La mise en place d’une politique de maintenance prévisionnelle est un défi majeur dans l’industrie qui tente de réduire le plus possible les frais relatifs à la maintenance. En effet, les systèmes sont de plus en plus complexes et demandent un suivi de plus en plus poussé afin de rester opérationnels et sécurisés. Une maintenance prévisionnelle nécessite d’une part d’évaluer l’état de dégradation des composants du système, et d’autre part de pronostiquer l’apparition future d’une panne. Plus précisément, il s’agit d’estimer le temps restant avant l’arrivée d’une défaillance, aussi appelé Remaining Useful Life ou RUL en anglais. L’estimation d’une RUL constitue un réel enjeu car la pertinence et l’efficacité des actions de maintenance dépendent de la justesse et de la précision des résultats obtenus. Il existe de nombreuses méthodes permettant de réaliser un pronostic de durée de vie résiduelle, chacune avec ses spécificités, ses avantages et ses inconvénients. Les travaux présentés dans ce manuscrit s’intéressent à une méthodologie générale pour estimer la RUL d’un composant. L’objectif est de proposer une méthode applicable à un grand nombre de cas et de situations différentes sans nécessiter de modification majeure. De plus, nous cherchons aussi à traiter plusieurs types d’incertitudes afin d’améliorer la justesse des résultats de pronostic. Au final, la méthodologie développée constitue une aide à la décision pour la planification des opérations de maintenance. La RUL estimée permet de décider de l’instant optimal des interventions nécessaires, et le traitement des incertitudes apporte un niveau de confiance supplémentaire dans les valeurs obtenues. / Predictive maintenance strategies can help reduce the ever-growing maintenance costs, but their implementation represents a major challenge. Indeed, it requires to evaluate the health state of the component of the system and to prognosticate the occurrence of a future failure. This second step consists in estimating the remaining useful life (RUL) of the components, in Other words, the time they will continue functioning properly. This RUL estimation holds a high stake because the precision and accuracy of the results will influence the relevance and effectiveness of the maintenance operations. Many methods have been developed to prognosticate the remaining useful life of a component. Each one has its own particularities, advantages and drawbacks. The present work proposes a general methodology for component RUL estimation. The objective i to develop a method that can be applied to many different cases and situations and does not require big modifications. Moreover, several types of uncertainties are being dealt With in order to improve the accuracy of the prognostic. The proposed methodology can help in the maintenance decision making process. Indeed, it is possible to select the optimal moment for a required intervention thanks to the estimated RUL. Furthermore, dealing With the uncertainties provides additional confidence into the prognostic results.

Thesis (MScEng)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Physical Asset Management (PAM) is becoming a greater concern for companies in industry today. The widely accepted British Standards Institutes’ specification for optimized management of physical assets and infrastructure is PAS55. According to PAS55, PAM is the “systematic and co-ordinated activities and practices through which an organization optimally manages its physical assets, and their associated performance, risks and expenditures over their life cycle for the purpose of achieving its organizational strategic plan”. One key performance area of PAM is Asset Care Plans (ACP). These plans are maintenance strategies which improve or ensure acceptable asset reliability and performance during its useful life. Maintenance strategies such as Condition Based Maintenance (CBM) acts upon Condition Monitoring (CM) data, disregarding the previous failure histories of an asset. Other maintenance strategies, such as Usage Based Maintenance (UBM), is based on previous failure histories, and does not consider CM data. Regression models make use of both CM data and previous failure histories to develop a model which represents the underlying failure behaviour of the asset under study. These models can be of high value in ACP development due to the fact that Residual Useful Life (RUL) can be estimated and/or the long term life cycle cost can be optimized. The objective of this thesis was to model historical failure data and CM data well enough so that RUL or optimized preventive maintenance instant estimations can be made. These estimates were used in decision models to develop maintenance schedules, i.e. ACPs. Several regression models were evaluated to determine the most suitable model to achieve the objectives of this thesis. The model found to be most suitable for this research project was the Proportional Hazards Model (PHM). A comprehensive investigation on the PHM was undertaken focussing on the mathematics and the practical implementation thereof. Data obtained from the South African mining industry was modelled with the Weibull PHM. It was found that the developed model produced estimates which were accurate representations of reality. These findings provide an exciting basis for the development of futureWeibull PHMs that could result in huge maintenance cost savings and reduced failure occurrences. / AFRIKAANSE OPSOMMING: Fisiese Bate Bestuur (FBB) is besig om ’n groter bekommernis vir maatskappye in die bedryf te word. Die Britse Standaarde Instituut se spesifikasie vir optimale bestuur van fisiese bates en infrastruktuur is PAS55. Volgens PAS55 is FBB die “sistematiese en gekoördineerde aktiwiteite en praktyke wat deur ’n organisasie optimaal sy fisiese bates, hul verwante prestasie, risiko’s en uitgawes vir die doel van die bereiking van sy organisatoriese strategiese plan beheer oor hul volle lewensiklus te bestuur”. Een Sleutel Fokus Area (SFA) van FBB is Bate Versorgings Plan (BVP) ontwikkeling. Hierdie is onderhouds strategieë wat bate betroubaarheid verbeter of verseker tydens die volle bruikbare lewe van die bate. Een onderhoud strategie is Toestands Gebasseeerde Onderhoud (TGO) wat besluite baseer op Toestand Monitering (TM) informasie maar neem nie die vorige falingsgeskiedenis van die bate in ag nie. Ander onderhoud strategieë soos Gebruik Gebasseerde Onderhoud (GGO) is gebaseer op historiese falingsdata maar neem nie TM inligting in ag nie. Regressiemodelle neem beide TM data en historiese falings geskiedenis data in ag ten einde die onderliggende falings gedrag van die gegewe bate te verteenwoordig. Hierdie modelle kan baie nuttig wees vir BVP ontwikkeling te danke aan die feit dat Bruikbare Oorblywende Lewe (BOL) geskat kan word en/of die langtermyn lewenssilus koste geoptimeer kan word. Die doelwit van hierdie tesis was om historiese falingsdata en TT data goed genoeg te modelleer sodat BOL of optimale langtermyn lewensiklus kostes bepaal kan word om opgeneem te word in BVP ontwikkeling. Hierdie bepalings word dan gebruik in besluitnemings modelle wat gebruik kan word om onderhoud skedules op te stel, d.w.s. om ’n BVP te ontwikkel. Verskeie regressiemodelle was geëvalueer om die regte model te vind waarmee die doel van hierdie tesis te bereik kan word. Die mees geskikte model vir die navorsingsprojek was die Proporsionele Gevaarkoers Model (PGM). ’n Omvattende ondersoek oor die PGM is onderneem wat fokus op die wiskunde en die praktiese implementering daarvan. Data is van die Suid-Afrikaanse mynbedryf verkry en is gemodelleer met behulp van die Weibull PGM. Dit was bevind dat die ontwikkelde model resultate geproduseer het wat ’n akkurate verteenwoordinging van realiteit is. Hierdie bevindinge bied ’n opwindende basis vir die ontwikkeling van toekomstige Weibull Proporsionele Gevaarkoers Modelle wat kan lei tot groot onderhoudskoste besparings en minder onverwagte falings.

