Dissertations / Theses on the topic 'Artificial Additives'

Submitted by Marco Antonio Oliveira da Silva (maosilva@ipen.br) on 2017-11-01T17:31:29Z No. of bitstreams: 0
Made available in DSpace on 2017-11-01T17:31:29Z (GMT). No. of bitstreams: 0
Os níveis de radioatividade natural do planeta e suas eventuais consequencias são objeto de estudo da radioproteção ambiental. Nos últimos anos, as agências internacionais ligadas à proteção radiológica têm debatido as práticas até então estipuladas, no que diz respeito à proteção da fauna e flora, cuja filosofia acreditava que as recomendações sugeridas para a proteção do homem asseguravam que outras espécies estariam também protegidas. Assim, são necessários estudos sobre as concentrações de atividade dos radionuclídeos dispersos no meio ambiente, assim como as doses absorvidas por organismos de diferentes ecossistemas, pela exposição interna e externa. O Brasil possui a segunda maior população de cães e gatos do mundo e produz anualmente mais de 2 milhões de toneladas de rações. O presente trabalho investigou os níveis de radioatividade presentes em rações comerciais para cães e gatos, por meio da espectrometria gama de alta resolução. Os resultados mostraram concentrações abaixo da MDA para radionuclídeos artificiais e baixas concentrações para radionuclídeos naturais, cujos valores variaram de 0,9 ± 0,3 Bq/kg a 5,1 ± 0,7 Bq/kg para o 226Ra, de 1,2 ± 0,4 Bq/kg a 11,1 ± 1,0 Bq/kg para o 232Th e de 156 ± 7 Bq/kg a 410 ± 19 Bq/kg para o 40K. Para verificar a composição de alguns minerais, foi empregada a técnica por EDXRF e, utilizando estatística multivariada, foi possível verificar as correlações entre os radionuclídeos e o conteúdo mineral encontrado. A boa correlação que foi observada entre as concentrações de 226Ra, 232Th e cálcio, pode estar associada ao uso de farinhas de carne e ossos na fabricação das rações. As doses internas para alguns órgãos foram inferidas pelo método de Monte Carlo, obtendo valores menores que 1 μGy/dia. Em síntese, os resultados mostraram que os níveis de atividade encontrados nas rações são baixos o suficiente para concluir que as marcas de ração canina avaliadas não fornecem riscos radiológicos para os animais que as consomem.
Tese (Doutorado em Tecnologia Nuclear)
IPEN/T
Instituto de Pesquisas Energéticas e Nucleares - IPEN-CNEN/SP

Coatings for exterior wood have two basic functions. One isto give an aesthetically acceptable surface appearance andcolour. The other is to provide protection against wooddegradation by microbiological or physical attack. Theseprotective properties, often called the barrier properties,play an important role in the selection of proper material forsupreme durability. The assessment of these barrier propertiesis of great importance. Within the CEN Technical Committee 139,Working Group 2, Coatings for exterior wood test methods andperformance specifications have been established. Forassessment of water protection efficiency a standard testprocedure, EN 927-5, has been launched. The present work hasfocused on its applicability in assessing water protectionefficiency in relation to the degradation of coatings duringexposure.

Assessments according to EN 927-5 is shown to givesignificant differences in water absorption values fordifferent types of coatings on wood. The proposed performancespecifications in ENV 927 - 2 for the water absorption valuesfor coatings to be used in different constructions seem to beset at acceptable levels. It has been shown that there is agood correlation between the level of water absorption andpractical experience of the performance of paints inScandinavia. However, it has also been shown that thecombination of a standard procedure for water absorptionmeasurement and an artificial weathering procedure gives moreinformation regarding expected durability and longtermperformance than a single measurement of water absorption onfresh, unweathered coated wood. A combination of waterabsorption measurement and artificial weathering could become auseful tool in product development as well as in benchmarking.Together with statistical tools, such as reliability-basedservice life prediction methodologies for predicting theservice life of coating systems a reduction in testing timesmay be achieved.

Surface-active substances in coatings have a negative effecton the coatings ability to prevent water ingress, which mostprobably is due to the hydrophilic character of thesesubstances. The presence of these substances, which are commonin waterborne coatings, alters the moisture sorptioncharacteristics of wood. Considering that these substancesoccur in waterborne coatings, may be mobilised during and afterfilm formation and accumulate at the coating/substrateinterface, there is a great probability that these substanceschange the moisture sorption characteristics of the woodsubstrate in an unfavourable way and create unexpected dynamicmoisture conditions at the coating/wood interface.

Computerized tomography has been found to be a valuable,non-destructive tool for visualising the spatial moisturedistribution of water and moisture in coated wood panels. Themethod is suitable for recurrent testing of a specimen exposedto a series of processes of wetting, drying. This thesis showsthat high moisture contents occur locally at and around cracks.In these areas there is an increased risk of internal tensionand stress resulting in crack initiation and propagation andthat high moisture contents may occur in the first fewmillimetres under waterborne coatings despite intact coatingfilms. Even with good barrier properties of the coating,moisture may accumulate by water-vapour absorption in air gapsbehind the cladding thus causing favourable conditions formicrobiological colonization.

The work that has been carried out regarding assessment ofthe water protection efficiency shows promising resultsregarding the possibility to use reliability-based service lifeprediction methodology for the assessment. The aim of futurework will be to establish more reliable techniques andprotocols for assessing service life expectancy and durability,especially for waterborne coatings with special focus ontendencies to early failure and robustness of the coatingsystems.

Keywords:Coatings, surfactants, water absorption value,EN 927, paint, additives, moisture dynamics, absorption,desorption, artificial weathering, artificial exposure,computerized tomography, MRI.

Aus der Einführung: "Stehen wir am Rande einer bio-nanotechnologischen getriebenen Revolution, die unsere Art zu leben, zu arbeiten und miteinander umzugehen grundlegend verändern wird? Welchem gesellschaftspolitischen, wirtschaftlichen und technologischen Wandel haben wir uns zu stellen? Langfristige Entwicklungszyklen (Kondratieff, Schumpeter) führen zur nachhaltigen Weiterentwicklung der Zivilisation. Mittelfristige Entwicklungen wie die Trends Globalisierung, Urbanisierung, Digitalisierung (Miniaturisierung) und Humanisierung (Individualisierung), die immer stärker unser Umfeld und Handeln beeinflussen führen zu ganzheitlichen, weltumspannenden Grundtendenzen der gesellschaftlichen Weiterentwicklung. Die technologischen "Enabler" Computing, Biotechnology, Artifical Intelligence, Robotik, Nanotechnology, Additive Manufacturing und Design Thinking wirken beschleunigend auf die gesellschaftlichen Entwicklungen ein. Die technologischen Möglichkeiten beschleunigen sowohl gesellschaftspolitische Zyklen und zivilisatorische Anpassungen. Durch rasanten technologischen, wissenschaftlichen Fortschritt, zunehmende Globalisierungswirkungen, beschleunigte Urbanisierung und aber auch politischer Interferenzen sind die Veränderungsparameter eines dynamischen Geschäftsumfelds immer schnellere Transformationen ausgesetzt. Alle diese Richtungen zeigen das unsere gesellschaftliche Entwicklung inzwischen stark durch die Technik getrieben ist. Ob dies auch heißt, dass wir den Punkt der Singularität (Kurzweil) absehbar erreichen ist dennoch noch offen. ..."

Artificial reefs have been deployed throughout the world’s oceans to act as habitat and fishing enhancement tools. To expand current research on the role of artificial reefs in the marine community, ordination and multivariate regression methods were used here to analyze survey data of natural and artificial reefs. The reefs, located in the Northern Gulf of Mexico (NGOM) and on the West Florida Shelf (WFS), had been previously surveyed from 2004 to 2015 using remote operated vehicle and stationary video techniques. This study tested the hypothesis that similar functional roles are accounted for at both natural and artificial reef sites even if species composition varies. Secondly, it examines the role of environment and fisheries in determining the assemblages. Artificial reefs tended to host communities that were as biodiverse as natural reefs, although not necessarily composed of the same species. Results of an ordination confirmed that as the classification was broadened from the level of species, to family, to functional group, the assemblages on each reef type (natural vs. artificial and NGOM vs WFS) appeared more similar. Dominant groups were present at all levels of classification and included the families Lutjanidae and Carangidae, as well as functional groups Red Snapper and Small Reef Fish. Both natural and artificial reefs tended to be dominated by one of the following: Lutjanidae, Carangidae, or Small Reef Fish, although a continuous gradient was found across the extremes of natural versus artificial reefs. Generalized Additive Models were developed to examine the influence of reef type, location, environment and fishing intensity covariates. Results indicated that for both natural and artificial reefs, the abundance of families and functional groups can be influenced by environmental factors. In both cases, there is strong spatial autocorrelation suggesting connectivity with neighboring reefs.

Human beings have a touch and force estimation mechanism beneath their feet. They use this feeling of touch and force to maintain balance, walk, run and perform various agile motions. This paper presents a new sensor platform beneath the humanoid feet, enabled by a pragmatic model based compliant foot design and sensor configuration that mimics the human tactile sensory system for contact force measurement in humanoid robots. Unlike previous force sensor based approaches, the system is defined as a total and sufficient method of Ground Reaction Force (GRF) and Zero Moment Point (ZMP) measurement for balancing and walking using contact force feedback in mid to full sized humanoids. The conventional systems for the GRF and ZMP measurement are made of heavy metallic parts that tend to be bulky and vulnerable to inertial noises upon high acceleration. In addition to low cost and reliable operation, the proposed system can withstand shock and enable agile motion much like humans do with their footpad. The proposed foot is manufactured using state-of-the-art technique with elastomer padding which not only protects the sensors but also acts as a compliance beneath the foot giving integrity in structural design. This composite layer provides compliance and traction for foot collision while the contact surfaces are sampled for pressure distribution which can be mapped into three axis force and ZMP. A single step training process is required to relate the sensor readings to force measurement. The system’s capability of contact force measurement, subsequent ZMP estimation is experimentally verified with the application of appropriate software. Moreover, a simulation study has been conducted via Finite Element Analysis (FEA) of the footpad structure to analyze the proposed footpad structure. The experimental results demonstrate why this can be a major step toward a biomimetic, affordable yet robust contact force and ZMP measurement method for humanoid robots. This work was supported by the Office of Naval Research, Grant N00014-15-1-2128 as part of development of Project SAFFiR (Shipboard Autonomous Firefighting Robot).
Master of Science
How we interact with the surfaces in contact with us has a crucial role for balancing and walking with agility. The biological touch and force measurement systems in human is currently unmatched, not even mimicked in a significant way in the state-of-the-art humanoid robots’ systems. Human beings use this feeling of touch and force beneath the feet to maintain balance, walk, run and perform various agile motions. This research aims to find a holistic system in humanoid robot’s feet design that can mimic this human characteristics of force estimation beneath the feet and using that estimation for balancing and walking. A practical model based sensor configuration is derived from the rigorous study of human and humanoid robot’s feet contact with the ground. The sensors are tactile in nature, and unlike previous below feet based approaches, the system is defined as a total and sufficient system of Ground Reaction Force (GRF) and Center of Pressure (CoP) measurement. The conventional systems for this purpose are not only highly expensive but also having error in quantification during accelerated movement. The proposed foot is designed following the practical model derived and manufactured using the state-of-the-art mechanism for having a soft cushion between the sensors and the contact surfaces. In addition to low cost and reliable operation, the proposed system can withstand shock and enable agile motion much like humans do with their footpad. The quantification of the forces and pressure from the sensor readings and developed using appropriate software and algorithms. The system’s capability of contact force measurement, subsequent Center of Pressure measurement is experimentally verified with the application of appropriate software. Moreover, a simulation study has been conducted of the footpad structure to analyze the proposed footpad structure. The experimental results demonstrate why this can be a major step toward a biomimetic, affordable yet robust contact force and Center of Pressure measurement method for human-like robots.

