Given that achieving nominal (all dimensions are theoretically perfect) geometry is challenging during building construction, understanding and anticipating sources of geometric variation through tolerances modeling and allocation is critical. However, existing building modeling environments lack the ability to support coordinated, incremental and systematic specification of manufacturing and construction requirements. This issue becomes evident when adding multi-material systems produced off site by different vendors during building erection. Current practices to improve this situation include costly and time-consuming operations that challenge the relationship among the stakeholders of a project. As one means to overcome this issue, this research proposes the development of a knowledge-aided modeling framework that integrates a parametric CAD tool with a system modeling application to assess variability in building construction. The CAD tool provides robust geometric modeling capabilities, while System Modeling allows for the specification of feature-based manufacturing requirements aligned with construction standards and construction processes know-how. The system facilitates the identification of conflicting interactions between tolerances and manufacturing specifications of building material systems. The expected contributions of this project are the representation of manufacturing knowledge and tolerances interaction across off-site building subsystems to identify conflicting manufacturing requirements and minimize costly construction errors. The proposed approach will store and allocate manufacturing knowledge as Model-Based Systems Engineering (MBSE) design specifications for both single and multiple material systems. Also, as new techniques in building design and construction are beginning to overlap with engineering methods and standards (e.g. in-factory prefabrication), this project seeks to create collaborative scenarios between MBSE and Building Information Modeling (BIM) based on parametric, simultaneous, software integration to reduce human-to-data translation errors, improving model consistency among domains. Important sub-stages of this project include the comprehensive review of modeling and allocation of tolerances and geometric deviations in design, construction and engineering; an approach for model integration among System Engineering models, mathematical engines and BIM (CAD) models; and finally, a demonstration computational implementation of a System-level tolerances modeling and allocation approach.

In pneumatic pressure control applications the influence of tubes that connect the valve with the control volume ist mainly neglected. This can lead to stability and robustness issues and limit either control performance or tube length. Modeling and considering tube behavior in controller design procedure allows longer tubes while maintaining the required performance and robustness properties without need for manual tuning. The author\'s previously published Simplified Fluid Transmission Line Model and the proposed model-based controller design enable the specification of a desired pressure trajectory in the control volume while the pressure sensor is mounted directly at the valve. Thus wiring effort is reduced as well as cost and the chance of cable break or sensor disturbance. In order to validate the simulated results the proposed control scheme is implemented on a real-time system and compared to a state-of-the-art pressure regulating valve

An efficient system development needs reuse, traceability and understanding. Today, specifications are usually written in text documents. Reuse means a copy and paste of suitable specifications. Traceability is the textual note that references to affected requirements. Achieving a full context understanding requires reading hundreds of pages in a variety of documents. Changing one textual requirement in complex systems can be very time-consuming. Model-based systems engineering (MBSE) addresses these issues. There, an integrated system model is used for the design, analysis, communication and system specification and shall contribute to handling the system complexity. This paper shows aspects of this approach in the development of a wheel loader\'s attachment system. Customer requirements will be used to derive a specification model. Based on this, the author introduces the system and software architecture. The connection between requirement and architecture leads to a traceable system design and produces the huge advantage of MBSE.

Building Information Modeling, or BIM, is a digital representation of physical and functional characteristics of a facility that serves as a shared resource for information for decision-making throughout the project lifecycle (National Institute of Building Sciences, 2007). The masonry construction industry currently suffers from the lack of BIM integration. Where other industries and trades have increased productivity by implementing standards for software-enhanced workflows, masonry construction has failed to adopt information tools and processes. New information technology and process modeling tools have grown in popularity and their use is helping to understand and improve construction processes. The Systems Modeling Language, or SysML, is one of the process modeling tools we can use to model and analyze the various processes and workflows. In this research, a case study methodology was applied to analyze the masonry construction industry to understand the current state of masonry construction processes and workflows. This thesis reviews these concepts and the applied case studies which are necessary to move forward with the implementation of BIM for masonry.

