Le corps humain (adulte) est composé de 206 os (“Anatomy and Physiology | Simple Book Production” n.d.) qui sont des tissus denses et composent la majeure partie du squelette humain. Le squelette, étant hautement vascularisé, est l’endroit le plus communément affecté par le cancer métastatique (Coleman 1997). L’apparition de ces métastases osseuses fragilise l’os et peut provoquer des fractures pathologiques. Toutefois la prédiction de telles fractures est difficile et loin d’être automatique. Une possibilité pour créer un outil de diagnostic plus performant serait les simulations éléments finis (FEA en anglais pour « Finite Elements Analysis »). Des études ont montré que la FEA spécifique au patient était capable de surpasser l’expertise des cliniciens dans le cas d’étude ex vivo avec défauts osseux induits mécaniquement (dont Derikx et al. 2012). Les recherches portant sur le cancer osseux sont toutefois dur à mettre en place, les échantillons étant rare. De manière à contourner la difficulté de trouver des échantillons humains rarement disponibles, la souris a été utilisé comme modèle squelettique dans plusieurs cas, incluant la tenue mécanique d’os atteint de métastases ex vivo (Mann et al. 2008). Ainsi, de manière à pouvoir étudier l’implication du tissu métastatique dans la résistance globale de l’os sur échantillons réels, nous avons utilisé ce modèle animal pour créer des échantillons tumoraux.Notre but était double : premièrement, quantifier l’apport de la prise en compte des propriétés mécaniques de la métastase dans la résistance globale de l’os. Deuxièmement, statuer sur le fait qu’un modèle plus simple que celui proposé dans la littérature (reposant sur des propriétés purement élastiques plutôt qu’élasto-plastiques (Eggermont et al. 2018) pouvait permettre d’améliorer la prédiction de fractures pathologiques.Tout d’abord, les résultats obtenus avec nos modèles hétérogènes (ne prenant pas en compte la tumeur) ont montré une bonne consistance avec la littérature, la corrélation entre tous les modèles hétérogènes (n=43 pattes) quant à la fracture simulée et expérimentale étant du même ordre de grandeur que celles d’une étude analogue menée sur vertèbres de souris (Nyman et al. 2015). Ensuite, le modèle prenant en compte les propriétés des tumeurs n’as pas permis d’améliorer la prédiction de fracture, au contraire, la moyenne des différences de ces modèles étant de 30±21% (n=11 pattes tumorales) contre 12±9% (n=43 pattes). De plus le modèle spécifique (prenant en compte le module des tumeurs) étant plus difficile à obtenir que le modèle hétérogène (ne nécessitant pas de segmentation entre os et tumeur), le premier ne semble pas être judicieux dans la prédiction de fracture d’os long présentant des lyses osseuses. Enfin, un critère de détection reposant sur la différence entre valeurs de forces ultimes globale et locale a permis de détecter la majorité des instabilités mécaniques constatées dans cette étude (sensibilité de 85% et spécificité de 100%). Un autre critère, basé sur le ratio entre poids des individus et la force ultime locale prédite via FEA a permis de correctement diagnostiquer l’ensemble des cas (100% de sensibilité et de spécificité). Ce résultat pourrait s’avérer être d’une grande aide quant à la prise de décision d’intervention chirurgicale dans le cas d’os long atteints de métastases osseuses. Bien sûr, avant cela la route à parcourir reste longue, ce résultat devant d’abord être confirmé cliniquement (possiblement en ayant recours à l’étude d’un cohorte rétrospective, comme cela a déjà pu être fait dans d’autres études (Eggermont et al. 2018). Cette