Machine learning-based systems are rapidly becoming popular because it has been realized that machines are more efficient and effective than humans at performing certain tasks. Although machine learning algorithms are extremely popular, they are also very literal and undeviating. This has led to a huge research surge in the field of interpretability in machine learning to ensure that machine learning models are reliable, fair, and can be held liable for their decision-making process. Moreover, in most real-world problems just making predictions using machine learning algorithms only solves the problem partially. Time series is one of the most popular and important data types because of its dominant presence in the fields of business, economics, and engineering. Despite this, interpretability in time series is still relatively unexplored as compared to tabular, text, and image data. With the growing research in the field of interpretability in machine learning, there is also a pressing need to be able to quantify the quality of explanations produced after interpreting machine learning models. Due to this reason, evaluation of interpretability is extremely important. The evaluation of interpretability for models built on time series seems completely unexplored in research circles. This thesis work focused on achieving and evaluating model agnostic interpretability in a time series forecasting problem.  The use case discussed in this thesis work focused on finding a solution to a problem faced by a digital consultancy company. The digital consultancy wants to take a data-driven approach to understand the effect of various sales related activities in the company on the sales deals closed by the company. The solution involved framing the problem as a time series forecasting problem to predict the sales deals and interpreting the underlying forecasting model. The interpretability was achieved using two novel model agnostic interpretability techniques, Local interpretable model- agnostic explanations (LIME) and Shapley additive explanations (SHAP). The explanations produced after achieving interpretability were evaluated using human evaluation of interpretability. The results of the human evaluation studies clearly indicate that the explanations produced by LIME and SHAP greatly helped lay humans in understanding the predictions made by the machine learning model. The human evaluation study results also indicated that LIME and SHAP explanations were almost equally understandable with LIME performing better but with a very small margin. The work done during this project can easily be extended to any time series forecasting or classification scenario for achieving and evaluating interpretability. Furthermore, this work can offer a very good framework for achieving and evaluating interpretability in any machine learning-based regression or classification problem.
Maskininlärningsbaserade system blir snabbt populära eftersom man har insett att maskiner är effektivare än människor när det gäller att utföra vissa uppgifter. Även om maskininlärningsalgoritmer är extremt populära, är de också mycket bokstavliga. Detta har lett till en enorm forskningsökning inom området tolkbarhet i maskininlärning för att säkerställa att maskininlärningsmodeller är tillförlitliga, rättvisa och kan hållas ansvariga för deras beslutsprocess. Dessutom löser problemet i de flesta verkliga problem bara att göra förutsägelser med maskininlärningsalgoritmer bara delvis. Tidsserier är en av de mest populära och viktiga datatyperna på grund av dess dominerande närvaro inom affärsverksamhet, ekonomi och teknik. Trots detta är tolkningsförmågan i tidsserier fortfarande relativt outforskad jämfört med tabell-, text- och bilddata. Med den växande forskningen inom området tolkbarhet inom maskininlärning finns det också ett stort behov av att kunna kvantifiera kvaliteten på förklaringar som produceras efter tolkning av maskininlärningsmodeller. Av denna anledning är utvärdering av tolkbarhet extremt viktig. Utvärderingen av tolkbarhet för modeller som bygger på tidsserier verkar helt outforskad i forskarkretsar. Detta uppsatsarbete fokuserar på att uppnå och utvärdera agnostisk modelltolkbarhet i ett tidsserieprognosproblem.  Fokus ligger i att hitta lösningen på ett problem som ett digitalt konsultföretag står inför som användningsfall. Det digitala konsultföretaget vill använda en datadriven metod för att förstå effekten av olika försäljningsrelaterade aktiviteter i företaget på de försäljningsavtal som företaget stänger. Lösningen innebar att inrama problemet som ett tidsserieprognosproblem för att förutsäga försäljningsavtalen och tolka den underliggande prognosmodellen. Tolkningsförmågan uppnåddes med hjälp av två nya tekniker för agnostisk tolkbarhet, lokala tolkbara modellagnostiska förklaringar (LIME) och Shapley additiva förklaringar (SHAP). Förklaringarna som producerats efter att ha uppnått tolkbarhet utvärderades med hjälp av mänsklig utvärdering av tolkbarhet. Resultaten av de mänskliga utvärderingsstudierna visar tydligt att de förklaringar som produceras av LIME och SHAP starkt hjälpte människor att förstå förutsägelserna från maskininlärningsmodellen. De mänskliga utvärderingsstudieresultaten visade också att LIME- och SHAP-förklaringar var nästan lika förståeliga med LIME som presterade bättre men med en mycket liten marginal. Arbetet som utförts under detta projekt kan enkelt utvidgas till alla tidsserieprognoser eller klassificeringsscenarier för att uppnå och utvärdera tolkbarhet. Dessutom kan detta arbete erbjuda en mycket bra ram för att uppnå och utvärdera tolkbarhet i alla maskininlärningsbaserade regressions- eller klassificeringsproblem.

In this dissertation, generalized additive neural networks (GANNs) and multilayer perceptrons (MLPs) are studied and compared as prediction techniques. MLPs are the most widely used type of artificial neural network (ANN), but are considered black boxes with regard to interpretability. There is currently no simple a priori method to determine the number of hidden neurons in each of the hidden layers of ANNs. Guidelines exist that are either heuristic or based on simulations that are derived from limited experiments. A modified version of the neural network construction with cross–validation samples (N2C2S) algorithm is therefore implemented and utilized to construct good MLP models. This algorithm enables the comparison with GANN models. GANNs are a relatively new type of ANN, based on the generalized additive model. The architecture of a GANN is less complex compared to MLPs and results can be interpreted with a graphical method, called the partial residual plot. A GANN consists of an input layer where each of the input nodes has its own MLP with one hidden layer. Originally, GANNs were constructed by interpreting partial residual plots. This method is time consuming and subjective, which may lead to the creation of suboptimal models. Consequently, an automated construction algorithm for GANNs was created and implemented in the SAS R statistical language. This system was called AutoGANN and is used to create good GANN models. A number of experiments are conducted on five publicly available data sets to gain insight into the similarities and differences between GANN and MLP models. The data sets include regression and classification tasks. In–sample model selection with the SBC model selection criterion and out–of–sample model selection with the average validation error as model selection criterion are performed. The models created are compared in terms of predictive accuracy, model complexity, comprehensibility, ease of construction and utility. The results show that the choice of model is highly dependent on the problem, as no single model always outperforms the other in terms of predictive accuracy. GANNs may be suggested for problems where interpretability of the results is important. The time taken to construct good MLP models by the modified N2C2S algorithm may be shorter than the time to build good GANN models by the automated construction algorithm
Thesis (M.Sc. (Computer Science))--North-West University, Potchefstroom Campus, 2011.

En aquest treball es defineix el concepte de generador additiu de t–normes i de t–conormes discretes. S’hi estableixen resultats generals sobre la generació additiva de disjuncions i les caracteritzacions dels generadors de les t–conormes bàsiques. Es planteja un algorisme per a decidir quan una disjunció és additivament generable, basat en l’algorisme Gamma de la teoria de convexitat. S’estudia la relació que hi ha entre la generació additiva amb la suma ordinal i amb l’anidament. S’introdueixen els conceptes de generador concau i generador convex. S’estudia la generació additiva de les disjuncions i les t–conormes suaus i bivalents sobre L*. S’insisteix amb l’aplicabilitat de la generació additiva quan es tracta de manejar la condició de T–transitivitat per a relacions d’indistingibilitat discretes. Finalment, s’estudia la relació que hi ha entre la generació additiva d’una t–conorma S i les propietats de l’S–implicació corresponent. Amb motiu de les propietats d’ordre i modus ponens generalitzat, es defineixen els generadors mixtos