Mechanical ventilation (MV) is the primary form of therapeutic support for patients with acute respiratory failure (ARF) or acute respiratory distress syndrome (ARDS) until the underlying disease is resolved. However, as patient disease state and response to MV are highly variable, clinicians often rely on experience to set MV. The result is more variable care, as there are currently no standard approaches to MV settings. As a result of the common occurrence of MV and variability in care, MV is one of the most expensive treatments in critical care. Thus, an approach capable of guiding patient-specific MV is required and this approach could potentially save significant cost. This research focuses on developing models and model-based approaches to analyse and guide patient-specific MV care. Four models and metrics are developed, and each model is tested in experimental or clinical trials developed for the purpose. Each builds the understanding and methods necessary for an overall approach to guide MV in a wide range of patients. The first model, a minimal recruitment model, captures the recruitment of an injured lung and its response to positive end expiratory pressure (PEEP). However, the model was only previously validated in diagnosed ARDS patients, and was not proven to capture behaviours seen in healthy patients. This deficiency could potentially negate its ability to track disease state, which is crucial in providing rapid diagnosis and patient-specific MV in response to changes in patient condition. Hence, the lack of validation in disease state progression monitoring from ARDS to healthy, or vice-versa, severely limits its application in real-time monitoring and decision support. To address this issue, an experimental ARDS animal model is developed to validate the model across the transition between healthy and diseased states. The second model, a single compartment linear lung model, models the lung as a conducting airway connected to an elastic compartment. This model is used to estimate the respiratory mechanics (Elastance and Resistance) of an ARDS animal model during disease progression and recruitment manoeuvres. This model is later extended to capture high resolution, patient-specific time-varying respiratory mechanics during each breathing cycle. This extended model is tested in ARDS patients, and was used to titrate patient-specific PEEP using a minimum elastance metric that balances recruitment and the risk of lung overdistension and ventilation-induced injury. Studies have revealed that promoting patients to breathe spontaneously during MV can improve patient outcomes. Thus, there is significant clinical trend towards using partially assisted ventilation modes, rather than fully supported ventilation modes. In this study, the patient-ventilator interaction of a state of the art partially assisted ventilation mode, known as neurally adjusted ventilatory assist (NAVA), is investigated and compared with pressure support ventilation (PS). The matching of patient-specific inspiratory demand and ventilator supplied tidal volume for these two ventilation modes is assessed using a novel Range90 metric. NAVA consistently showed better matching than PS, indicating that NAVA has better ability to provide patient-specific ventilator tidal volume to match variable patient-specific demand. Hence, this new analysis highlights a critical benefit of partially assisted ventilation and thus the need to extend model-based methods to this patient group. NAVA ventilation has been shown to improve patient-ventilator interaction compared to conventional PS. However, the patient-specific, optimal NAVA level remains unknown, and the best described method to set NAVA is complicated and clinically impractical. The Range90 metric is thus extended to analyse the matching ability of different NAVA levels, where it is found that response to different NAVA levels is highly patient-specific. Similar to the fully sedated MV case, and thus requiring models and metrics to help titrate care. More importantly, Range90 is shown to provide an alternative metric to help titrate patient-specific optimal NAVA level and this analysis further highlights the need for extended model-based methods to better guide these emerging partially assisted MV modes. Traditionally, the respiratory mechanics of the spontaneously breathing (SB) patient cannot be estimated without significant additional invasive equipment and tests that interrupt normal care and are clinically intensive to carry out. Thus, respiratory mechanics and model-based methods are rarely used to guide partially assisted MV. Thus, there is significant clinical interest to use respiratory mechanics to guide MV in SB patients. The single compartment model is extended to effectively capture the trajectory of time-varying elastance for SB patients. Results show that without additional invasive equipment, the model was able estimate unique and clinically useful respiratory mechanics in SB patients. Hence, the extended single compartment model can be used as ‘a one model fits all’ means to guide patient-specific MV continuously and consistently, for all types of patient and ventilation modes, without interrupting care. Overall, the model-based approaches presented in this thesis are capable of capturing physiologically relevant patient-specific parameters, and thus, characterise patient disease state and response to MV. With additional, larger scale clinical trials to test the performance and the impact of model-based methods on clinical outcome, the models can aid clinicians to guide MV decision making in the heterogeneous ICU population. Hence, this thesis develops, extends and validates several fundamental model-based metrics, models and methods to enable standardized patient-specific MV to improve outcome and reduce the variability and cost of care.

