L'inférence statistique est un domaine complexe et en constante évolution. Ce mémoire traitera de l'inférence sur la fonction de densité d'une variable aléatoire. Nous partirons de plusieurs résultats connus et développerons une analyse de ces résultats dans le cadre paramétrique avec une approche bayésienne. Nous nous aventurerons aussi dans les problèmes avec espace paramétrique restreint. L'objectif du travail est de trouver les meilleurs estimateurs possibles considérant l'information a priori et l'observation de variables tirées d'une densité faisant intervenir le paramètre. Le chapitre 1 traitera de notions d'inférence bayésienne, de choix de perte évaluant la performance d'un estimateur et possédant des propriétés recherchées. Le chapitre 2 concernera l'estimation ponctuelle du paramètre. En particulier, nous aborderons l'estimateur de James-Stein et trouverons des conditions suffisantes pour la minimaxité et la dominance d'estimateurs en remarquant la forme particulière de ceux-ci. Une condition remontera même à la loi a priori utilisée. Le chapitre 3 établira des liens entre l'estimation ponctuelle et l'estimation par densité prédictive pour le cas multinormal. Des conditions seront aussi établies pour la minimaxité et la dominance. Nous comparerons nos estimateurs à l'estimateur de Bayes découlant d'une loi a priori non informative et démontrerons les résultats par des exemples. Le chapitre 4 considérera le problème dans un cadre plus général où le paramètre d'intérêt pourra être un paramètre de position ou d'échelle. Des liens entre ces deux problèmes seront énoncés et nous trouverons des conditions sur la famille de densités étudiée pour trouver des estimateurs minimax. Quelques exemples concluront cette section. Finalement, le chapitre 5 est l'intégrale de l'article déposé en collaboration avec Tatsuya Kubokawa, Éric Marchand et William E. Strawderman, concernant l'ensemble du problème étudié dans ce mémoire, à savoir l'estimation par densité prédictive dans un espace paramétrique restreint.

There are existing standards and guidelines for the effective management of infrastructure through infrastructure asset management planning (IAM). However, few if any of these standards explicitly address the financial implications associated with the uncertainty that underlies the risk associated with service provision. Without credibly quantifying the potential implications of this network variable uncertainty (i.e. an extreme weather event that affects the performance and costs of many segments within the study network, or the introduction of a new technology that may impact the network cost estimates) infrastructure management systems may actually regularly and significantly over or under estimate the actual financial requirements required to provide services. Therefore, financial projections may actually include a systematic bias. It was hypothesized that a model could be developed that quantifies and communicates the financial implications of network variable uncertainty within the IAM context. A model was developed to demonstrate how network variable uncertainty could be included in financial planning for infrastructure networks. The model was able to: (1) be applied to various types of infrastructure networks, (2) incorporate network variable uncertainty, (3) compare alternatives and scenarios, and (4) support effective communication of results. The outputs of the model were the average network annual worth (AW) and network present worth (PW). These outputs, along with tornado plots, risks curves, level of service dashboards, and existing budget levels, were used to communicate the impacts of the network variable uncertainty on the financial projections. The model was developed using Excel tools linked to DPL software to utilize probabilistic methods. The Life Cycle Cost (LCC) portion of the model was successfully verified against an existing infrastructure costing tool, the Land and Infrastructure Resiliency Assessment (LIRA) tool developed by the Agri-Environmental Services Branch of Agriculture and Agri-Food Canada. The impact of the network variable uncertainty within the variables was also quantified in terms of levels of service provided by the organization. The developed model was first applied to a hypothetical twelve segment road network for illustrative purposes. For the hypothetical road network there were four events, representing network variable uncertainty, that were considered. These decisions or events included the: (1) decision to implement a new technology, (2) event of changing standards, (3) event of increased material costs, and (4) occurrence of an extreme rainfall event. The hypothetical network illustrated that if the defined decisions or events occurred then the expected network AW would increase by 41%. The impacts of decisions or events on the hypothetical network levels of service, stemming from network variable uncertainty, were also considered. The measured levels of service for the hypothetical network included the network financial sustainability indicator (an indicator reflecting the network current budget divided by the network annual worth as a percentage) and the frequency of blading of the roads. The model was next applied to a case study using the Town of Shellbrook sanitary main network. The Town has a large quantity of aging mains which were constructed in the 1960’s and are expected to require renewal in the near term. The network variable uncertainty for the case study resulted from the potential decision to implement a new trenchless technology for the renewal of sanitary mains. The new technology was expected to decrease the renewal costs. However, there was uncertainty as to what percentage of the sanitary mains would be found to be suitable for the new technology. Using the model it was determined that if the decision was made to implement the new technology, there would be an expected reduction of 17% in the network AW. The levels of service that were used for the Shellbrook case study were the network financial sustainability indicator (annual budget / network AW) and the meeting of standards set by regulating bodies. It was determined that the network financial sustainability indicator was sensitive to the decision to implement the trenchless technology, while the meeting of regulating bodies was not. If the decision was made to implement the new technology the network sustainability indicator would be expected to increase from 28% (if the new technology was not implemented) to 34% (if the new technology were implemented). The model was finally applied to a case study looking at the RM of Wilton gravel road network. The network variable uncertainty for this case study resulted from the potential increase in gravel material costs. The network variable uncertainty represented the magnitude of the annual increase in gravel costs. Given the event of increasing gravel costs the expected network AW would increase by 14%. The levels of service indicators used for the RM of Wilton case study were the network financial sustainability indicator and the frequency of blading. It was determined that the network financial sustainability indicator was sensitive to the event (increasing gravel costs), while the frequency of blading was not directly impacted (although it may be indirectly impacted). If the event of increasing gravel costs were to occur then the network financial sustainability indicator would be expected to decrease from 59% (if gravel costs did not increase) to 52% (if gravel costs did increase). This research proved that the hypothesis was correct, and that a model could be developed that quantified and communicated the financial implications and level of service impacts of network variable uncertainty for IAM planning. This research illustrated and quantified that IAM planning without accounting for network variable uncertainty, such as: (1) changing technology, (2) changing standards, (3) increasing material costs, and (4) extreme weather events, managers may introduce a systematic bias into long term planning. Network variable uncertainty can significantly impact the projected expenditures required for the long term provision of services. Infrastructure managers and decision makers need to manage infrastructure in a sustainable way over the long term in the face of uncertainty. It is necessary that decision makers have information regarding the impacts of network variable uncertainty on both LCCs and levels of service to make fully informed decision.