Face aux problématiques de gestion des sédiments de dragage portuaire, aujourd'hui, il apparait nécessaire de trouver des solutions potentielles de valorisation permettant de répondre efficacement à ces problématiques. La valorisation en génie civil, des sédiments des port de Saint Louis et de Dunkerque, présente une solution alternative à la gestion de ces matériaux. les travaux menés dans le cadre de cette thèse se focalisent d'une part, sur la valorisation des sédiments de Saint louis en technique routière et d'autre part, sur la valorisation des sédiments de Dunkerque en béton ainsi qu'en granulats artificiels. Après la caractérisation physico-chimique, minéralogique et mécanique, l'impact environnemental des sédiments bruts du Port de Saint Louis et le potentiel d'utilisation de ces matériaux en technique routière sont évalués. L'étude de formulation de matériaux, pour une utilisation en couche de fondation, s'est fondé sur une méthode expérimentale de détermination de compacité maximale. les mélanges granulaires optimaux remplissent les conditions d'utilisation en couche de fondation sont ensuite évalué au travers d'essais de lixiviation. La partie béton concerne la formulation des bétons à base de sédiment. Après la caractérisation de ces matériaux, trois bétons ont fait l'objet d'une étude de durabilité face à l'attaque sulfatique externe associé à une caractérisation non destructive en vue d'étudier l'influence d'incorporation des sédiments sur les propriétés des bétons. La partie granulats artificiels présente d'étude de faisabilité de granulats à base de sédiments, l'assiette granulaire et la technique de big-bag sont les deux méthodes de confection utilisées
In front of problems of management of harbour dredging sediment, today, it appears necessary to find potential solutions of valosisation allowing to answer effectively these problems. the valorisation in civil engineering, of not immergeables sediments of ports of Dunkirk and Saint Louis, presents an alternative solution to the management of these materials. the works undertaken within the framework of this thesis are focused on the one hand, on the valorisation of Saint Louis sediment in road construction, and the other hand on the valorisation of Dunkirk sediment in concrete as well as artificials agregates. After physicochimical characterisation, mineralogical and mechanical, environmental impact of raw sediment of Saint Louis harbour and the potential use of these materials in road constuction are evaluated. the study of formulation of materials, for use in layer fondation, was based on an experimental method of determination of maximum compactness. The optimal granular mixtures fulfilling the terms of a use in a layer fondation are the evaluated through leaching tests. The concrete party concerns the formulation of the concretes containing Dunkerk sediment. After the characterisation of these materials, three concretes were the object of a durability study vis-a-vis the external sulphate attack associated with a not destructive characterisation in order to study the influence of incorporation of sediment on properties of concretes. The artificial aggregates party presents the feasibility study of aggregates with sediment, the granular plate and the big-bag technique are two making method used

This thesis presents an extension to the Temporal Fast Downward planning system that integrates motion planning in it and algorithms for generating two types of temporal macro operators expressible in PDDL2.1. The extension to the Temporal Fast Downward planning system includes, in addition to the integration of motion planning itself, an extension to the context-enhanced additive heuristic that uses information from the motion planning part to improve the heuristic estimate. The temporal macro operators expressible in PDDL2.1 are, to the author's knowledge, an area that is not studied within the context of plan repair before. Two types of temporal macro operators are presented along with algorithms for automatically constructing and using them when solving plan repair problems by replanning. Both the heuristic extension and the temporal macro operators were evaluated in the context of simulated unmanned aerial vehicles autonomously executing reconnaissance missions to identify targets and avoiding threats in unexplored areas. The heuristic extension was proved to be very helpful in the scenario. Unfortunately, the evaluation of the temporal macro operators indicated that the cost of introducing them is higher than the gain of using them for the scenario.

In the context of decarbonizing the electricity generation of French non-interconnected territories, the knowledge of future electricity demand, in particular annual and peak demand in the long-term, is crucial to design new renewable energy infrastructures. So far, these territories, mainly islands located in the Pacific and Indian ocean, relies mainly on fossil fuels powered facilities. Energy policies envision to widely develop renewable energies to move towards a low-carbon electricity mix by 2028.  This thesis focuses on the long-term forecasting of hourly electricity demand. A methodology is developed to design and select a model able to fit accurately historical data and to forecast future demand in these particular territories. Historical data are first analyzed through a clustering analysis to identify trends and patterns, based on a k-means clustering algorithm. Specific calendar inputs are then designed to consider these first observations. External inputs, such as weather data, economic and demographic variables, are also included.  Forecasting algorithms are selected based on the literature and they are than tested and compared on different input datasets. These input datasets, besides the calendar and external variables mentioned, include different number of lagged values, from zero to three. The combination of model and input dataset which gives the most accurate results on the testing set is selected to forecast future electricity demand. The inclusion of lagged values leads to considerable improvements in accuracy. Although gradient boosting regression features the lowest errors, it is not able to detect peaks of electricity demand correctly. On the contrary, artificial neural network (ANN) demonstrates a great ability to fit historical data and demonstrates a good accuracy on the testing set, as well as for peak demand prediction. Generalized additive model, a relatively new model in the energy forecasting field, gives promising results as its performances are close to the one of ANN and represent an interesting model for future research.  Based on the future values of inputs, the electricity demand in 2028 in Réunion was forecasted using ANN. The electricity demand is expected to reach more than 2.3 GWh and the peak demand about 485 MW. This represents a growth of 12.7% and 14.6% respectively compared to 2019 levels.
I samband med utfasningen av fossila källor för elproduktion i franska icke-sammankopplade territorier är kunskapen om framtida elbehov, särskilt årlig förbrukning och topplast på lång sikt, avgörande för att utforma ny infrastruktur för förnybar energi. Hittills är dessa territorier, främst öar som ligger i Stilla havet och Indiska oceanen, beroende av anläggningar med fossila bränslen. Energipolitiken planerar att på bred front utveckla förnybar energi för att gå mot en koldioxidsnål elmix till 2028.  Denna avhandling fokuserar på den långsiktiga prognosen för elbehov per timme. En metod är utvecklad för att utforma och välja en modell som kan passa korrekt historisk data och för att förutsäga framtida efterfrågan inom dessa specifika områden. Historiska data analyseras först genom en klusteranalys för att identifiera trender och mönster, baserat på en k-means klusteralgoritm. Specifika kalenderinmatningar utformas sedan för att beakta dessa första observationer. Externa inmatningar, såsom väderdata, ekonomiska och demografiska variabler, ingår också.  Prognosalgoritmer väljs utifrån litteraturen och de testas och jämförs på olika inmatade dataset. Dessa inmatade dataset, förutom den nämnda kalenderdatan och externa variabler, innehåller olika antal fördröjda värden, från noll till tre. Kombinationen av modell och inmatat dataset som ger de mest exakta resultaten på testdvärdena väljs för att förutsäga framtida elbehov. Införandet av fördröjda värden leder till betydande förbättringar i exakthet. Även om gradientförstärkande regression har de lägsta felen kan den inte upptäcka toppar av elbehov korrekt. Tvärtom, visar artificiella neurala nätverk (ANN) en stor förmåga att passa historiska data och visar en god noggrannhet på testuppsättningen, liksom för förutsägelse av toppefterfrågan. En generaliserad tillsatsmodell, en relativt ny modell inom energiprognosfältet, ger lovande resultat eftersom dess prestanda ligger nära den för ANN och representerar en intressant modell för framtida forskning.  Baserat på de framtida värdena på indata, prognostiserades elbehovet 2028 i Réunion med ANN. Elbehovet förväntas nå mer än 2,3 GWh och toppbehovet cirka 485 MW. Detta motsvarar en tillväxt på 12,7% respektive 14,6% jämfört med 2019 års nivåer.