The carbonation in concrete is caused by the penetration of environmental carbonic anhydride (CO2) in concrete by diffusion, being considered one of more important structures pathologies. It occurs naturally in concrete structures, from the surface and causes the alkalis neutralization reactions, depassivating the rebars and its corrosion possibility. In this study the concrete investigation with 14 years old composed by binary and ternary mixtures of pozzolans as silica fume (10%), fly ash (25%), rice husk ash (25%), fly ash and silica fume (15+10)% and fly ash with rice husk ash (10+15)%. The natural carbonation depths were measured in cylindrical specimens with 0,5, 1, 2, 4 and 14 years readings, exposed in lab internal environment, in normal conditions of temperature and CO2 concentrations. Natural carbonation coefficients in the five cited ages were calculated, in equality of water/binder (w/b) relationship (0.35, 0.45 and 0.55), and 50 and 60 MPa axial compressive strength, and the results were compared with the short term tests (accelerated). Microstructure analysis by means of remained calcium hydroxide, hydrated compounds by DRX, porosity by Hg intrusion porosimetry and visual analysis by MEV and EDS were accomplished. Among the pozzolans mixtures the silica fume (10%) presented lower carbonation for w/b 0.35 and 0.45. For w/b 0.55 the better performance occurred to the fly ash and rice husk ash ternary mixture. The relationships between accelerated and natural carbonation coefficients showed decrease of the natural coefficients related to the accelerated ones, in a relation that varied between 1,0 and 2,0 (mean) for the pozzolanic mixtures. In 50 and 60 MPa compressive strength equality, the carbonation depended of the pozzolan type and content, being influenced by the chemical and physical properties of each one in particular. The microstructure tests results confirmed the CH decrease in the carbonated layer and the CaCO3 increase, and the hydrated silicates and silicoaluminates depolymerization. According NBR 15575-5 was observed that it is possible to obtain concretes with until 25% of pozzolans with performance of the project useful life, in front of carbonation, of 60 years. / A penetração de anidrido carbônico (CO2) ambiental por difusão no concreto ocasiona a carbonatação, sendo considerada uma das patologias mais importantes das estruturas. Acontece naturalmente nas estruturas de concreto, a partir da superfície e ocasiona reações de neutralização dos álcalis, despassivando a armadura e a possibilidade de sua corrosão. Neste estudo apresenta-se a investigação de concretos com 14 anos de idade, compostas de misturas binárias e ternárias de pozolanas sendo sílica ativa(10%), cinza volante(25%), cinza de casca de arroz(25%), cinza volante e sílica ativa (15+10)% e cinza volante com cinza de casca de arroz (10+15)%. As profundidades de carbonatação natural foram medidas em corpos de prova cilíndricos, com leituras a 0,5, 1, 2, 4 e 14 anos, expostos em ambiente interno de laboratório, em condições normais de temperatura e concentração de CO2. Calculou-se os coeficientes de carbonatação natural nas 5 idades citadas, em igualdade de relação a/ag (0,35, 0,45 e 0,55) e de resistência à compressão axial de 50 e 60 MPa, e comparou-se com os resultados de curto prazo (acelerado). Realizou-se análise da microestrutura por meio do teor de hidróxido remanescente, dos compostos hidratados por difração de raios-X, porosidade por porosimetria por intrusão de mercúrio e análise visual por microscopia eletrônica de varredura com EDS. O concreto de referência apresentou menor coeficiente de carbonatação nas três relações a/ag estudadas. Entre as misturas com pozolanas a sílica ativa (10%) apresentou menor carbonatação para a/ag 0,35 e 0,45. Para a/ag 0,55 o melhor desempenho aconteceu para a mistura ternaria de cinza volante e cinza de casca de arroz. As relações entre os ensaios acelerado e naturais mostraram decréscimo dos coeficientes de carbonatação natural em relação aos acelerados, numa relação que variou, em média, entre 1,0 e 2,0 para as mistura pozolânicas. Em igualdade de resistência à compressão de 50 e 60 MPa a carbonatação dependeu do tipo e do teor de pozolana, sendo influenciada pelas propriedades químicas e físicas de cada uma em particular. Os resultados dos ensaios da microestrutura confirmaram o decréscimo do CH nas camadas carbonatadas e acréscimo de CaCO3 e a despolimerização dos silicatos e silicoaluminatos de cálcio hidratados. Segundo a NBR 15575-2 observou-se que é possível obter-se concretos com até 25% de pozolanas, com desempenho à vida útil de projeto, frente a carbonatação, de 60 anos.