The aim of this thesis is to determine the changes in BOLD signal of the human brain during various stages of reinforcement learning. In order to accomplish that goal two probabilistic reinforcement-learning tasks were developed and assessed with healthy participants by using functional magnetic resonance imaging (fMRI). For both experiments the brain imaging data of the participants were analysed by using a combination of univariate and model–based techniques. In Experiment 1 there were three types of stimulus-response pairs where they predict either a reward, a neutral or a monetary loss outcome with a certain probability. The Experiment 1 tested the following research questions: Where does the activity occur in the brain for expecting and receiving a monetary reward and a punishment ? Does avoiding a loss outcome activate similar brain regions as gain outcomes and vice a verse does avoiding a reward outcome activate similar brain regions as loss outcomes? Where in the brain prediction errors, and predictions for rewards and losses are calculated? What are the neural correlates of reward and loss predictions for reward and loss during early and late phases in learning? The results of the Experiment 1 have shown that expectation for reward and losses activate overlapping brain areas mainly in the anterior cingulate cortex and basal ganglia but outcomes of rewards and losses activate separate brain regions, outcomes of losses mainly activate insula and amygdala whereas reward activate bilateral medial frontal gyrus. The model-based analysis also revealed early versus late learning related changes. It was found that predicted-value in early trials is coded in the ventro-medial orbito frontal cortex but later in learning the activation for the predicted value was found in the putamen. The second experiment was designed to find out the differences in processing novel versus familiar reward-predictive stimuli. The results revealed that dorso-lateral prefrontal cortex and several regions in the parietal cortex showed greater activation for novel stimuli than for familiar stimuli. As an extension to the fourth research question of Experiment 1, reward predictedvalues of the conditional stimuli and prediction errors of unconditional stimuli were also assessed in Experiment 2. The results revealed that during learning there is a significant activation of the prediction error mainly in the ventral striatum with extension to various cortical regions but for familiar stimuli no prediction error activity was observed. Moreover, predicted values for novel stimuli activate mainly ventro-medial orbito frontal cortex and precuneus whereas the predicted value of familiar stimuli activates putamen. The results of Experiment 2 for the predictedvalues reviewed together with the early versus later predicted values in Experiment 1 suggest that during learning of CS-US pairs activation in the brain shifts from ventro-medial orbito frontal structures to sensori-motor parts of the striatum.

Recursive partitioning algorithms separate a feature space into a set of disjoint rectangles. Then, usually, a constant in every partition is fitted. While this is a simple and intuitive approach, it may still lack interpretability as to how a specific relationship between dependent and independent variables may look. Or it may be that a certain model is assumed or of interest and there is a number of candidate variables that may non-linearily give rise to different model parameter values. We present an approach that combines generalized linear models with recursive partitioning that offers enhanced interpretability of classical trees as well as providing an explorative way to assess a candidate variable's influence on a parametric model. This method conducts recursive partitioning of a the generalized linear model by (1) fitting the model to the data set, (2) testing for parameter instability over a set of partitioning variables, (3) splitting the data set with respect to the variable associated with the highest instability. The outcome is a tree where each terminal node is associated with a generalized linear model. We will show the methods versatility and suitability to gain additional insight into the relationship of dependent and independent variables by two examples, modelling voting behaviour and a failure model for debt amortization. / Series: Research Report Series / Department of Statistics and Mathematics