碩士
國立臺北科技大學
化學工程研究所
102
The most common artificial joint material is ultra-high molecular weight polyethylene (UHMWPE) and Cobalt-Chromium-Molybdenum alloy (CoCrMo alloy). UHMWPE wear debris could lead to osteolysis and that result in loosening and thus failure of the implant. Periprosthetic fluid presents between artificial joint surfaces and acts as a lubricant which prevents wear and reduces friction. From previous studies, the most significant combination of lubricant composition is bovine calf serum containing 4.5mg/mL of alginic acid and 4.5mg/mL of carrageenan. Therefore, our experiment will investigate lubrication mechanism of alginic acid and carrageenan between proteins and adjust alginic acid and carrageenan different combinations of concentrations to do friction test. In this study, pin-on-disc friction test and reciprocating wear testing machine were carried out to investigate two kinds of biomolecules additives to change different combinations of concentrations on friction and wear behavior of artificial joints. We observed when bovine serum containing both alginic acid and carrageenan, the coefficient of friction is lower than carrageenan. The most significant combination of lubricant composition is bovine calf serum containing 2.25mg/mL of alginic acid and 12.5mg/mL of carrageenan. And this lubricant formulation also reduces the wear. And then investigate two biomolecules additives alginic acid and carrageenan between protein molecules interactions by using rheological properties and quartz crystal microbalance. The results show that viscosity is not the main reason to affect the coefficient of friction; we surmise the adsorption behavior of molecules is resulting the friction coefficient decrease. In the future, we can use the FITC investigate the lubrication mechanism , and artificial joint simulator testing in accordance with ISO standards and verified, this benefit in patients with artificial joints and prolong the life of artificial joints.

碩士
國立臺北科技大學
化學工程研究所
101
Ultra-high molecular molecular weight polyethylene (UHMWPE) and Co-Cr-Mo alloy are common artificial joint materials.Wear debris of UHMWPE will induce immune response of human body which leads to osteolysis.Moreover, it will accelerate the loosening of artificial joint and reduce the lifetime of artificial joints. The compositions of human synovial fluid are abundant. From previous studies, we found out that the key moleculars affecting the tribological behavior of artificial joints are albumin and hyaluronic acid. In this study, we applied the pin-on-disc friction tests to screen out the potential bio-molecular additives to reduce friction.We applied 25v/v% bovine calf serum as lubricant which is used in artificial joint simulator.Hyaluronic acid, carboxymethyl cellulose, alginic acid sodium, carrageenan are applied as the additives in BCS in this study. We investigated and analyzed the tribiological behavior of lubricants with various compositions. The result showed that alginic acid sodium is the most effective biomolecular additive. The lowest of friction of coefficient will be shown when adding 12.5mg/ml of alginic acid sodium. Furthermore, we investigated the difference in the lubrication characteristics of artificial joint materials when albumin solution was used as lubricant. The testing results show that the lubrication can be improved after adding the single molecule into albumin under boundary lubrication. Adding alginic acid could lead to the lowest friction of coefficient which is similar to the result when BCS was used as lubricant. Moreover, this study suggests that the most effective combination of lubricant composition is bovine calf serum containing 4.5mg/ml of alginic acid and 4.5mg/ml of carrageenan. It shall be further designed to run the artificial joint simulator tests to verify the phenomena observed in this study.

In the present study, the potential of low-melting point lipids, namely methyl palmitate (MP) and menhaden stearine (MS), were studied for delivery of water-soluble materials to fish larvae. Addition of the solid emulsifier, sorbitan monopalmitate (SMP) resulted in higher RE compared to the liquid emulsifier sorbitan sesquioleate (SSO). Additions of SMP were found to have a positive effect on the stability of lipid spray beads (LSB) at concentrations >10% (w/w) of lipid. LSB were shown to retain riboflavin with the highest RE when the lipid matrix was composed of a mixture of MP+10% SMP+1O% ethyl cellulose (w/w). Use of MP facilitated digestive breakdown of LSB by larvae of zebrafish, Brachydanio rerio, and glowlight tetra, Hemigrammus erythrozonus, as indicated by release of riboflavin from ingested LSB. LSB composed of MS performed better than LSB composed of MP when particulate glycine was incorporated within LSB. Highest retention efficiencies (RE) and delivery efficiencies (DE) were achieved by LSB composed of 100% MS. RE and DE of LSB were further improved when glycine was encapsulated as aqueous droplets rather than in a particulate form. The overall performance of LSB containing tyrosine showed that nutrients with similar water-solubilities to tyrosine can be very effectively delivered to early fish larvae. Visual observation indicated that LSB composed of 100% MS were broken down by larvae of clownfish, Amphiprion percula. Zein-bound particles (ZBP) were not efficient in delivering riboflavin to fish larvae due to high leaching rates. In an attempt to improve ZBP, we developed complex particles (CP) containing LSB incorporated in a dietary mixture bound by zein, to obtain a single particle type. The potential of LSB for delivery of free amino acids (FAA) was evaluated when LSB were incorporated within CP. RE and DE for tyrosine were significantly greater than those for alanine, glycine and serine. There was a significant inverse correlation between RE for FAA and their solubilities where FAA with low solubilites resulted in higher RE. Development of CP with high RE and DE for amino acids is an important step towards formulating a diet to allow early weaning of fish larvae onto artificial diets.
Graduation date: 2003

碩士
大同大學
生物工程學系(所)
102
Food additives has been widespread used .But, illegal or abuse of food additives has led to the rampancy of food fraud. The physiological function of intellectually disabled individuals is greatly affected by their eating habit, and todays’ food package labeling comes in all sorts that easily confuse people. Hence, it is a pressing issue to understand intellectually disabled individuals’ ability to choose and buy food and identify food additives. This study aimed to investigate the intellectually disabled students’ eating behavior, labeling of artificial foods with additives, and food purchase intention-using a special education school as example. (1) The intellectually disabled students from different grades reached significant standards in “labeling of food additives” and “awareness of food additives”; the intellectually disabled students with varying degrees of barriers reached significant standards in “labeling of food additives”, “awareness of food additives”, and “food purchase intention”. (2) Labeling and knowledge of food additives showed a highly positive correlation(r=.703,p=.000);labeling of food additives and food purchase intention showed a highly positive correlation (r=.711,p=.000); the relationship between knowledge of food additives and food purchase intention also showed a highly positive correlation(r=.736,p=.000).

碩士
國立臺灣大學
機械工程學研究所
104
The cornea is the outermost part of the eye and a highly transparent organization without vessels. However, due to injury, infection or disease, the cornea can cause bleaching or transmittance decline. In the worldwide, more than millions people are blind due to corneal problems. The global need for artificial cornea is driven by both the population that cannot tolerate donor corneas and the severe shortage of donor corneas. Additive manufacturing provides the opportunity to produce substitutes of the native tissues, and, in turn, to produce customized tissue constructs. This study aims to analyze and design various additive manufacturing processes for artificial cornea. COMSOL simulation presents the deformation of the corneal structure during fabrication. Moreover, 3D printing enables accurate temperature and pressure control in construction on corneal structure during dispensing and photo-curing of diacrylate-terminated Poloxamer 407 (P407DA) hydrogel. With the precise control at ambient temperature (15°C) and additional air pressure support (45 Pa), fabricated corneas can be formed with a smooth surface and light transmission all over 82% in the range of visible light by additive manufacturing process.