Prognostics and health management (PHM) is a major tool enabling systems to evaluate their reliability in real-time operation. Despite ground-breaking advances in most engineering and scientific disciplines during the past decades, reliability engineering has not seen significant breakthroughs or noticeable advances. Therefore, self-awareness of the embedded system is also often required in the sense that the system should be able to assess its own health state and failure records, and those of its main components, and take action appropriately. This thesis presents a radically new prognostics approach to reliable system design that will revolutionise complex power electronic systems with robust prognostics capability enhanced Insulated Gate Bipolar Transistors (IGBT) in applications where reliability is significantly challenging and critical. The IGBT is considered as one of the components that is mainly damaged in converters and experiences a number of failure mechanisms, such as bond wire lift off, die attached solder crack, loose gate control voltage, etc. The resulting effects mentioned are complex. For instance, solder crack growth results in increasing the IGBT’s thermal junction which becomes a source of heat turns to wire bond lift off. As a result, the indication of this failure can be seen often in increasing on-state resistance relating to the voltage drop between on-state collector-emitter. On the other hand, hot carrier injection is increased due to electrical stress. Additionally, IGBTs are components that mainly work under high stress, temperature and power consumptions due to the higher range of load that these devices need to switch. This accelerates the degradation mechanism in the power switches in discrete fashion till reaches failure state which fail after several hundred cycles. To this end, exploiting failure mechanism knowledge of IGBTs and identifying failure parameter indication are background information of developing failure model and prognostics algorithm to calculate remaining useful life (RUL) along with ±10% confidence bounds. A number of various prognostics models have been developed for forecasting time to failure of IGBTs and the performance of the presented estimation models has been evaluated based on two different evaluation metrics. The results show significant improvement in health monitoring capability for power switches. Furthermore, the reliability of the power switch was calculated and conducted to fully describe health state of the converter and reconfigure the control parameter using adaptive algorithm under degradation and load mission limitation. As a result, the life expectancy of devices has been increased. These all allow condition-monitoring facilities to minimise stress levels and predict future failure which greatly reduces the likelihood of power switch failures in the first place.

Submitted by Erika Demachki (erikademachki@gmail.com) on 2014-10-23T18:20:00Z No. of bitstreams: 2 Dissertação - Murillo Batista Ferreira - 2013.pdf: 2726162 bytes, checksum: b4ad5f67e5b5eff91f35b0296db5a36e (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Approved for entry into archive by Jaqueline Silva (jtas29@gmail.com) on 2014-10-23T18:46:59Z (GMT) No. of bitstreams: 2 Dissertação - Murillo Batista Ferreira - 2013.pdf: 2726162 bytes, checksum: b4ad5f67e5b5eff91f35b0296db5a36e (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) / Made available in DSpace on 2014-10-23T18:46:59Z (GMT). No. of bitstreams: 2 Dissertação - Murillo Batista Ferreira - 2013.pdf: 2726162 bytes, checksum: b4ad5f67e5b5eff91f35b0296db5a36e (MD5) license_rdf: 23148 bytes, checksum: 9da0b6dfac957114c6a7714714b86306 (MD5) Previous issue date: 2013-04-29 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Among lots of pathologies that attack the concrete structures stands out the corrosion of the reinforcements, that may be triggered by the phenomenon of carbonation of the concrete, which consists in a physico-chemical process between the carbon dioxide (CO2) from the atmosphere and the cement paste compounds, which has as main result the calcium carbonate precipitation (CaCO3) in a region of the concrete cover, with the formation of a layer that has less alkalinity than that not affected by this phenomenon. The carbonation moves from outside to inside the concrete through a carbonated front, that when it reaches the depth of the reinforcement causes destabilization of the protective passive layer, providing the onset of the corrosion. How many researches are conducted by accelerated carbonation tests, it is of paramount importance that the evaluation process by comparing it with data obtained from samples that have suffered natural carbonation. Facing this reality, this research is conducted based on samples that are exposed in the natural environment (external housed) for a period of ten years, with the general objective of evaluating the results (coefficients and carbonation depths) of natural carbonation process of concrete with different mineral additions, water/binder ratios and curing conditions (moist and dry) over this period. For this realization, samples of concrete produced in 2002, produced in Castro's dissertation (2003). The variables included in the experimental program are: concretes with five types of mineral additions (silica fume (10%), rice husk ash (10%), metakaolin (10%), fly ash (25%), blast furnace slag (65%), these amounts being in partial replacement of cement mass) and the reference concrete, without mineral addition; three water/binder relationships (0.40 , 0.55 and 0.70) and two curing conditions (moist and dry). Measurements were made of natural carbonation through of pH indicator based on phenolphthalein, elaborating a database along time, through measurements with periods of exposure of 0.25 years (91 days), 0.50 years (182 days), 0.82 years (301 days), 3 years, 7 years, 9 years and 10 years, allowing comparative performance evaluation of various concrete containing mineral additions, taking as a base carbonated depths and natural carbonation coefficients data. The results showed that there are significant differences of behavior between different concretes studied and that in the overall analysis of the carbonation natural coefficients, the reference concrete showed the best alone performance, showing that the alkaline reserve is still the predominant effect in terms of against carbonation, but the microstructure refinement provided by additions (physical effect) approaches such chemical aspect, as detected, for example, in concrete with metakaolin. In the correlation between the accelerated carbonation coefficients and the natural carbonation ones, was verified a relation of direct proportionality between the coefficients found for the two procedures, with the coefficient signifying accelerated approximately 7.4 times the natural ratio. The useful life predictions showed periods of time until the carbonation front reaches the cover thickness ranging from 4 years up to 241 years, for the different concretes and situations analyzed. / Dentre as diversas patologias que atacam as estruturas de concreto destaca-se a corrosão das armaduras, que pode ser desencadeada pelo fenômeno da carbonatação do concreto, que consiste em um processo físico-químico entre o gás carbônico (CO2) presente na atmosfera e os compostos da pasta de cimento, tendo como resultado principal a precipitação do carbonato de cálcio (CaCO3) em uma região do cobrimento, com a constituição de uma camada que passa a ter uma alcalinidade significativamente menor do aquela não afetada por esse fenômeno. A carbonatação avança de fora para dentro do concreto, por meio de uma frente carbonatada, que quando atinge a profundidade das armaduras provoca desestabilização da camada passiva protetora, propiciando assim o início da corrosão. Como muitas pesquisas são conduzidas por ensaios de carbonatação acelerada, é de suma importância a avaliação desse processo confrontando-o com dados obtidos de amostras que sofreram cabonatação natural. Diante dessa realidade, esta pesquisa se conduz baseada em amostras que estão expostas no ambiente natural (abrigado externo) por um período de dez anos, com o objetivo geral de avaliar os resultados obtidos (profundidades e coeficientes de carbonatação) do processo de carbonatação natural de concretos com diferentes adições minerais, relações água/aglomerante e processos de cura (úmida e seca) ao longo desse período. Para sua realização, foram utilizados corpos de prova de concreto armado produzidos no ano de 2002, fruto da dissertação de Castro (2003). As variáveis contempladas no programa experimental são: concretos com cinco tipos de adições minerais (sílica ativa (10%), cinza de casca de arroz (10%), metacaulim (10%), cinza volante (25%) e escória alto forno (65%), sendo estes teores em substituição parcial da massa de cimento) e mais o concreto de referência, sem adição mineral; três relações água/aglomerante (0,40; 0,55 e 0,70) e dois procedimentos de cura (úmida e seca). Foram realizadas medidas de carbonatação natural por meio de indicador de pH à base de fenolftaleína, elaborando-se um banco de dados ao longo do tempo, por meio de medidas realizadas com os tempos de exposição de 0,25 anos (91 dias), 0,50 anos (182 dias), 0,82 anos (301 dias), 3 anos, 7 anos, 9 anos e 10 anos, permitindo assim uma avaliação comparativa de desempenho dos diversos concretos contendo adições minerais, tendo-se como base os valores de profundidade carbonatada e os coeficientes de carbonatação natural. Os resultados mostraram que existem diferenças significativas de comportamento entre os diversos concretos estudados e que na análise global dos coeficientes de carbonatação natural, o concreto de referência apresentou isoladamente o melhor desempenho, demonstrando que a reserva alcalina ainda é o efeito preponderante em termos de combate à carbonatação, mas que o refinamento da microestrutura proporcionado pelas adições (efeito físico) se aproxima desse aspecto químico, como se tem, por exemplo, nos concretos com metacaulim. Na correlação entre os coeficientes de carbonatação acelerado e natural, verificou-se uma relação de direta proporcionalidade entre os coeficientes encontrados para os dois procedimentos, com o coeficiente acelerado significando aproximadamente 7,4 vezes o coeficiente natural. Os prognósticos de vida útil mostraram períodos de tempo estimado até a frente de carbonatação atingir a espessura de cobrimento variando desde 4 anos até 241 anos, para os diversos concretos e situações analisadas.

Tato práce se zabývá vytvořením simulačního modelu dvojčinného pneumatického pístu s mechanickou sestavou, včetně modelů snímačů, s následujícím odhadem parametrů a aproximací chování demonstračního zařízení. Dalším cílem je prezentace různých přístupů prediktivní údržby na datové sadě měřené na demonstračním zařízení. Na měřený datový soubor se aplikovaly signal-based techniky bez použití simulačního modelu a model-based metody, které vyžadují použití simulačního modelu. Výsledkem této práce je ověření možnosti monitorování stavu zařízení pomocí nainstalovaných senzorů a vyhodnocení efektivity senzorů z hlediska přesnosti a finančních nákladů.