Legacy systems contain critical and complex business code that has been in use for a long time. This code is difficult to understand, maintain, and evolve, in large part due to crosscutting concerns: software system features, such as persistence, logging, and error handling, whose implementation is spread across multiple modules. Aspect-oriented techniques separate crosscutting concerns from the base code, using separate modules called aspects and, thus, simplifying the legacy code. Aspect mining techniques identify aspect candidates so that the legacy code can be refactored into aspects. This study investigated an automated aspect mining method in which a vector-space model clustering approach was used with model-based clustering. The vector-space model clustering approach has been researched for aspect mining using a number of different heuristic clustering methods and producing mixed results. Prior to this study, this model had not been researched with model-based algorithms, even though they have grown in popularity because they lend themselves to statistical analysis and show results that are as good as or better than heuristic clustering methods. This study investigated the effectiveness of model-based clustering for identifying aspects when compared against heuristic methods, such as k-means clustering and agglomerative hierarchical clustering, using six different vector-space models. The study's results indicated that model-based clustering can, in fact, be more effective than heuristic methods and showed good promise for aspect mining. In general, model-based algorithms performed better in not spreading the methods of the concerns across the multiple clusters but did not perform as well in not mixing multiple concerns in the same cluster. Model-based algorithms were also significantly better at partitioning the data such that, given an ordered list of clusters, fewer clusters and methods would need to be analyzed to find all the concerns. In addition, model-based algorithms automatically determined the optimal number of clusters, which was a great advantage over heuristic-based algorithms. Lastly, the study found that the new vector-space models performed better, relative to aspect mining, than previously defined vector-space models.

When building aircrafts, structural components must be designed for high strength, low cost, and easy assembly.To meet these conditions structural components are often based upon previous designs, even if a new component is developed.Refining previous designs can be a good way of preserving knowledge but can also limit the exploration of new design concepts. Currently the design process for structural aircraft components at SAAB is managed by design engineers. The design engineer is responsible for ensuring the design meets requirements from several different disciplines such as structural analysis, manufacturing, tool design, and assembly.Therefore, the design engineer needs to have good communication with all disciplines and an effective flow of information. The previous design is refined, it is then reviewed and approved by adjacent disciplines.Reviewing designs is an iterative process, and when several disciplines are involved it quickly becomes time consuming.Any time the design is altered it has to be reviewed once more by all disciplines to ensure the change is acceptable.So there is a need for further customizing the design concept to decrease the number of iterations when reviewing. Design Space Exploration DSE is a well known method to explore design alternatives before implementation and is used to find new concepts.This thesis investigates if DSE can be used to facilitate the design process of structural aircraft components and if it can support the flow of information between different disciplines.To find a suitable discipline to connect with design a prestudy is conducted, investigating what information affect structural design and how it is managed.The information flow is concluded in a schematic diagram where structural analysis is chosen as additional discipline. By using topology optimization in a DSE, design and structural analysis are connected.The design space can be explored with regards to structural constraints.The thesis highlights the possibilities of using DSE with topology optimization for developing structural components and proposes a method for including it in the design process.

Nesta tese, com a ambição de desenvolver um pacote de controle que poderia ser implementado a baixo custo nas indústrias brasileiras, é estudado o controle preditivo (MPC) de sistemas estáveis e instáveis com modelos incertos, baseado em um modelo de realinhamento e o uso de técnicas de desigualdades matriciais lineares (LMI), para resolução de problemas de controle robusto. Na primeira parte da tese, a aplicabilidade de um controlador baseado em um modelo de realinhamento é considerada. Como o controlador preditivo baseado em um modelo de realinhamento não requer o uso de um observador de estados, espera-se que seja mais eficiente e mais robusto a distúrbios não medidos que um controlador que precise de um observador de estados. Essa hipótese é testada comparando-se o desempenho e a robustez a distúrbios não medidos deste controlador com um controlador que requer o uso de um observador de estados, através da simulação do controle de um separador polipropileno/propano (PP) industrial. Por outro lado, a desvantagem de um controlador baseado em um modelo de realinhamento é a sensibilidade da construção do modelo e das matrizes do controlador a erros numéricos que aumentam com o tamanho do sistema. As etapas do algoritmo de controle mais sensíveis a erros numéricos são então destacadas com o objetivo de discutir a aplicabilidade de um controlador baseado em um modelo de realinhamento a sistemas de grande porte. Além disso, sempre que possível, os métodos utilizados para reduzir a sensibilidade das etapas problemáticas a erros numéricos são apresentados. A segunda parte da tese trata da solução do problema de controle robusto preditivo baseado em um modelo de sistema incerto. Este problema é usualmente abordado através da inclusão de restrições não lineares sobre os custos associados aos modelos da planta no problema de controle, de tal modo que a ação de controle é obtida a cada instante de tempo através da resolução de um problema de otimização não linear, cujo custo computacional pode ser proibitivo para sistemas de grande porte. Nesta parte da tese, o problema de MPC robusto é então reformulado como um problema baseado em LMIs, que pode ser resolvido com uma fração do esforço computacional. A abordagem proposta é comparada com o MPC robusto convencional e testada através da simulação do controle de um reator e de um separador C3/C4 industriais. Finalmente, na terceira parte da tese, o MPC de sistemas estáveis e instáveis com incertezas no modelo que permite a saturação das entradas manipuladas é abordado. Um subconjunto das entradas manipuladas é alocado ao controle das saídas instáveis através de uma lei de realimentação de estados, enquanto as outras entradas são deixadas livres para controlar as saídas estáveis restantes. Assume-se que as saídas são controladas dentro de faixas e que os setpoints das saídas são tratados como entradas manipuladas adicionais. Os controladores desenvolvidos nesta parte da tese permitem a saturação das entradas associadas às saídas instáveis através da manipulação das entradas livres e dos setpoints das saídas instáveis. A viabilidade recursiva dos controladores desenvolvidos é garantida, permitindo-se que os setpoints das saídas instáveis deixem temporariamente suas faixas. O desempenho dos controladores desenvolvidos é testado através da simulação do controle de dois processos integradores da indústria química. / In this thesis, with the ambition of developing a homemade control package that could be implemented at a low cost in Brazilian industries, the model predictive control (MPC) of uncertain stable and unstable systems based on a realigned state space representation and on the use of the Linear Matrix Inequality (LMI) techniques is addressed. In the first part of this thesis, the practical implementability of a controller based on a realigned model is considered. Since a Model Predictive Controller based on a realigned model does not require the use of a state observer, it is expected to be more efficient and more robust to unmeasured disturbances than a controller requiring the use of a state observer. This assumption is tested by comparing the performance and the robustness to unmeasured disturbances of a controller based on a realigned model with a controller requiring the use of a state observer through the simulation of the control of a nonlinear industrial propylene/propane (PP) splitter. On the other hand, a disadvantage of a controller based on a realigned model is that the construction of the model and controller matrices is very sensitive to numerical errors. The steps of the control algorithm that are more sensitive to numerical errors are then highlighted in order to discuss the practical applicability of an MPC based on the realigned model to large-scale systems. Also, wherever possible, some ways of decreasing the sensitiveness of the problematic steps to numerical errors are presented. The second part of this thesis addresses the solution to the problem of robust MPC of systems with model uncertainty. The usual approach of dealing with this kind of problem is through the inclusion of nonlinear cost constraints in the control problem so that the control action is obtained at each sampling time as the solution to a nonlinear programming (NLP) problem that, for high order systems, can be computationally expensive. In this part of the thesis, the robust MPC problem is then recast as an LMI problem that can be solved in real time with a fraction of the computer effort. The proposed approach is compared with the conventional robust MPC and tested through the simulation of the control of a reactor system and a C3/C4 splitter of the process industry. Finally, in the third part of the thesis, the model predictive control of uncertain process systems with stable and unstable outputs that allows input saturation is addressed. A subset of the manipulated inputs is allocated to the control of the unstable outputs through a state feedback control law while the other inputs are left free to control the remaining stable outputs. It is assumed that the outputs are controlled inside zones and that the output setpoints are treated as additional free manipulated inputs. The proposed controllers allow the saturation of the inputs related to the unstable outputs by manipulating the free inputs and the setpoints of the unstable outputs. The recursive feasibility of the controllers is guaranteed by allowing the setpoints of the unstable outputs to temporarily leave their bounds. The performance of the proposed approach is tested through the simulation of the control of two integrating chemical process systems.