Cette thèse envisage un ensemble de méthodes permettant aux algorithmes d'apprentissage statistique de mieux traiter la nature séquentielle des problèmes de gestion de portefeuilles financiers. Nous débutons par une considération du problème général de la composition d'algorithmes d'apprentissage devant gérer des tâches séquentielles, en particulier celui de la mise-à-jour efficace des ensembles d'apprentissage dans un cadre de validation séquentielle. Nous énumérons les desiderata que des primitives de composition doivent satisfaire, et faisons ressortir la difficulté de les atteindre de façon rigoureuse et efficace. Nous poursuivons en présentant un ensemble d'algorithmes qui atteignent ces objectifs et présentons une étude de cas d'un système complexe de prise de décision financière utilisant ces techniques. Nous décrivons ensuite une méthode générale permettant de transformer un problème de décision séquentielle non-Markovien en un problème d'apprentissage supervisé en employant un algorithme de recherche basé sur les K meilleurs chemins. Nous traitons d'une application en gestion de portefeuille où nous entraînons un algorithme d'apprentissage à optimiser directement un ratio de Sharpe (ou autre critère non-additif incorporant une aversion au risque). Nous illustrons l'approche par une étude expérimentale approfondie, proposant une architecture de réseaux de neurones spécialisée à la gestion de portefeuille et la comparant à plusieurs alternatives. Finalement, nous introduisons une représentation fonctionnelle de séries chronologiques permettant à des prévisions d'être effectuées sur un horizon variable, tout en utilisant un ensemble informationnel révélé de manière progressive. L'approche est basée sur l'utilisation des processus Gaussiens, lesquels fournissent une matrice de covariance complète entre tous les points pour lesquels une prévision est demandée. Cette information est utilisée à bon escient par un algorithme qui transige activement des écarts de cours (price spreads) entre des contrats à terme sur commodités. L'approche proposée produit, hors échantillon, un rendement ajusté pour le risque significatif, après frais de transactions, sur un portefeuille de 30 actifs.
This thesis considers a number of approaches to make machine learning algorithms better suited to the sequential nature of financial portfolio management tasks. We start by considering the problem of the general composition of learning algorithms that must handle temporal learning tasks, in particular that of creating and efficiently updating the training sets in a sequential simulation framework. We enumerate the desiderata that composition primitives should satisfy, and underscore the difficulty of rigorously and efficiently reaching them. We follow by introducing a set of algorithms that accomplish the desired objectives, presenting a case-study of a real-world complex learning system for financial decision-making that uses those techniques. We then describe a general method to transform a non-Markovian sequential decision problem into a supervised learning problem using a K-best paths search algorithm. We consider an application in financial portfolio management where we train a learning algorithm to directly optimize a Sharpe Ratio (or other risk-averse non-additive) utility function. We illustrate the approach by demonstrating extensive experimental results using a neural network architecture specialized for portfolio management and compare against well-known alternatives. Finally, we introduce a functional representation of time series which allows forecasts to be performed over an unspecified horizon with progressively-revealed information sets. By virtue of using Gaussian processes, a complete covariance matrix between forecasts at several time-steps is available. This information is put to use in an application to actively trade price spreads between commodity futures contracts. The approach delivers impressive out-of-sample risk-adjusted returns after transaction costs on a portfolio of 30 spreads.

Dissertação de Mestrado Integrado em Engenharia Mecânica apresentada à Faculdade de Ciências e Tecnologia
Com o desenvolvimento tecnológico, os processos de Fabrico Aditivo têm alcançado uma importância acrescida no mundo da indústria quando comparados com outros processos de fabrico. Tal é devido ao facto de para além de serem capazes de produzir estruturas tridimensionais complexas, que não são obtidas por nenhuma outra tecnologia, também o processo produz uma quantidade mínima, ou mesmo nula, de desperdício. A quantidade de matéria-prima que é utilizada é menor quando comparada com outros processos, o que leva à produção de partes, componentes e dispositivos mais económicos. Estes fatores têm uma elevada importância, de tal forma que se sobrepõem a algumas desvantagens associadas, em alguns casos, como à qualidade do acabamento superficial e à tolerância geométrica. Atualmente, os processos de Fabrico Aditivo têm a oportunidade de criar impacto no mundo da produção, sendo um elemento fulcral da Indústria 4.0De todas as tecnologias associada ao fabrico aditivo, a Modelagem por Deposição Fundida (Fused Deposition Modeling – FDM), comummente designada por Impressão 3D, é a mais fácil de utilizar, e que requere equipamentos mais acessíveis e de fácil aquisição. Esta dissertação centra-se na utilização desta tecnologia na produção de protetores bocais para atletas. O processo atual de produção destes dispositivos não utiliza um design adequado, e a produção personalizada envolve um elevado custo. Através do fabrico por FDM, é possível produzir protetores bocais completamente personalizados, por um custo muito menor e com um mínimo de desperdício. Um dos outros objetivos é a utilização de mono ou multi-materiais poliméricos que possam ser substitutos adequados ao material que atualmente é utilizado na produção de protetores bocais, EVA – copolímero de Acetato- de Vinilo de Etileno.Os materiais que foram objeto de estudo desta dissertação foram o copolímero Acrilonitrilo Butadieno Estireno (ABS), Poliestireno de Elevado Impacto (HIPS), Poli(metil metacrilato) (PMMA) e a Poliuretana Termoplástica (TPU). A variação das propriedades mecânicas com a utilização dos dispositivos foi avaliada através do envelhecimento com uma solução de saliva artificial. As propriedades mecânicas foram avaliadas através de testes de Impacto Transversal e testes de Flexão em Três Pontos (Three Point Bending - 3PB) em provetes impressos em mono e em multi-material (estruturas sandwich), antes e após o processo de envelhecimento em saliva artificial. Os provetes testados foram impressos segundo as normas ASTM D790 e Charpy ISO179.
With the advances in technology, Additive Manufacturing (AM) processes have been gaining an increased importance in the world of industry when compared to other manufacturing processes. This is due to the fact that AM is able to produce parts and components with complex geometries unachievable by other technologies, while generating little or no waste during and after production. When compared to other manufacturing processes, AM uses less raw material, which lowers the production costs. The high importance of these factors overcome the drawbacks that are sometimes associated with the quality of the surface finish and geometry tolerance of printed parts. Nowadays, AM processes have the opportunity to have an impact in the manufacturing world, being a core element of the Industry 4.0.From every available process used in Additive Manufacturing, Fused Deposition Modelling (FDM), commonly known as 3D Printing, is the one which requires the less amount of equipment, and is one of the easiest, if not the easiest to use of all processes, and the equipment needed to produce parts, components or devices through this process are easily available. The main topic of this dissertation is the use of this technology with the aim of producing mouthguards for athletes. The current processes to create mouthguards do not produce devices with the adequate design and customized production has a high cost. The materials used have some hindrances associated with them, and the technologies used in the processing produce a high amount of waste. Through FDM processing, it is possible to obtain completely customized mouthguards with minimal waste. There is also the focus on the comprehension of which material and material combination will suit better this application, in order to have a reliable substitute for the current material employed in the production of mouthguards, which is EVA – copolymer of ethylene-vinyl acetate.The used materials were the copolymer Acrylonitrile-Butadiene-Styrene (ABS), High Impact Polystyrene (HIPS), Poly(methyl methacrylate) (PMMA) and Thermoplastic Polyurethane (TPU). The influence of the use on the mechanical properties was also evaluated through an aging process with an artificial saliva solution. For both mono and sandwich multi-material combinations, before and after the saliva influence, the mechanical properties were evaluated through Transverse Impact Testing and Flexural Testing (Three-Point Bending – 3PB). The tested specimens were printed according to the standards ASTM D790 and Charpy ISO179.