Assessment of machine health and prediction of future failures are critical for maintenance decisions. Many of the existing methods use unsupervised techniques to construct health indicators by measuring the disparity between the current state and either the healthy or the faulty states of the system. This approach can work well, but if the resulting health indicators are insufficient there is no easy way to steer the algorithm towards better ones. In this thesis a new method for health indicator construction is investigated that aims to solve this issue. It is based on measuring distance after transforming the sensor data into a new space using a feed-forward neural network. The feed-forward neural network is trained using a multi-objective optimization algorithm, NSGA-II, to optimize criteria that are desired in a health indicator. Thereafter the constructed health indicator is passed into a gated recurrent unit for remaining useful life prediction. The approach is compared to benchmarks on the NASA Turbofan Engine Degradation Simulation dataset and in regard to the size of the neural networks, the model performs relatively well, but does not outperform the results reported by a few of the more recent methods. The method is also investigated on a simulated dataset based on elevator weights with two independent failures. The method is able to construct a single health indicator with a desirable shape for both failures, although the latter estimates of time until failure are overestimated for the more rare failure type. On both datasets the health indicator construction method is compared with a baseline without transformation function and does in both cases outperform it in terms of the resulting remaining useful life prediction error using the gated recurrent unit. Overall, the method is shown to be flexible in generating health indicators with different characteristics and because of its properties it is adaptive to different remaining useful life prediction methods. / Estimering av maskinhälsa och prognos av framtida fel är kritiska steg för underhållsbeslut. Många av de befintliga metoderna använder icke-väglett (unsupervised) lärande för att konstruera hälsoindikatorer som beskriver maskinens tillstånd över tid. Detta sker genom att mäta olikheter mellan det nuvarande tillståndet och antingen de friska eller fallerande tillstånden i systemet. Det här tillvägagångssättet kan fungera väl, men om de resulterande hälsoindikatorerna är otillräckliga så finns det inget enkelt sätt att styra algoritmen mot bättre. I det här examensarbetet undersöks en ny metod för konstruktion av hälsoindikatorer som försöker lösa det här problemet. Den är baserad på avståndsmätning efter att ha transformerat indatat till ett nytt vektorrum genom ett feed-forward neuralt nätverk. Nätverket är tränat genom en multi-objektiv optimeringsalgoritm, NSGA-II, för att optimera kriterier som är önskvärda hos en hälsoindikator. Därefter används den konstruerade hälsoindikatorn som indata till en gated recurrent unit (ett neuralt nätverk som hanterar sekventiell data) för att förutspå återstående livslängd hos systemet i fråga. Metoden jämförs med andra metoder på ett dataset från NASA som simulerar degradering hos turbofan-motorer. Med avseende på storleken på de använda neurala nätverken så är resultatet relativt bra, men överträffar inte resultaten rapporterade från några av de senaste metoderna. Metoden testas även på ett simulerat dataset baserat på elevatorer som fraktar säd med två oberoende fel. Metoden lyckas skapa en hälsoindikator som har en önskvärd form för båda felen. Dock så överskattar den senare modellen, som använde hälsoindikatorn, återstående livslängd vid estimering av det mer ovanliga felet. På båda dataseten jämförs metoden för hälsoindikatorkonstruktion med en basmetod utan transformering, d.v.s. avståndet mäts direkt från grund-datat. I båda fallen överträffar den föreslagna metoden basmetoden i termer av förutsägelsefel av återstående livslängd genom gated recurrent unit- nätverket. På det stora hela så visar sig metoden vara flexibel i skapandet av hälsoindikatorer med olika attribut och p.g.a. metodens egenskaper är den adaptiv för olika typer av metoder som förutspår återstående livslängd.

