locations and preloads using an impact hammer for excitation and a triaxial acceleration sensor. The frequency response functions show major changes for the various boundary conditions with respect to their frequency-dependent compositions. The results demonstrate how the utilised method enables the identification of changes in local pressure marks regarding the variation of position and magnitude.

This project aimed to classified artificiality gener-ated,fake, and authentic,true, telecommunication signals, basedupon their frequency response, using methods from deep learn-ing. Another goal was to accomplish this with the least amountof dimension of data possible. The datasets used contained of anequal amount of measured, provided by Ericsson, and generated,by a WINNER II implementation in Matlab, frequency responses.It was determined that a normalized version of the absolute valueof the complex frequency response was enough information for afeedforward network to do a sufficient classification. To improvethe accuracy of the network we did a hyperparameter search,which allowed us to reach an accuracy of 90 percent on our testdataset. The results show that it is possible for neural networksto differentiate between true and fake telecommunication signalsbased on their frequency response, even if it is hard for a humanto tell the difference. / Målet med det här projektet var att klassificera artificiellt genererade signaler, falska, och riktiga, sanna, telekommunikation signaler med hjälp av signalernas frekvens- svar med djup inlärningsmetoder, deep learning. Ett annat mål med projektet var att klassificera signalerna med minsta möjliga antalet dimensioner av datan. Datasetet som användes bestod av till hälften av uppmät data som Ericsson har tillhandahållit, och till hälften av generad data ifrån en WINNER II modell implementerad i Matlab. En slutsats som kunde dras är att en normaliserad version av beloppet av det komplexa frekvenssvaret innehöll tillräckligt med information för att träna ett feedforward nätverk till att uppnå en hög klassificeringssäkerhet. För att vidare öka tillförlitligheten av nätverket gjordes en hyperparametersökning, detta ökade tillförligheten till 90 procent för testdataseten. Resultaten visar att det är möjligt för neurala nätverk att skilja mellan sanna och falska telekommunikations- signaler baserat på deras frekvenssvar, även om det är svårt för människor att skilja signalerna åt. / Kandidatexjobb i elektroteknik 2020, KTH, Stockholm

Duration tuning in the mammalian inferior colliculus (IC) is created by the interaction of excitatory and inhibitory synaptic inputs. We used extracellular recording and paired-tone stimulation to measure the strength and time-course of the contralateral inhibition and offset-evoked excitation underlying duration-tuned neurons (DTNs) in the IC of the awake bat. The onset time of a short, best duration (BD), excitatory probe tone was varied relative to the onset of a longer-duration, non-excitatory (NE) suppressor tone. Spikes evoked by the roving BD tone were suppressed or facilitated when the stationary NE tone was varied in frequency or amplitude. When the NE tone frequency was presented away from the cell’s best excitatory frequency (BEF) or at lower SPLs, the onset of inhibition was relatively constant whereas the offset and duration of inhibition decreased. Excitatory and inhibitory frequency response areas were measured and best inhibitory frequencies matched best excitatory frequencies; however, inhibitory bandwidths were broader than excitatory bandwidths. Excitatory rate-level and inhibitory suppression-level functions were also measured and the dynamic ranges and inflection points were similar, which is hypothesized to play a role in the level tolerance of responses measured from DTNs. We compared the latency of offset-locked facilitation to the onset or offset of inhibition as a function of frequency and amplitude; we found that the facilitation was more related to the onset of inhibition. Moreover, facilitation typically preceded the offset of inhibition – suggesting that it is a separate excitatory input to DTNs and not a rebound from inhibition. We conclude that DTNs receive inputs that generate and preserve temporal selectivity. / Dissertation / Doctor of Philosophy (PhD)