Indiana University-Purdue University Indianapolis (IUPUI)
Powder bed fusion (PBF) process is a metal additive manufacturing process, which can build parts with any complexity from a wide range of metallic materials. Research in the PBF process predominantly focuses on the impact of a few parameters on the ultimate properties of the printed part. The lack of a systematic approach to optimizing the process parameters for a better performance of given material results in a sub-optimal process limiting the potential of the application. This process needs a comprehensive study of all the influential parameters and their impact on the mechanical and microstructural properties of a fabricated part. Furthermore, there is a need to develop a quantitative system for mapping the material properties and process parameters with the ultimate quality of the fabricated part to achieve improvement in the manufacturing cycle as well as the quality of the final part produced by the PBF process. To address the aforementioned challenges, this research proposes a framework to optimize the process for 316L stainless steel material. This framework characterizes the influence of process parameters on the microstructure and mechanical properties of the fabricated part using a series of experiments. These experiments study the significance of process parameters and their variance as well as study the microstructure and mechanical properties of fabricated parts by conducting tensile, impact, hardness, surface roughness, and densification tests, and ultimately obtain the optimum range of parameters. This would result in a more complete understanding of the correlation between process parameters and part quality. Furthermore, the data acquired from the experiments are employed to develop an intelligent parameter suggestion multi-layer feedforward (FF) backpropagation (BP) artificial neural network (ANN). This network estimates the fabrication time and suggests the parameter setting accordingly to the user/manufacturers desired characteristics of the end-product. Further, research is in progress to evaluate the framework for assemblies and complex part designs and incorporate the results in the network for achieving process repeatability and consistency.

Powder bed fusion (PBF) process is a metal additive manufacturing process, which can build parts with any complexity from a wide range of metallic materials. Research in the PBF process predominantly focuses on the impact of a few parameters on the ultimate properties of the printed part. The lack of a systematic approach to optimizing the process parameters for a better performance of given material results in a sub-optimal process limiting the potentialof the application. This process needs a comprehensive study of all the influential parameters and their impact on the mechanical and microstructural properties of a fabricated part. Furthermore, there is a need to develop a quantitative system for mapping the material properties and process parameters with the ultimate quality of the fabricated part to achieve improvement in the manufacturing cycle as well as the quality of the final part produced by the PBF process. To address the aforementioned challenges, this research proposes a framework to optimize the process for 316L stainless steel material. This framework characterizes the influence of process parameters on the microstructure and mechanical properties of the fabricated part using a series of experiments. These experiments study the significance of process parameters and their variance as well as study the microstructure and mechanical properties of fabricated parts by conducting tensile, impact, hardness, surface roughness, and densification tests, and ultimately obtain the optimum range of parameters. This would result in a more complete understanding of the correlation between process parameters and part quality. Furthermore, the data acquired from the experimentsare employed to develop an intelligent parameter suggestion multi-layer feedforward (FF) backpropagation (BP) artificial neural network (ANN). This network estimates the fabrication time and suggests the parameter setting accordingly to the user/manufacturers desired characteristics of the end-product. Further, research is in progress to evaluate the framework for assemblies and complex part designs and incorporate the results in the network for achieving process repeatability and consistency.

abstract: Additive manufacturing (AM) has been extensively investigated in recent years to explore its application in a wide range of engineering functionalities, such as mechanical, acoustic, thermal, and electrical properties. The proposed study focuses on the data-driven approach to predict the mechanical properties of additively manufactured metals, specifically Ti-6Al-4V. Extensive data for Ti-6Al-4V using three different Powder Bed Fusion (PBF) additive manufacturing processes: Selective Laser Melting (SLM), Electron Beam Melting (EBM), and Direct Metal Laser Sintering (DMLS) are collected from the open literature. The data is used to develop models to estimate the mechanical properties of Ti-6Al-4V. For this purpose, two models are developed which relate the fabrication process parameters to the static and fatigue properties of the AM Ti-6Al-4V. To identify the behavior of the relationship between the input and output parameters, each of the models is developed on both linear multi-regression analysis and non-linear Artificial Neural Network (ANN) based on Bayesian regularization. Uncertainties associated with the performance prediction and sensitivity with respect to processing parameters are investigated. Extensive sensitivity studies are performed to identify the important factors for future optimal design. Some conclusions and future work are drawn based on the proposed study with investigated material.
Dissertation/Thesis
Masters Thesis Mechanical Engineering 2020

Dissertação de Mestrado Integrado em Engenharia Mecânica apresentada à Faculdade de Ciências e Tecnologia
Este trabalho pretende conjugar uma bordagem multidisciplinar ao incorporar duas áreas científicas distintas: a Engenharia Mecânica e a Medicina Dentária. O objetivo principal era avaliar a viabilidade de processar por fabricação aditiva, utilizando a tecnologia normalmente designada por impressão 3D, próteses totais mandibulares, utilizando diferentes materiais poliméricos. Deste modo pretendia-se estudar a substituição dos materiais cerâmicos atualmente utilizados por alternativas mais baratas, de maior facilidade de fabrico e cujas tecnologias envolvidas fossem mais sustentáveis do ponto de vista ambiental.O estudo foi iniciado pela caracterização química, térmica e mecânica dos filamentos de ABS, HIPS e PMMA. Após estudos preliminares para otimização dos parâmetros de impressão foram impressos provetes para a avaliação das propriedades mecânicas, nomeadamente da resistência ao impacto. A etapa seguinte consistiu na impressão das arcadas dentárias superiores e inferiores e avaliação macroscópica das mesmas. Quando comparadas com outras obtidas por fresagem e fabricação aditiva a partir de pós de polímero, as desenvolvidas no presente trabalho mostram um acabamento superior, apesar de a qualidade de impressão selecionada para o seu processamento ter sido a qualidade normal (altura de camada de 0,1 mm) e não a de elevada qualidade (altura de camada de 0,05 mm). O trabalho ficaria concluído pela avaliação do comportamento mecânico das próteses híbridas (arcadas com os implantes metálicos) em testes à compressão. Esta avaliação seria efetuada antes e após estudos de envelhecimento em saliva artificial. Esta última etapa não foi possível de ser efetuada devido aos constrangimentos impostos pela pandemia designada COVID-19.
This work aims to combine a multidisciplinary approach by incorporating two distinct scientific areas: Mechanical Engineering and Dentistry. The main objective was to evaluate the feasibility of processing by additive manufacturing, using the technology normally called 3D printing, mandibular total prostheses, using different polymeric materials. In this way it was intended to study the replacement of the ceramic materials currently used by cheaper alternatives, which are easier to manufacture and whose technologies involved were more sustainable from an environmental point of view.The study was initiated by the chemical, thermal and mechanical characterization of the ABS, HIPS and PMMA filaments. After preliminary studies to optimize the printing parameters, test pieces were printed for the evaluation of the mechanical properties, namely the impact resistance. The next step consisted of printing the upper and lower dental arches and macroscopic evaluation of them. When compared with others obtained by milling and additive manufacturing from polymer powders, those developed in the present work show a superior finish, although the print quality selected for processing was normal quality (layer height 0, 1 mm) and not high quality (layer height 0.05 mm). The work would be concluded by the evaluation of the mechanical behavior of hybrid prostheses (arches with metallic implants) in compression tests. This evaluation would be carried out before and after studies of aging in artificial saliva. This last step was not possible due to the constraints imposed by the pandemic designated as COVID-19.