Predictive maintenance plays a crucial role in preventing unexpected equipment failures and maintaining assets in good operating conditions in various systems. One such scenario where predictive maintenance has been widely used is in battery management systems for electronic vehicles based on lithium batteries, where the risk of failure can be reduced by predicting the remaining useful life of the lithium battery. This project developed a DL model based on Long Short-Term Memory networks which was able to generalize new and various kinds of battery. The model was implemented on a low-cost, low-power using embedded artifcial intelligence, which enables local model execution, reducing costs, time, and risks associated with transferring data to the cloud. To further optimize the model and reduce its memory usage, quantization was applied before porting it to an embedded system based on the STM32 MCU. The results show that the model migration was successful, with low memory cost and no signifcant degradation in accuracy. Finally, the memory usage of the prediction model was also analyzed. / Predictiv underhåll har en avgörande roll för att förebygga oväntade utrustningsfel och bibehålla tillgångar i god driftsvillkor i olika system. Ett scenario där predictivt underhåll har använts mycket är i batterihanteringssystem för elfordon baserade på litiumbatterier, där risken för fel kan reduceras genom att förutsäga den återstående användbarhetsperioden för litiumbatteriet. I det här projektet utvecklades djupinlärningsprediktiva modeller med hjälp av Keras sekventiella modell för att representera en ferlagersneural nätverk och en Lång Korttidsminne modell för tidserieprediktion. Dessa modeller implementerades på en lågkostnad, låglägesmikrokontroller med inbyggd artifcial intelligence, vilket möjliggör lokal modellkörning, vilket reducerar kostnader, tid och risker med att överföra data till molnet. För att ytterligare optimera modellen och minska dess minnesfotavtryck tillämpades kvantisering innan den portades till en inbyggd system baserat på STM32 mikrokontroller. Resultaten visar att modellmigrationen var framgångsrik, med låg minneskostnad och ingen signifkant försämring av precisionen. Slutligen analyserades även minnesanvändningen av prediktionsmodellen.

The growing demand for sustainable technology has led to an increased application of power electronics. As these devices are often exposed to harsh conditions, their reliability is a primary concern for both manufacturers and users. Addressing these reliability challenges involves a set of activities known as Prognostics and Health Management (PHM). In PHM, predicting the Remaining Useful Life (RUL) is crucial. This prediction relies on identifying failure precursors, which signify the presence of degradation. These precursors are then used to construct a degradation model that enables the prediction of the remaining time that the device can work before failure. The project focuses on examining a MOSFET aging dataset from the NASA PCoE dataset depository and a diode aging dataset from Fraunhofer ENAS. The prediction of the remaining useful life of devices using failure precursors has been done by applying recurrent neural network (RNN) methods. However, the prediction results from a single feature is significantly deviated from the actual values. To improve the prediction, the age of the device was proposed as an additional feature. RNNs with a similar number of weights and RNNs with the same hyperparameters are implemented and their performance is evaluated by the accuracy of prediction. The results show that all the RNN models implemented manage to capture the characteristics of the aging data. Despite its simpler structure, the vanilla RNN manages to produce a comparable result with the GRU and LSTM by simpler mechanism and less number of weights. The results also reveal that the characteristics of the data have a significant impact on the final results. / Den växande efterfrågan på hållbar teknik har lett till en ökad tillämpning av kraftelektronik. Eftersom dessa enheter ofta utsätts för tuffa förhållanden är deras tillförlitlighet ett primärt bekymmer för både tillverkare och användare. Att ta itu med dessa tillförlitlighetsutmaningar innebär en uppsättning aktiviteter som kallas Prognostics and Health Management (PHM). I PHM är det avgörande att förutsäga det återstående användbara livet (RUL). Denna förutsägelse bygger på identifiering av felprekursorer, som anger förekomsten av nedbrytning. Dessa prekursorer används sedan för att konstruera en nedbrytningsmodell som möjliggör förutsägelse av den återstående tiden som enheten kan fungera innan fel. Projektet fokuserar på att undersöka en MOSFET-åldringsdataset från NASA PCoE-datauppsättningen och en diodåldringsdataset från Fraunhofer ENAS. Förutsägelsen av den återstående livslängden för enheter som använder felprekursorer har gjorts genom att använda metoder för återkommande neurala nätverk (RNN). Förutsägelseresultatet från en enskild funktion avviker dock avsevärt från de faktiska värdena. För att förbättra förutsägelsen föreslogs enhetens ålder som en extra funktion. RNN med ett liknande antal vikter och RNN med samma hyperparametrar implementeras och deras prestanda utvärderas av förutsägelsens noggrannhet. Resultaten visar att alla implementerade RNN-modeller lyckas fånga egenskaperna hos åldrande data. Trots sin enklare struktur lyckas vanilj RNN producera ett jämförbart resultat med GRU och LSTM genom enklare mekanism och färre antal vikter. Resultaten visar också att uppgifternas egenskaper har en betydande inverkan på de slutliga resultaten.

