No description available.

<p>Model based design has had a large impact on the process of software development in many different industries. A lack of formality in these environments can lead to incorrect software and does not facilitate the formal analysis of created models. A formal tool known as tabular expressions have been successfully used in developing safety critical systems, however insufficient tool support has hampered their wider adoption. To address this shortfall we have developed the Tabular Expression Toolbox for Matlab/Simulink.</p> <p>We have developed an intuitive user interface that allows users to easily create, modify and check the completeness and disjointness of tabular expressions using the theorem prover PVS or SMT solver CVC3. The tabular expressions are translated to m-functions allowing their seamless use with Matlab's simulation and code generation. We present a method of generating counter examples for incorrect tables and a means of effectively displaying this information to the user. We provide support for modelling inputs as floating point numbers, through subtyping a user can show the properness of a table using a more concrete representation of data. The developed tools and processes have been used in the modelling of a nuclear shutdown system as a case study of the practicality and usefulness of the tools.</p> / Master of Applied Science (MASc)

<p>Evidence suggests that a change in the firing rate of dopamine (DA) cells is a major neurobiological correlate of learning. The Temporal Difference (TD) learning algorithm provides a popular account of the DA signal as conveying the error between expected and actual rewards. Other accounts have attempted to code the DA firing pattern as conveying surprise or salience. The DA mediated cells have also been implicated in several neuropsychological disorders such as obsessive compulsive disorder and schizophrenia. Compelling neuropsychological explanations of the DA signal also frame it as conveying salience. A model-based reinforcement learning algorithm using a salience signal analogous to dopamine neurons was built and used to model existing animal behavioral data.</p> <p>Different reinforcement learning models were then compared under conditions of altered DA firing patterns. Several differing predictions of the TD model and the salience model were compared against animal behavioral data in an obsessive compulsive disorder (OCD) model using a dopamine agonist. The results show that the salience model predictions more accurately model actual animal behavior.</p> <p>The role of context in the salience model is different than the standard TD-learning algorithm. Several predictions of the salience model for how people should respond to context shifts of differing salience were tested against known behavioral correlates of endogenous dopamine levels. As predicted, individuals with behavioral traits correlated with higher endogenous dopamine levels are far more sensitive to low salience context shifts than those with correlates to lower endogenous dopamine levels. This is a unique prediction of the salience model for the DA signal which allows for better integration of reinforcement learning models and neuropsychological frameworks for discussing the role of dopamine in learning, memory and behavior.</p> / Doctor of Science (PhD)

Growth mixture models (GMMs) are used to model intra-individual change and inter-individual differences in change and to detect underlying group structure in longitudinal studies. Regularly, these models are fitted under the assumption of normality, an assumption that is frequently invalid. To this end, this thesis focuses on the development of novel non-elliptical growth mixture models to better fit real data. Two non-elliptical growth mixture models, via the multivariate skew-t distribution and the generalized hyperbolic distribution, are developed and applied to simulated and real data. Furthermore, these two non-elliptical growth mixture models are extended to accommodate missing values, which are near-ubiquitous in real data. Recently, finite mixtures of non-elliptical distributions have flourished and facilitated the flexible clustering of the data featuring longer tails and asymmetry. However, in practice, real data often have missing values, and so work in this direction is also pursued. A novel approach, via mixtures of the generalized hyperbolic distribution and mixtures of the multivariate skew-t distributions, is presented to handle missing values in mixture model-based clustering context. To increase parsimony, families of mixture models have been developed by imposing constraints on the component scale matrices whenever missing data occur. Next, a mixture of generalized hyperbolic factor analyzers model is also proposed to cluster high-dimensional data with different patterns of missing values. Two missingness indicator matrices are also introduced to ease the computational burden. The algorithms used for parameter estimation are presented, and the performance of the methods is illustrated on simulated and real data. / Thesis / Doctor of Philosophy (PhD)

Software has emerged as a significant part of many domains, including financial service platforms, social networks, medical devices and vehicle control. In critical domains, standards organizations have responded to this by creating regulations to address issues such as safety, security and privacy. In this context, compliance of software with standards has emerged as a key issue. For companies, compliance is a complex and costly goal to achieve and is often accomplished by producing so-called assurance cases, which demonstrate that the system indeed satisfies the property imposed by a standard (e.g., safety, security, privacy) by linking evidence to support claims made about the system. However, as systems undergo evolution for a variety of reasons, including fixing bugs, adding functionality or improving system quality, maintaining assurance cases multiplies the effort. Increasingly, models and model-driven engineering are being used as a means to facilitate communication and collaboration between the stakeholders in the compliance value chain and, further, to introduce automation into regulatory compliance tasks. A complexity problem also exists with the proliferation of software models in model-based software development, and the field of Model Management has emerged to address this challenge. Model Management focuses on a high-level view in which entire models and their relationships (i.e., mappings between models) can be manipulated using specialized operators to achieve useful outcomes. In this thesis, we exploit this connection between model driven engineering and regulatory compliance, and explore how to use Model Management techniques to address software compliance management issues, focusing on assurance case change impact assessment, evolution and reuse. We support the presented approach with tooling and a case study. Although the main contributions of this thesis are not domain specific, for validation, we ground our approaches in the automotive domain and the ISO 26262 standard for functional safety of road vehicles. / Thesis / Doctor of Philosophy (PhD)

An increase in consumer demand and scarcity of available resources has led industrialists to hunt for solutions related to the automation of traditional manufacturing and production processes, optimizing resource consumption while improving the overall efficiency of the process. The resultant revolution brought forward the concept of cyber-physical production systems. Furthermore, industries within the private sector have integrated artificial intelligence with their traditional production processes as Cobots (collaborative robots), thus introducing the concept of Autonomous Cyber-Physical Production Systems. Although these systems maximize the production or manufacturing process while efficiently using the available resources, the machine learning component integrated into the traditional cyber-physical production system brings about trust-related issues due to its possible lack of predictability and transparency. Implementing trust-related attributes within autonomous cyber-physical production systems alone cannot overcome the highlighted problem. Therefore, a detailed risk assessment is required to identify and assess any trust-related risks in the system, especially at the early stages of the software development life cycle, to avoid major incidents and reduce maintenance costs. