Indiana University-Purdue University Indianapolis (IUPUI) / This thesis studies critical zones of automated vehicles. The goal is for the automated vehicle to complete a car-following or lane change maneuver without collision. For instance, the automated vehicle should be able to indicate its interest in changing lanes and plan how the maneuver will occur by using model predictive control theory, in addition to the autonomous vehicle toolbox in Matlab. A test bench (that includes a scenario creator, motion logic and planner, sensors, and radars) is created and used to calculate the parameters of a critical zone. After a trajectory has been planned, the automated vehicle then attempts the car following or lane change while constantly ensuring its safety to continue on this path. If at any point, the lead vehicle brakes or a trailing vehicle accelerates, the automated vehicle makes the decision to either brake, accelerate, or abandon the lane change.

Artificial intelligence (AI) and predictive modeling are tools used to diagnose a disease, determine how much a home is worth, estimate insurance risks, and detect fraud. AI and predictive modeling are so ubiquitous that they can be why one gets spam and why spam is automatically deleted. Information science integrates interdisciplinary elements of data-driven, behavioral, design, interpretive, and analytical research methodologies to design and understand interactions between digital media, information systems, and humans. This research focuses on the interaction between humans, AI, and predictive models. This research proposes a theoretical framework and a conceptual research model to understand the antecedents to reliance on AI and predictive modeling. The dissertation follows a traditional format that includes three studies. Study 1 employed a deductive quantitative research approach as a survey to model the relationship between trust in science and reliance on formal news sources. Study 2 employed a deductive quantitative research approach as a survey to understand the impact of framing questions and consider an alternative method of measuring society's reliance on science using predictive models. Study 3 employed a deductive quantitative research approach in the form of a survey to posit a new model based on the first two studies. This study benefited from a Toulouse Graduate School grant to fund research using the crowdsourcing platform https://lucidtheorem.com/ to generate a stratified sample of the U.S. population.

No description available.

Magnetic Bearing Systems have been receiving a great deal of research attention for the past decades. Its inherent nonlinearity and open-loop instability are challenges for controller design. This thesis investigates and designs model predictive control strategy for an experimental Active Magnetic Bearing (AMB) laboratory system. A host-target development environment of real-time control system with hardware in the loop (HIL) is implemented. In this thesis, both continuous and discrete time model predictive controllers are studied. In the first stage, local MPC controllers are applied to control the AMB system; and in the second stage, concept of supervisory controller design is then investigated and implemented. Contributions of the thesis can be summarized as follows; 1. A Discrete time Model Predictive Controller has been developed and applied to the active magnetic bearing system. 2. A Continuous time Model Predictive Controller has been developed and applied to the active magnetic bearing system. 3. A frequency domain identification method using FSF has been applied to pursue model identification with respect to local MPC and magnetic bearing system. 4. A supervisory control strategy has been applied to pursue a two stages model predictive control of active magnetic bearing system.

Modern technologies coupled with advanced research have allowed model predictive control (MPC) to be applied to new and often experimental systems. The purpose of this research is to develop a model predictive control algorithm for tip-jet reaction drive system. This system's faster dynamics require an extremely short sampling rate, on the order of 20ms, and its slower dynamics require a longer prediction horizon. This coupled with the fact that the tip-jet reaction drive system has multiple control inputs makes the integration of an online MPC algorithm challenging. In order to apply a model predictive control to the system in question, an algorithm is proposed that combines multiplexed inputs and a feasible cooperative MPC algorithm. In the proposed algorithm, it is hypothesized that the computational burden will be reduced from approximately Hp(Nu + Nx)3 to pHp(Nx+1)3 while maintaining control performance similar to that of a centralized MPC algorithm. To capture the performance capability of the proposed controller, a comparison its performance to that of a multivariable proportional-integral (PI) controller and a centralized MPC is executed. The sensitivity of the proposed MPC to various design variables is also explored. In terms of bandwidth, interactions, and disturbance rejection, the proposed MPC was very similar to that of a centralized MPC or PI controller. Additionally in regards to sensitivity to modeling error, there is not a noticeable difference between the two MPC controllers. Although the constraints are handled adequately for the proposed controller, adjustments can be made in the design and sizing process to improve the constraint handling, so that it is more comparable to that of the centralized MPC. Given these observations, the hypothesis of the dissertation has been confirmed. The proposed MPC does in fact reduce computational burden while maintaining close to centralized MPC performance.