Transportsektorn är en av världens största producenter av växthusgas därav är dess avkarbonisering essentiell för att uppnå Parisavtalets mål för CO2-emissioner. Ett viktigt steg för att uppnå dessa mål utförs genom elektrifiering. Litium-jon-batterier (eng. litium-ion batteries, ’LIB’) har blivit väldigt populära energilagringssystem för batteridrivna elektriska fordon (eng. battery electric vehicles, ’BEV’) men tenderar att åldras, precis som alla andra batterier. Därav krävs forskning kring batteriföråldring på grund av nedbrytningsprocessernas inverkan på prissättningen, prestationerna och miljöpåverkan av BEV. Olika modeller används för att beskriva batteriernas åldrande. Datadrivna modeller som förutspår batteriers livstid ökar i popularitet vars noggrannhet och prestationer till stor del beror på indatats kvalitet. Formatet för tidsinhämtade data kräver enorma mängder lagringsutrymme, hög processkapacitet och längre processer; något ’reducerad’ eller ’aggregerad’ data delvis åtgärdar. Denna avhandling fokuserar på att utveckla en metodik för användning av dynamiska fordonsdata i ’aggregerad’ form. Tidsloggade data inhämtade från kallklimatstesting av Scanias BEV-prototyp användes varav interaktionseffekterna mellan diverse fordonsparametrar samt deras effekt på batteriåldring utifrån en batteriåldringsmodell analyserades. Olika tillvägagångssätt för strukturering av dynamiska fordonsdata i modellen undersöktes också. Tolv aggregeringsscenarion designades och testades. Dessutom valdes tre scenarion för uppskattningar och jämförelser av återstående användbar livslängd (eng. remaining-useful-life, ’RUL’) tillsammans med resultat från tidsinhämtade data. Slutligen drogs slutsatser om: parameterinteraktioner, struktur av dynamiska fordonsdata och RUL. Flera framtida utvecklingsområden har också föreslagits bland annat: tester av andra aggregeringstekniker, utöka modellen till tjänstefordon samt kategorisera användningsbeteenden av fordon för att förbättra RUL-uppskattningar. / The transport sector is one of the world’s largest greenhouse gas producing sector and it’s decarbonisation is imperative to achieve the CO2 emission targets set by the Paris Agreement. One important step towards achieving these targets is through electrification of the sector. Lithium-ion batteries (LIBs) have become very popular energy storage systems for battery electric vehicles (BEVs). However, LIBs like all other batteries, tend to age. Hence, the study of the battery ageing phenomena is very essential since the degradation in battery characteristics hugely determines the cost, performance and the environmental impact of BEVs. Different modelling approaches are used to represent battery ageing behaviour. Data-driven models for predicting the lifetime of batteries are becoming popular. However, the accuracy and performance of data-driven models largely depends upon the quality of data being used as the input. Time-sampled format of logging data results in huge data files requiring enormous amounts of storage space, high processing power requirements and longer processing times. Instead, using data in a ’reduced’ or ‘aggregated’ form can help in addressing these issues. This thesis work focuses on developing a methodology for using dynamic vehicle data in an ‘aggregated’ form. Time-sampled data from a Scania prototype BEV truck, recorded during cold climate test, was used. The interaction effects between various vehicle parameters and their effect on battery ageing in a battery ageing model were analyzed. Different approaches to structuring dynamic vehicle data for use in the model were also studied. Twelve aggregation scenarios were designed and tested. Furthermore, three scenarios were selected for making remaining-useful-life (RUL) estimations and compared alongside time-sampled data results. Finally, conclusions about parameter interactions, structuring of dynamic vehicle data and RUL estimations were drawn. Several next steps for future work have also been suggested such as testing other aggregation techniques, extending the model to vehicle fleets and categorizing vehicle usage behaviours to make better RUL estimations.