[pt] Este trabalho tem como objetivo construir um classificador baseado em regras de inferência fuzzy, as quais são extraídas a partir de máquinas de vetor suporte (SVMs) e ajustadas com o auxílio de um algoritmo genético. O classificador construído visa a diagnosticar transformadores de potência. As SVMs são sistemas de aprendizado baseados na teoria do aprendizado estatístico e apresentam boa habilidade de generalização em conjuntos de dados reais. SVMs, da mesma forma que redes neurais (RN), geram um modelo caixa preta, isto é, um modelo que não explica o processo pelo qual sua saída é obtida. Entretanto, para alguns problemas, o conhecimento sobre como a classificação foi obtida é tão importante quanto a classificação propriamente dita. Alguns métodos propostos para reduzir ou eliminar essa limitação já foram desenvolvidos, embora sejam restritos à extração de regras simbólicas, isto é, contêm funções ou intervalos nos antecedentes das regras. No entanto, a interpretabilidade de regras simbólicas ainda é reduzida. De forma a aumentar a interpretabilidade das regras, o modelo FREx_SVM foi desenvolvido. Neste modelo as regras fuzzy são extraídas a partir de SVMs treinadas. O modelo FREx_SVM pode ser aplicado a problemas de classificação com n classes, não sendo restrito a classificações binárias. Entretanto, apesar do bom desempenho do modelo FREx_SVM na extração de regras linguísticas, o desempenho de classificação do sistema de inferência fuzzy obtido é ainda inferior ao da SVM, uma vez que as partições (conjuntos fuzzy) das variáveis de entrada são definidas a priori, permanecendo fixas durante o processo de aprendizado das regras. O objetivo desta dissertação é, portanto, estender o modelo FREx_SVM, de forma a permitir o ajuste automático das funções de pertinência das variáveis de entrada através de algoritmos genéticos. Para avaliar o desempenho do modelo estendido, foram realizados estudos de caso em dois bancos de dados: Iris, como uma base benchmark, e a análise de resposta em frequência. A análise de resposta em frequência é uma técnica não invasiva e não destrutiva, pois preserva as características dos equipamentos. No entanto, o diagnóstico é feito de modo visual comparativo e requer o auxílio de um especialista. Muitas vezes, este diagnóstico é subjetivo e inconclusivo. O ajuste automático das funções de pertinência correspondentes aos conjuntos fuzzy associados às variáveis de entrada reduziu o erro de classificação em até 13,38 por cento em relação à configuração sem este ajuste. Em alguns casos, o desempenho da configuração com ajuste das funções de pertinência supera até mesmo aquele obtido pela própria SVM. / [en] This work aims to develop a classifier model based on fuzzy inference rules, which are extracted from support vector machines (SVMs) and optimized by a genetic algorithm. The classifier built aims to diagnose power transformers. The SVMs are learning systems based on statistical learning theory and have provided good generalization performance in real data sets. SVMs, as artificial neural networks (NN), generate a black box model, that is, a model that does not explain the process by which its output is obtained. However, for some applications, the knowledge about how the classification was obtained is as important as the classification itself. Some proposed methods to reduce or eliminate this limitation have already been developed, although they are restricted to the extraction of symbolic rules, i.e. contain functions or ranges in the rules´ antecedents. Nevertheless, the interpretability of symbolic rules is still reduced. In order to increase the interpretability of the rules, the FREx_SVM model was developed. In this model the fuzzy rules are extracted from trained SVMs. The FREx_SVM model can be applied to classification problems with n classes, not being restricted to binary classifications. However, despite the good performance of the FREx_SVM model in extracting linguistic rules, the classification performance of fuzzy classification system obtained is still lower than the SVM, since the partitions (fuzzy sets) of the input variables are predefined at the beginning of the process, and are fixed during the rule extraction process. The goal of this dissertation is, therefore, to extend the FREx_SVM model, so as to enable the automatic adjustment of the membership functions of the input variables through genetic algorithms. To assess the performance of the extended model, case studies were carried out in two databases: iris benchmark and frequency response analysis. The frequency response analysis is a noninvasive and non-destructive technique, because it preserves the characteristics of the equipment. However, the diagnosis is carried out by visual comparison and requires the assistance of an expert. Often, this diagnosis is subjective and inconclusive. The automatic adjustment of the membership functions associated with input variables reduced the error up to 13.38 per cent when compared to the configuration without this optimization. In some cases, the classification performance with membership functions optimization exceeds even those obtained by SVM.