An increasing number of applications ranging from multi-vehicle systems, large-scale process control systems, transportation systems to smart grids call for the development of cooperative control theory. Meanwhile, when designing the cooperative controller, the state and control constraints, ubiquitously existing in the physical system, have to be respected. Model predictive control (MPC) is one of a few techniques that can explicitly and systematically handle the state and control constraints. This dissertation studies the robust MPC and distributed MPC strategies, respectively. Specifically, the problems we investigate are: the robust MPC for linear or nonlinear systems, distributed MPC for constrained decoupled systems and distributed MPC for constrained nonlinear systems with coupled system dynamics. In the robust MPC controller design, three sub-problems are considered. Firstly, a computationally efficient multi-stage suboptimal MPC strategy is designed by exploiting the j-step admissible sets, where the j-step admissible set is the set of system states that can be steered to the maximum positively invariant set in j control steps. Secondly, for nonlinear systems with control constraints and external disturbances, a novel robust constrained MPC strategy is designed, where the cost function is in a non-squared form. Sufficient conditions for the recursive feasibility and robust stability are established, respectively. Finally, by exploiting the contracting dynamics of a certain type of nonlinear systems, a less conservative robust constrained MPC method is designed. Compared to robust MPC strategies based on Lipschitz continuity, the strategy employed has the following advantages: 1) it can tolerate larger disturbances; and 2) it is feasible for a larger prediction horizon and enlarges the feasible region accordingly. For the distributed MPC of constrained continuous-time nonlinear decoupled systems, the cooperation among each subsystems is realized by incorporating a coupling term in the cost function. To handle the effect of the disturbances, a robust control strategy is designed based on the two-layer invariant set. Provided that the initial state is feasible and the disturbance is bounded by a certain level, the recursive feasibility of the optimization is guaranteed by appropriately tuning the design parameters. Sufficient conditions are given ensuring that the states of each subsystem converge to the robust positively invariant set. Furthermore, a conceptually less conservative algorithm is proposed by exploiting the controllability set instead of the positively invariant set, which allows the adoption of a shorter prediction horizon and tolerates a larger disturbance level. For the distributed MPC of a large-scale system that consists of several dynamically coupled nonlinear systems with decoupled control constraints and disturbances, the dynamic couplings and the disturbances are accommodated through imposing new robustness constraints in the local optimizations. Relationships among, and design procedures for the parameters involved in the proposed distributed MPC are derived to guarantee the recursive feasibility and the robust stability of the overall system. It is shown that, for a given bound on the disturbances, the recursive feasibility is guaranteed if the sampling interval is properly chosen. / Graduate / 0548 / 0544 / 0546 / liuxiaotao1982@gmail.com

L'augmentation des données disponibles dans presque tous les domaines soulève la nécessité d'utiliser des algorithmes pour l'analyse automatisée des données. Cette nécessité est mise en évidence dans la maintenance prédictive, où l'objectif est de prédire les pannes des systèmes en observant continuellement leur état, afin de planifier les actions de maintenance à l'avance. Ces observations sont générées par des systèmes de surveillance habituellement sous la forme de séries temporelles et de journaux d'événements et couvrent la durée de vie des composants correspondants. Le principal défi de la maintenance prédictive est l'analyse de l'historique d'observation afin de développer des modèles prédictifs.Dans ce sens, l'apprentissage automatique est devenu omniprésent puisqu'il fournit les moyens d'extraire les connaissances d'une grande variété de sources de données avec une intervention humaine minimale. L'objectif de cette thèse est d'étudier et de résoudre les problèmes dans l'aviation liés à la prévision des pannes de composants à bord. La quantité de données liées à l'exploitation des avions est énorme et, par conséquent, l'évolutivité est une condition essentielle dans chaque approche proposée.Cette thèse est divisée en trois parties qui correspondent aux différentes sources de données que nous avons rencontrées au cours de notre travail. Dans la première partie, nous avons ciblé le problème de la prédiction des pannes des systèmes, compte tenu de l'historique des Post Flight Reports. Nous avons proposé une approche statistique basée sur la régression précédée d'une formulation méticuleuse et d'un prétraitement / transformation de données. Notre méthode estime le risque d'échec avec une solution évolutive, déployée dans un environnement de cluster en apprentissage et en déploiement. À notre connaissance, il n'y a pas de méthode disponible pour résoudre ce problème jusqu'au moment où cette thèse a été écrite.La deuxième partie consiste à analyser les données du livre de bord, qui consistent en un texte décrivant les problèmes d'avions et les actions de maintenance correspondantes. Le livre de bord contient des informations qui ne sont pas présentes dans les Post Flight Reports bien qu'elles soient essentielles dans plusieurs applications, comme la prédiction de l'échec. Cependant, le journal de bord contient du texte écrit par des humains, il contient beaucoup de bruit qui doit être supprimé afin d'extraire les informations utiles. Nous avons abordé ce problème en proposant une approche basée sur des représentations vectorielles de mots. Notre approche exploite des similitudes sémantiques, apprises par des neural networks qui ont généré les représentations vectorielles, afin d'identifier et de corriger les fautes d'orthographe et les abréviations. Enfin, des mots-clés importants sont extraits à l'aide du Part of Speech Tagging.Dans la troisième partie, nous avons abordé le problème de l'évaluation de l'état des composants à bord en utilisant les mesures des capteurs. Dans les cas considérés, l'état du composant est évalué par l'ampleur de la fluctuation du capteur et une tendance à l'augmentation monotone. Dans notre approche, nous avons formulé un problème de décomposition des séries temporelles afin de séparer les fluctuations de la tendance en résolvant un problème convexe. Pour quantifier l'état du composant, nous calculons à l'aide de Gaussian Mixture Models une fonction de risque qui mesure l'écart du capteur par rapport à son comportement normal. / The increase of available data in almost every domain raises the necessity of employing algorithms for automated data analysis. This necessity is highlighted in predictive maintenance, where the ultimate objective is to predict failures of hardware components by continuously observing their status, in order to plan maintenance actions well in advance. These observations are generated by monitoring systems usually in the form of time series and event logs and cover the lifespan of the corresponding components. Analyzing this history of observation in order to develop predictive models is the main challenge of data driven predictive maintenance.Towards this direction, Machine Learning has become ubiquitous since it provides the means of extracting knowledge from a variety of data sources with the minimum human intervention. The goal of this dissertation is to study and address challenging problems in aviation related to predicting failures of components on-board. The amount of data related to the operation of aircraft is enormous and therefore, scalability is a key requirement in every proposed approach.This dissertation is divided in three main parts that correspond to the different data sources that we encountered during our work. In the first part, we targeted the problem of predicting system failures, given the history of Post Flight Reports. We proposed a regression-based approach preceded by a meticulous formulation and data pre-processing/transformation. Our method approximates the risk of failure with a scalable solution, deployed in a cluster environment both in training and testing. To our knowledge, there is no available method for tackling this problem until the time this thesis was written.The second part consists analyzing logbook data, which consist of text describing aircraft issues and the corresponding maintenance actions and it is written by maintenance engineers. The logbook contains information that is not reflected in the post-flight reports and it is very essential in several applications, including failure prediction. However, since the logbook contains text written by humans, it contains a lot of noise that needs to be removed in order to extract useful information. We tackled this problem by proposing an approach based on vector representations of words (or word embeddings). Our approach exploits semantic similarities of words, learned by neural networks that generated the vector representations, in order to identify and correct spelling mistakes and abbreviations. Finally, important keywords are extracted using Part of Speech Tagging.In the third part, we tackled the problem of assessing the health of components on-board using sensor measurements. In the cases under consideration, the condition of the component is assessed by the magnitude of the sensor's fluctuation and a monotonically increasing trend. In our approach, we formulated a time series decomposition problem in order to separate the fluctuation from the trend by solving a convex program. To quantify the condition of the component, we compute a risk function which measures the sensor's deviation from it's normal behavior, which is learned using Gaussian Mixture Models.

Youth offending is an increasingly major problem in many countries and cultures. Several theories imply that a subset of young people display delinquent behaviour at a young age and go on to have an extensive and serious criminal career. Recently, there has been interest in the literature in identifying these young people early on and carrying out interventions in order to deter them from a criminal career. Many studies have examined the development and usefulness of actuarial measures of risk of future violence or recidivism in adult offenders. However, the same attention has not been paid to the youth offender population. The present study gathered data from the population (N = 4307) of all young persons in New Zealand whose antisocial behaviour resulted in a Youth Justice intake from the Department of Child, Youth, and Family (CYF) in 2002. Information was obtained about this population from the CYF database, CYRAS, and from the Police National Intelligence Application database for a stratified random sample (N = 500). Three models were developed using Hierarchical Cox regression to predict recidivism, and they each used a different definition of recidivism. The performance of the models was assessed using ROC analysis and they were found to predict recidivism with a moderately good level of accuracy. A validation sample (N = 500), different from the sample on which the models were developed, was used to further assess the performance of the models by showing that they were able to generalize to a new data set and continue to perform at an adequate level. An actuarial model, like the one developed in the present study, could be used to help make decisions about which young people within the Youth Justice System require intervention in order to reduce the likelihood of subsequent reoffending.

No description available.

According to some (e.g. Friston, 2010) predictive processing (PP) models of cognition have the potential to offer a grand unifying theory of cognition. The framework defines a flexible architecture governed by one simple principle - minimise error. The process of Bayesian inference used to achieve this goal results in an ongoing flow of prediction that both makes sense of perception and unifies it with action. Such a provocative and appealing theory naturally has caused ripples in philosophical circles, prompting several commentaries (e.g. Hohwy, 2012; Clark, 2016). This thesis tackles one outstanding philosophical problem in relation to PP - the question of mental representation. In attempting to understand the nature of mental representations in PP systems I touch on several contentious points in philosophy of cognitive science, including the explanatory power of mechanisms vs. dynamics, the internalism vs. externalism debate, and the knotty problem of proper biological function. Exploring these issues enables me to offer a speculative solution to the question of mental representation in PP systems, with further implications for understanding mental representation in a broader context. The result is a conception of mind that is deeply continuous with life. With an explanation of how normativity emerges in certain classes of self-maintaining systems of which cognitive systems are a subset. We discover the possibility of a harmonious union between mechanics and dynamics necessary for making sense of PP systems, each playing an indispensable role in our understanding of their internal representations.