étude vient d’être initiée dans le cas du projet MEKANOS (étude multicentrique en France) porté par le Professeur Cyrille Confavreux (rhumatologue) / The human body (adult) is composed of 206 bones ("Anatomy and Physiology | Simple Book Production" n.d.) that are dense tissues and make up the bulk of the human skeleton. The skeleton, being highly vascularized, is the most commonly affected site for metastatic cancer (Coleman 1997). The development of these bone metastases weakens the bone and can cause pathological fractures. However, the prediction of such fractures is difficult and far from automatic. One possibility for creating a more powerful diagnostic tool would be finite element simulations (FEA). Studies have shown that patient-specific FEA is able to surpass the expertise of clinicians in the case of ex vivo studies with mechanically induced bone defects (including Derikx et al., 2012). Research on bone cancer, however, is hard to put in place as samples are rare. In order to overcome the difficulty of finding human samples that are rarely available, the mouse has been used as a skeletal model in several cases, including the mechanical resistance of bones with ex vivo metastases (Mann et al., 2008). Thus, in order to study the involvement of metastatic tissue in the overall bone resistance of real samples, we used this animal model to create tumor samples. Our goal was twofold: first, to quantify the contribution of taking into account the mechanical properties of metastasis in the overall resistance of the bone. Secondly, to see if a simpler model than that proposed in the literature (based on purely elastic rather than elastoplastic properties (Eggermont et al., 2018) could improve the prediction of pathological fractures. First, the results obtained with our heterogeneous models (not taking tumor into account) showed a good consistency with the literature, the correlation between all the heterogeneous models (n = 43 legs) regarding the agreement of simulated and experimental fracture were of the same order of magnitude as a similar study conducted on mouse vertebrae (Nyman et al., 2015). Then, the model taking into account the properties of the tumors did not make it possible to improve the fracture prediction. The average of the differences of models taking tumor into account being of 30 ± 21% (n = 11 tumor limbs) against 12 ± 9% (n = 43 limbs). In addition, the specific model (taking into account the modulus of the tumors) being more difficult to obtain than the heterogeneous model (not requiring segmentation between bone and tumor), the first does not seem to be a wise choice in the prediction of long bone fracture presenting bone lysis. Finally, a detection criterion based on the difference between global and local ultimate force values made it possible to detect the majority of the mechanical instabilities observed in this study (sensitivity of 85% and specificity of 100%). Another criterion, based on the ratio between individual weights and the local ultimate force predicted via FEA, made it possible to correctly diagnose all cases (100% sensitivity and specificity). This result could prove to be of great help in making surgical decision making in the case of long bone with bone metastases. Of course, before that, the road ahead is long, this result having to be clinically confirmed first (possibly through the study of a retrospective cohort, as has already been done in other studies (Eggermont et al., 2018). This study has just been initiated in the case of the project MEKANOS (multicenter study in France) led by Professor Cyrille Confavreux (rheumatologist)

No description available.

This study is concerned with developing a model to identify small-medium U.K. companies at risk of financial failure up to five years in advance. The importance of small companies in an economy, the impact of their failures, and the lack of failure research with respect to . this population, provided justification for this study. The research was undertaken in two stages. The first stage included a detailed description and discussion of the nature and role of small business in the UK economy, heir relevance, problems and Government involvement in this sector, together with literature review and assessment of past research relevant to this study. The second stage was involved with construction of the models using multiple discriminant analysis, applied to published accountancy data for two groups of failed and nonfailed companies. The later stage was performed in three parts : (1) evaluating five discriminant models for each of five years prior to failure; (2) testing the performance of each of the .five models over time on data not used . in their construction; (3) testing the discriminant models on a validation sample. The purpose was to establish the "best" discriminant model. "Best" was determined according to classification ability of the model and interpretation of variables. Finally a model comprising seven financial ratios measuring four aspects of a company's financial profile, such as profitability, gearing, capital turnover and liquidity was chosen. The model has shown to be a valid tool for predicting companies' health up to five years in advance. / Ministry of Higher Education and Scientific Research of the Iraqi Government.

In Nonlinear Finite Element Analysis (FEA) applied to structures, displacements at which the tangent stiffness matrix KT becomes singular are called critical points, and correspond to instabilities such as buckling or elastoplastic softening (e.g., necking). Prior to the introduction of Arc Length Methods (ALMs), critical points posed severe computational challenges, which was unfortunate since behavior at instabilities is of great interest as a precursor to structural failure. The original ALM was shown to be capable in some circumstances of continued computation at critical points, but limited success and unattractive features of the formulation were noted and addressed in extensive subsequent research. The widely used Crisfield Cylindrical and Spherical ALMs may be viewed as representing the 'state-of-the-art'. The more recent Stiff Arc Length method, which is attractive on fundamental grounds, was introduced in 2004, but without implementation, benchmarking or performance assessment. The present thesis addresses (a) implementation and (b) performance comparisons for the Crisfield and Stiff methods, using simple benchmarks formulated to incorporate elastoplastic softening. It is seen that, in contrast to the Crisfield methods, the Stiff ALM consistently continues accurate computation at, near and beyond critical points.

Submitted by Daniella Sodre (daniella.sodre@ufpe.br) on 2015-04-10T16:15:19Z No. of bitstreams: 2 dscidl.pdf: 3672005 bytes, checksum: 16e2ea719e96351a648acbff70be2fb0 (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) / Made available in DSpace on 2015-04-10T16:15:19Z (GMT). No. of bitstreams: 2 dscidl.pdf: 3672005 bytes, checksum: 16e2ea719e96351a648acbff70be2fb0 (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Previous issue date: 2013-02-27 / CNPq / Nesse trabalho são desenvolvidos modelos de quantificação de métricas de risco e confiabilidade para sistemas em diferentes etapas do ciclo de vida. Para sistemas na fase de projeto, um Algoritmo Genético Multiobjetivo (MOGA) é combinado à Simulação Discreta de Eventos (DES) a fim de prover configurações não-dominadas com relação à disponibilidade e ao custo. O MOGA + DES proposto incorpora Processos de Renovação Generalizados para modelagem de reparos imperfeitos e também indica o número ótimo de equipes de manutenção. Para a fase operacional é proposto um hibridismo entre MOGA e Inspeção Baseada no Risco para elaboração de planos de inspeção não-dominados em termos de risco e custo que atendem às normas locais. Regressão via Support Vector Machines (SVR) é aplicada nos casos em que a métrica relacionada à confiabilidade (variável resposta) de um sistema operacional é função de variáveis ambientais e operacionais com expressão analítica desconhecida. Otimização via Nuvens de Partículas é combinada à SVR para a seleção simultânea das variáveis explicativas mais relevantes e dos valores dos hiperparâmetros que aparecem no problema de treinamento de SVR. Com o objetivo de avaliar a incerteza relacionada à variável resposta, métodos bootstrap são combinados à SVR para a obtenção de intervalos de confiança e de previsão. São realizados experimentos numéricos e são apresentados exemplos de aplicação no contexto da indústria do petróleo. Os resultados obtidos indicam que os modelos propostos fornecem informações importantes para o planejamento de custos e para a implementação de ações apropriadas a fim de evitar eventos indesejados. --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------This work develops models for quantifying risk and reliability-related metrics of systems in different phases of their life cycle. For systems in the design phase, a Multi-Objective Genetic Algorithm (MOGA) is coupled with Discrete Event Simulation (DES) to provide non-dominated configurations with respect to availability and cost. The proposed MOGA + DES incorporates a Generalized Renewal Process to account for imperfect repairs and it also indicates the optimal number of maintenance teams. For the operational phase, a hybridism between MOGA and Risk-Based Inspection is proposed for the elaboration of non-dominated inspection plans in terms of risk and cost that comply with local regulations. Regression via Support Vector Machines (SVR) is applied when the reliability-related metric (response variable) of an operational system is function of a number of environmental and operational variables with unknown analytical relationship. A Particle Swarm Optimization is combined to SVR for the selection of the most relevant variables along with the tuning of the SVR hyperparameters that appear in its training problem. In order to assess the uncertainty related to the response variable, bootstrap methods are coupled with SVR to construct confidence and prediction intervals. Numerical experiments and application examples in the context of oil industry are provided. The obtained results indicate that the proposed frameworks give valuable information for budget planning and for the implementation of proper actions to avoid undesired events.