<p dir="ltr">Acoustic guitar soundboard material selection is based on selective evaluations that have been developed over centuries. These traditional practices are not conducive to the guitar industry we experience today because the supply of traditionally acceptable soundboard wood has decreased greatly. The purpose of this research was to develop an objective wood selection metric to determine the sound quality of an acoustic guitar’s soundboard. The metric would replace the subjective evaluations traditionally used to select materials for acoustic guitar soundboards.</p><p dir="ltr">The acoustic properties of sound radiation coefficient, material’s speed of sound, resonance and damping and the material properties of longitudinal and radial elastic modulus, density, and specific modulus were used in an attempt to construct a material selection metric. These variables were selected because the literature review revealed that these were the most critical variables in determining sound quality. The gaps in the literature were testing and analyzing samples that represented the true dimensions of an acoustic guitar soundboard blank and forming the metric. The literature revealed that the previous experimental studies did not have the appropriate test sample dimensions that correspond to the test samples evaluated by the subjective methods.</p><p dir="ltr">The methodology was carried out by using the objective testing counterparts to the subjective assessments found in the literature review. Instrumented hammer tap testing collected data to determine damping and resonance frequencies. A three-point static bending test collected data for longitudinal and radial elastic modulus. Mass and dimension measurements were recorded to calculate density. Calculations were done to compute the acoustic properties and specific modulus of the test samples. These variables were put into a table and underwent statistical analysis in the form of predictor correlation and logistical regression. The experimental variables were modeled against the subjective evaluation of an expert on the usability of the test samples.</p><p dir="ltr">Statistical analysis proved that the dataset did not show any significant separation between “good” or “bad” test samples or a significant correlation between the usability of the test sample and the variables in the dataset. The methodology did not produce an objective material selection metric to determine the sound quality of an acoustic guitar’s soundboard. Future research should include a wider range of measured frequencies and the collection of time domain data.</p>

This thesis aims to develop a new type of highly stretchable strain sensor to measure large deformation of a specimen subjected to dynamic loading. The sensor was based on the piezo-resistive response of carbon nanotube(CNT)/polydimethysiloxane (PDMS) composites thin films, some nickel particles were added into the sensor composite to improve the sensor performance. The piezo-resistive response of CNT composite gives high frequency response in strain measurement, while the ultra-soft PDMS matrix provides high flexibility and ductility for large strain measuring large strain (up to 26%) with an excellent linearity and a fast frequency response under quasi-static test, the delay time for high strain rate test is just 30 μs. This stretchable strain sensor is also able to exhibit much higher sensitivities, with a gauge factor of as high as 80, than conventional foil strain gauges.

Approved for public release, distribution is unlimited / In this thesis, the ocean was modeled as a waveguide with an ideal pressure - release surface, and an ideal rigid bottom. The ocean waveguide was then treated as a linear, time - invariant, space - variant (TISV) filter or communication channel. The filter is time - invariant because no motion was modeled and because the properties of the ocean were assumed to be constant. The filter is space - variant because of the presence of the two boundaries, that is, the ocean surface and ocean bottom. This thesis investigates the ocean as a linear TISV filter by evaluating 1) the complex frequency response, 2) the impulse response, and 3) the transfer function of the ocean with respect to depth. It is shown that the TISV impulse response of the ocean contains information that can be used to help localize a target in range and whether the target is above or below the receiver. Computer simulation results were obtained by evaluating the three filter functions for several different test cases. / Ensign, United States Navy

O ensaio de resposta em frequência em geradores síncronos vem ganhando espaço nas últimas décadas, porém o alto custo dos equipamentos empregados para realização do ensaio ainda é um empecilho, tanto para fabricantes como para consumidores finais. Este trabalho tem por objetivo complementar trabalhos anteriores, através do uso de redes neurais artificiais para identificação de padrões em ensaios realizados com equipamentos de baixo custo e baixa resolução. Através das redes neurais artificiais utilizadas é possível estimar novos pontos de ensaio sem que seja necessário um novo ensaio ou até mesmo um ensaio com equipamentos mais caros. Através de uma combinação de algoritmos de tratamento de dados, é possível a aproximação de modelos ensaiados com modelos teóricos e através desses resultados obter parâmetros elétricos dos geradores síncronos. / The frequency response test on synchronous generators has been increasing in recent decades, but the high cost of equipment used for conducting the test is still a stumbling block for both manufacturers and end-consumers. This dissertation aims to complement previous work, through the use of artificial neural networks to identify patterns in tests conducted with low-cost and low-resolution equipment. Through the artificial neural networks used it is possible to estimate a new set of test points without retesting is necessary or even a more expensive assay equipments. Through a combination of data processing algorithms, the approach tested models with theoretical models is possible and through these results to obtain electrical parameters of synchronous generators.