No description available.

Intermittent power sources such as wind and solar are increasingly penetrating electrical grids, mainly motivated by global warming concerns and government policies. These intermittent and non-dispatchable sources of power affect the operation and control of the power system because of the uncertainties associated with their output power. Depending on the penetration level of intermittent sources of power, the electric grid may experience considerable changes in power flows and synchronizing torques associated with system stability, because of the variability of the power injections, among several other factors. Thus, adequate and efficient techniques are required to properly analyze the system stability under such uncertainties. A variety of methods are available in the literature to perform power flow, transient, and voltage stability analyses considering uncertainties associated with electrical parameters. Some of these methods are computationally inefficient and require assumptions regarding the probability density functions (pdfs) of the uncertain variables that may be unrealistic in some cases. Thus, this thesis proposes computationally efficient Affine Arithmetic (AA)-based approaches for voltage and transient stability assessment of power systems, considering uncertainties associated with power injections due to intermittent sources of power. In the proposed AA-based methods, the estimation of the output power of the intermittent sources and their associated uncertainty are modeled as intervals, without any need for assumptions regarding pdfs. This is a more desirable characteristic when dealing with intermittent sources of power, since the pdfs of the output power depends on the planning horizon and prediction method, among several other factors. The proposed AA-based approaches take into account the correlations among variables, thus avoiding error explosions attributed to other self-validated techniques such as Interval Arithmetic (IA).

In this research, a multi-agent approach was applied to the design of a large axial flow compressor in order to optimise performance and to greatly enlarge the useful operating range of the machine. In this design a number of distributed software/hardware agents co-operate to control the internal geometry of the machine and thereby optimise the compressor characteristics in response to changes in flow conditions. The resulting machine is termed an ‘Intelligent Geometry Compressor’ (IGC). The design of a multi-agent system for the IGC was carried out in three main phases, each supported by computer simulation. In the first phase a steady-state model of the IGC was developed in which global control of the variable geometry is achieved by a single agent. This was used to help identify specific requirements for performance and the underlying parametric relationships. The subsequent phases incorporated additional agents into the machine design to meet these requirements. Initially, agents were deployed to optimise the settings of individual rows of stator vanes. In the final phase, the MAS was extended to incorporate agents into the machine design for the control of individual stator vanes. Simulation results were obtained which demonstrate the effectiveness of the intelligent geometry compressor in achieving delivery pressure regulation over a wide range of steady-state operating conditions whilst optimising overall machine efficiency and avoiding the occurrence of stall. Some of the implications for the physical design of an IGC arising from the MAS concept were briefly considered. The experience of the research supported by the specific results and observations from many simulation trials, led to the conclusion that multi-agent systems can provide an effective and novel alternative approach to the design of an intelligent geometry compressor. By implication, this conclusion may be extended to other intelligent machine applications where similar opportunity to apply a distributed control solution exists.

No description available.

No description available.