The ability to accurately predict an impending hard disk failure is important for reliable storage system design. The facility provided by most hard drive manufacturers, called S.M.A.R.T. (self-monitoring, analysis and reporting technology), has been shown by current research to have poor predictive value. The problem of finding alternatives to S.M.A.R.T. for predicting disk failure is an area of active research. In this work, we present a rule discovery methodology, and show that it is possible to construct decision support systems that can detect such failures using information recorded from live disks. It is desired that any such prediction methodology should have high accuracy and must have ease of interpretability. Black box models can deliver highly accurate solutions but do not provide an understanding of events which explains the decision given by it. To this end we explore rule based classifiers for predicting hard disk failures from various disk events. We show that it is possible to learn easy to understand rules from disk events. Our evaluation shows that our system can be tuned either to have a high failure detection rate (i.e., classify a bad disk as bad) or to have a low false alarm rate (i.e., not classify a good disk as bad). We also propose a modification of MLRules algorithm for classification of data with imbalanced class distributions. The existing algorithm, assuming relatively balanced class distributions and equal misclassfication costs, performs poorly in classification of such datasets. The performance can be considerably improved by introducing cost- sensitive learning to the existing framework.

No description available.

Diploma thesis focus on creation of neural network’s internal structure with goal of creation Artificial Neural Network capable of machine state monitoring and predicting its remaining usefull life. Main goal is creation of algorithm’s and library for design and learning of Artificial Neural Network, and deeper understanding of the problematics in the process, then by utilising existing libraries. Selected method was forward-propagation network with multi-layered perceptron architecture, and backpropagation learning. Achieved results was, that the network was able to determine parts state from vibration measurement and on its basis predict remaining usefull life.

Die Arbeit beschreibt verschiedene technische Ansätze zur Bestimmung der Alterung der Chip-Substrat-Verbindung. Eine der Schlüsseltechnologien ist hierbei die Bestimmung der virtuellen Sperrschichttemperatur. Es werden in der Arbeit verschiedene Möglichkeiten zur Bestimmung der virtuellen Chiptemperatur von IGBTs und der Alterung der Chip-Substrat-Verbindung vorgestellt und mit ihren Vor- und Nachteilen in der Umsetzbarkeit und in der Anwendbarkeit im Umrichter diskutiert. Besondere Betrachtung findet dabei unter anderem die technische Umsetzung einer Messmethode, die auf einer kurzzeitigen Belastung im aktiven Bereich mit anschließender Bestimmung des Abkühlverhaltens basiert. Über einen Vergleich mit dem ursprünglichen Abkühlverhalten ist es mit den vorgestellten Verfahren gut möglich, die Chipalterung zu detektieren. Weiterhin wird ein Verfahren vorgestellt, das die Bestimmung der virtuellen Chiptemperatur im Umrichter über eine Ermittlung der Millerplateauhöhe im Abschaltmoment des IGBTs ermöglicht.:1 Einleitung 2 Alterung von Leistungshalbleitern 3 Stand der Technik der Chiptemperaturbestimmung in der Umrichterschaltung 4 Untersuchungen temperaturabhängiger elektrischer Bauelementparameter 5 Wichtige Verfahren zur Bestimmung der Chiptemperatur in der Umrichterschaltung 6 Untersuchung eines Verfahrens zur Bestimmung der Chiptemperatur in der Umrichterschaltung mittels Millerplateauhöhe 7 Technische Umsetzbarkeit der gezeigten Messverfahren 8 Zusammenfassung und Ausblick A Anhang / This work describes different technical approaches to determine the aging of the chip-substrate interconnection. One of the key technologies here is the determination of the virtual junction temperature. Various possibilities for determining the virtual chip temperature of IGBTs and the aging of the chip-substrate interconnection are presented in the work and discussed with their advantages and disadvantages in terms of feasibility and applicability in the converter. Special consideration is given to the technical implementation of a measurement method based on a short-term load in the active area with subsequent determination of the cooling behavior. By comparing this with the original cooling behavior, it is possible to detect chip aging with the methods presented. Furthermore, a method is presented that enables the determination of the virtual chip temperature in the inverter via a determination of the Miller plateau height at the switch-off moment of the IGBT.:1 Einleitung 2 Alterung von Leistungshalbleitern 3 Stand der Technik der Chiptemperaturbestimmung in der Umrichterschaltung 4 Untersuchungen temperaturabhängiger elektrischer Bauelementparameter 5 Wichtige Verfahren zur Bestimmung der Chiptemperatur in der Umrichterschaltung 6 Untersuchung eines Verfahrens zur Bestimmung der Chiptemperatur in der Umrichterschaltung mittels Millerplateauhöhe 7 Technische Umsetzbarkeit der gezeigten Messverfahren 8 Zusammenfassung und Ausblick A Anhang