Artificial neural networks, including deep neural networks, play a central role in data-driven science due to their superior learning capacity and adaptability to different tasks and data structures. However, although quantitative uncertainty analysis is essential for training and deploying reliable data-driven models, the uncertainties in neural networks are often overlooked or underestimated in many studies, mainly due to the lack of a high-fidelity and computationally efficient uncertainty quantification approach. In this work, a novel uncertainty analysis scheme is developed. The Gaussian mixture model is used to characterize the probability distributions of uncertainties in arbitrary forms, which yields higher fidelity than the presumed distribution forms, like Gaussian, when the underlying uncertainty is multimodal, and is more compact and efficient than large-scale Monte Carlo sampling. The fidelity of the Gaussian mixture is refined through adaptive scheduling of the width of each Gaussian component based on the active assessment of the factors that could deteriorate the uncertainty representation quality, such as the nonlinearity of activation functions in the neural network.

Following this idea, an adaptive Gaussian mixture scheme of nonlinear uncertainty propagation is proposed to effectively propagate the probability distributions of uncertainties through layers in deep neural networks or through time in recurrent neural networks. An adaptive Gaussian mixture filter (AGMF) is then designed based on this uncertainty propagation scheme. By approximating the dynamics of a highly nonlinear system with a feedforward neural network, the adaptive Gaussian mixture refinement is applied at both the state prediction and Bayesian update steps to closely track the distribution of unmeasurable states. As a result, this new AGMF exhibits state-of-the-art accuracy with a reasonable computational cost on highly nonlinear state estimation problems subject to high magnitudes of uncertainties. Next, a probabilistic neural network with Gaussian-mixture-distributed parameters (GM-PNN) is developed. The adaptive Gaussian mixture scheme is extended to refine intermediate layer states and ensure the fidelity of both linear and nonlinear transformations within the network so that the predictive distribution of output target can be inferred directly without sampling or approximation of integration. The derivatives of the loss function with respect to all the probabilistic parameters in this network are derived explicitly, and therefore, the GM-PNN can be easily trained with any backpropagation method to address practical data-driven problems subject to uncertainties.

The GM-PNN is applied to two data-driven condition monitoring schemes of manufacturing processes. For tool wear monitoring in the turning process, a systematic feature normalization and selection scheme is proposed for the engineering of optimal feature sets extracted from sensor signals. The predictive tool wear models are established using two methods, one is a type-2 fuzzy network for interval-type uncertainty quantification and the other is the GM-PNN for probabilistic uncertainty quantification. For porosity monitoring in laser additive manufacturing processes, convolutional neural network (CNN) is used to directly learn patterns from melt-pool patterns to predict porosity. The classical CNN models without consideration of uncertainty are compared with the CNN models in which GM-PNN is embedded as an uncertainty quantification module. For both monitoring schemes, experimental results show that the GM-PNN not only achieves higher prediction accuracies of process conditions than the classical models but also provides more effective uncertainty quantification to facilitate the process-level decision-making in the manufacturing environment.

Based on the developed uncertainty analysis methods and their proven successes in practical applications, some directions for future studies are suggested. Closed-loop control systems may be synthesized by combining the AGMF with data-driven controller design. The AGMF can also be extended from a state estimator to the parameter estimation problems in data-driven models. In addition, the GM-PNN scheme may be expanded to directly build more complicated models like convolutional or recurrent neural networks.

Purdue School of Engineering and Technology, Indianapolis
Powder Bed Fusion Additive Manufacturing enables the fabrication of metal parts with complex geometry and elaborates internal features, the simplification of the assembly process, and the reduction of development time. However, the lack of consistent quality hinders its tremendous potential for widespread application in industry. This limits its ability as a viable manufacturing process particularly in the aerospace and medical industries where high quality and repeatability are critical. A variety of defects, which may be initiated during the powder-bed fusion additive manufacturing process, compromise the repeatability, precision, and resulting mechanical properties of the final part. The literature review shows that a non-uniform temperature distribution throughout fabricated layers is a significant source of the majority of thermal defects. Therefore, the work introduces an online thermography methodology to study temperature distribution, thermal evolution, and thermal specifications of the fabricated layers in powder-bed fusion process or any other thermal inherent AM process. This methodology utilizes infrared technique and segmentation image processing to extract the required data about temperature distribution and HAZs of the layer under fabrication. We conducted some primary experiments in the FDM process to leverage the thermography technique and achieve a certain insight to be able to propose a technique to generate a more uniform temperature distribution. These experiments lead to proposing an innovative chessboard scanning strategy called tessellation algorithm, which can generate more uniform temperature distribution and diminish the layer warpage consequently especially throughout the layers with either geometry that is more complex or poses relatively longer dimensions. In the next step, this work develops a new technique in ABAQUS to verify the proposed scanning strategy. This technique simulates temperature distribution throughout a layer printed by chessboard printing patterns in powder-bed fusion process in a fraction of the time taken by current methods in the literature. This technique compares the temperature distribution throughout a designed layer printed by three presented chessboard-scanning patterns, namely, rastering pattern, helical pattern, and tessellation pattern. The results confirm that the tessellation pattern generates more uniform temperature distribution compared with the other two patterns. Further research is in progress to leverage the thermography methodology to verify the simulation technique. It is also pursuing a hybrid closed-loop online monitoring and control methodology, which bases on the introduced tessellation algorithm and online thermography in this work and Artificial Neural Networking (ANN) to generate the most possible uniform temperature distribution within a safe temperature range layer-by-layer.