Variable Selection has always been a very important problem in statistics. We often meet situations where a huge data set is given and we want to find out the relationship between the response and the corresponding variables. With a huge number of variables, we often end up with a big model even if we delete those that are insignificant. There are two reasons why we are unsatisfied with a final model with too many variables. The first reason is the prediction accuracy. Though the prediction bias might be small under a big model, the variance is usually very high. The second reason is interpretation. With a large number of variables in the model, it's hard to determine a clear relationship and explain the effects of variables we are interested in. A lot of variable selection methods have been proposed. However, one disadvantage of variable selection is that different sizes of model require different tuning parameters in the analysis, which is hard to choose for non-statisticians. Xin and Zhu advocate variable ranking instead of variable selection. Once variables are ranked properly, we can make the selection by adopting a threshold rule. In this thesis, we try to rank the variables using Least Angle Regression (LARS). Some shrinkage methods like Lasso and LARS can shrink the coefficients to zero. The advantage of this kind of methods is that they can give a solution path which describes the order that variables enter the model. This provides an intuitive way to rank variables based on the path. However, Lasso can sometimes be difficult to apply to variable ranking directly. This is because that in a Lasso solution path, variables might enter the model and then get dropped. This dropping issue makes it hard to rank based on the order of entrance. However, LARS, which is a modified version of Lasso, doesn't have this problem. We'll make use of this property and rank variables using LARS solution path.

Heart disease is attributed as the highest cause of death in the world. Although this could be alleviated by heart transplantation, there is a chronic shortage of donor hearts and so mechanical solutions are being considered. Currently, many Ventricular Assist Devices (VADs) are being developed worldwide in an effort to increase life expectancy and quality of life for end stage heart failure patients. Current pre-clinical testing methods for VADs involve laboratory testing using Mock Circulation Loops (MCLs), and in vivo testing in animal models. The research and development of highly accurate MCLs is vital to the continuous improvement of VAD performance. The first objective of this study was to develop and validate a mathematical model of a MCL. This model could then be used in the design and construction of a variable compliance chamber to improve the performance of an existing MCL as well as form the basis for a new miniaturised MCL. An extensive review of literature was carried out on MCLs and mathematical modelling of their function. A mathematical model of a MCL was then created in the MATLAB/SIMULINK environment. This model included variable features such as resistance, fluid inertia and volumes (resulting from the pipe lengths and diameters); compliance of Windkessel chambers, atria and ventricles; density of both fluid and compressed air applied to the system; gravitational effects on vertical columns of fluid; and accurately modelled actuators controlling the ventricle contraction. This model was then validated using the physical properties and pressure and flow traces produced from a previously developed MCL. A variable compliance chamber was designed to reproduce parameters determined by the mathematical model. The function of the variability was achieved by controlling the transmural pressure across a diaphragm to alter the compliance of the system. An initial prototype was tested in a previously developed MCL, and a variable level of arterial compliance was successfully produced; however, the complete range of compliance values required for accurate physiological representation was not able to be produced with this initial design. The mathematical model was then used to design a smaller physical mock circulation loop, with the tubing sizes adjusted to produce accurate pressure and flow traces whilst having an appropriate frequency response characteristic. The development of the mathematical model greatly assisted the general design of an in vitro cardiovascular device test rig, while the variable compliance chamber allowed simple and real-time manipulation of MCL compliance to allow accurate transition between a variety of physiological conditions. The newly developed MCL produced an accurate design of a mechanical representation of the human circulatory system for in vitro cardiovascular device testing and education purposes. The continued improvement of VAD test rigs is essential if VAD design is to improve, and hence improve quality of life and life expectancy for heart failure patients.

Indonesia has been implementing methadone maintenance treatment (MMT) since January 2003 as a strategy to minimize HIV transmission among injecting drug users (IDU). Previous studies have shown the effectiveness of the program and also showed that the program had attracted many IDU to participate. However, the dropout rate, particularly in Jakarta clinics, was relatively high. The first aim of this study was to investigate the MMT retention rate and its predictive variables. The second aim was to examine the effects of remaining in the program on treatment outcomes. A six-month longitudinal prospective cohort study was conducted at the client level and a cross-sectional survey was carried out at the clinic level. Information from this study provides significant inputs for developing drug treatment policy and improving its quality of service in Indonesia. It also contributes to a better understanding of the substitution treatment implementation in Indonesia. The average 3-month treatment retention rate was 74.2 percent and the 6-month retention rate was 61.3 percent. There was no significant difference in retention rates between clinics. Significant predictors of treatment retention in MMT in Indonesia were size of dose, the interaction between take-home dose and clinic experience, age of participant, participant’s belief towards the program and perceived accessibility, while a variable representing perceived peer support unexpectedly predicted an increased likelihood of prematurely leaving the treatment. This study showed a marked reduction in the use of heroin and depression status and a significant improvement of self-efficacy at the follow up times among participants who continued in treatment. There were no significant differences in criminal involvement and physical health status between those who remained in treatment and the treatment dropouts in both follow-up interviews. Nevertheless, there was a significant improvement in physical health from baseline to follow-up in both groups. The study concluded that retention rates of MMT in Indonesia were comparable to those of similar programs in other countries. As previously reported in other settings, dose was the primary predictor of treatment retention in Indonesia. A policy of providing take-home doses, prescribed in experienced clinics,was also found to be a significant predictor of remaining in treatment. Further research, however, is still needed to explain some of the unexpected observations. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1522114 / Thesis (Ph.D.) -- University of Adelaide, School of Medical Sciences, 2010