This thesis proposes a deep learning anomaly detection pipeline to detect possible anomalies during the operation of a fleet of batteries and presents its development and evaluation. The pipeline employs sensors that connect to each battery in the fleet to remotely collect real-time measurements of their operating characteristics, such as voltage, current, and temperature. The deep learning based time-series anomaly detection model was developed using Variational Autoencoder (VAE) architecture that utilizes either Long Short-Term Memory (LSTM) or, its cousin, Gated Recurrent Unit (GRU) as the encoder and the decoder networks (LSTMVAE and GRUVAE). Both variants were evaluated against three well-known conventional anomaly detection algorithms Isolation Nearest Neighbour (iNNE), Isolation Forest (iForest), and kth Nearest Neighbour (k-NN) algorithms. All five models were trained using two variations in the training dataset (full-year dataset and partial recent dataset), producing a total of 10 different model variants. The models were trained using the unsupervised method and the results were evaluated using a test dataset consisting of a few known anomaly days in the past operation of the customer’s battery fleet. The results demonstrated that k-NN and GRUVAE performed close to each other, outperforming the rest of the models with a notable margin. LSTMVAE and iForest performed moderately, while the iNNE and iForest variant trained with the full dataset, performed the worst in the evaluation. A general observation also reveals that limiting the training dataset to only a recent period produces better results nearly consistently across all models. / Detta examensarbete föreslår en pipeline för djupinlärning av avvikelser för att upptäcka möjliga anomalier under driften av en flotta av batterier och presenterar dess utveckling och utvärdering. Rörledningen använder sensorer som ansluter till varje batteri i flottan för att på distans samla in realtidsmätningar av deras driftsegenskaper, såsom spänning, ström och temperatur. Den djupinlärningsbaserade tidsserieanomalidetekteringsmodellen utvecklades med VAE-arkitektur som använder antingen LSTM eller, dess kusin, GRU som kodare och avkodarnätverk (LSTMVAE och GRU) VAE). Båda varianterna utvärderades mot tre välkända konventionella anomalidetekteringsalgoritmer -iNNE, iForest och k-NN algoritmer. Alla fem modellerna tränades med hjälp av två varianter av träningsdatauppsättningen (helårsdatauppsättning och delvis färsk datauppsättning), vilket producerade totalt 10 olika modellvarianter. Modellerna tränades med den oövervakade metoden och resultaten utvärderades med hjälp av en testdatauppsättning bestående av några kända anomalidagar under tidigare drift av kundens batteriflotta. Resultaten visade att k-NN och GRUVAE presterade nära varandra och överträffade resten av modellerna med en anmärkningsvärd marginal. LSTMVAE och iForest presterade måttligt, medan varianten iNNE och iForest tränade med hela datasetet presterade sämst i utvärderingen. En allmän observation avslöjar också att en begränsning av träningsdatauppsättningen till endast en ny period ger bättre resultat nästan konsekvent över alla modeller.