Orientador: Samuel da Silva / Resumo: Muitas aplicações da engenharia envolvem estruturas essencialmente não-lineares onde várias técnicas têm sido recentemente estudadas e investigadas por muitos pesquisadores. Dentre as várias abordagems, as que usam séries de Volterra têm apresentado propriedades úteis para fornecer um melhor entendimento para identificação e análise. Neste contexto, a presente tese propõem novas contribuições em como usar as séries de Volterra para caracterização, identificação e análise dinâmica de sistemas não-lineares usando sinais de entrada e saída e sinais somente de saída. Inicialmente, apresenta-se uma metodologia para análise de sistemas mecânicos não-lineares através das funções de resposta em frequência de alta-ordem (HOFRFs) e o conceito de HOFRFs estendidas com dados apenas de saída é introduzido e descrito em detalhes. Após isso, uma abordagem para identificação de sistemas não-lineares com base nas séries de Volterra através da expansão na base ortonormal de Kautz é proposta. Essa técnica permite identificar os seus núcleos mais facilmente e permite separar as contribuições dos termos lineares e não-lineares usando somente sinais de saída. Além disso, uma metodologia para análise modal de sistemas fracamente não-lineares sujeito a excitações com vários níveis de amplitude é também apresentada. A contribuição desse novo método reside no fato de que as HOFRFs são simplesmente estimadas como função das FRFs lineares. Basicamente, essa metodologia estende o conceito ... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Most recent engineering applications involve structures essentially nonlinear where several techniques have been recently studied and investigated by many researchers. Among them, the methods based on Volterra series expansion have presented powerful properties to provide a better understanding for identification and analysis. In this context, the present thesis proposes new contributions in how to use Volterra series for characterization, identification and dynamical analysis of nonlinear systems based on input and output signals and output-only signals. Initially, a methodology for analysis of nonlinear mechanical systems through higher-order frequency response functions (HOFRFs) is presented and the concept of extended HOFRFs based on output-only is introduced and described in detail. Afterwards, an approach for identification of nonlinear systems based on Volterra series through the expansion onto orthonormal Kautz basis is proposed. This technique allows to identify the Volterra kernels easily and enable to split the contribution of the linear and nonlinear terms using input-output as well as output-only signals. Furthermore, a methodology for modal analysis of weakly nonlinear systems under multilevel excitation is also proposed. The contribution of this new approach lies in the fact that HOFRFs are simply computed as functions of the linear FRFs. Basically, it extends the conventional experimental modal analysis methods in order to characterize and treat no... (Complete abstract click electronic access below) / Doutor

Industrial robot manipulators are general-purpose machines used for industrial automation in order to increase productivity, flexibility, and quality. Other reasons for using industrial robots are cost saving, and elimination of heavy and health-hazardous work. Robot motion control is a key competence for robot manufacturers, and the current development is focused on increasing the robot performance, reducing the robot cost, improving safety, and introducing new functionalities. Therefore, there is a need to continuously improve the models and control methods in order to fulfil all conflicting requirements, such as increased performance for a robot with lower weight, and thus lower mechanical stiffness and more complicated vibration modes. One reason for this development of the robot mechanical structure is of course cost-reduction, but other benefits are lower power consumption, improved dexterity, safety issues, and low environmental impact. This thesis deals with three different aspects of modeling and control of flexible, i.e., elastic, manipulators. For an accurate description of a modern industrial manipulator, the traditional flexible joint model, described in literature, is not sufficient. An improved model where the elasticity is described by a number of localized multidimensional spring-damper pairs is therefore proposed. This model is called the extended flexible joint model. This work describes identification, feedforward control, and feedback control, using this model. The proposed identification method is a frequency-domain non-linear gray-box method, which is evaluated by the identification of a modern six-axes robot manipulator. The identified model gives a good description of the global behavior of this robot. The inverse dynamics control problem is discussed, and a solution methodology is proposed. This methodology is based on a differential algebraic equation (DAE) formulation of the problem. Feedforward control of a two-axes manipulator is then studied using this DAE approach. Finally, a benchmark problem for robust feedback control of a single-axis extended flexible joint model is presented and some proposed solutions are analyzed.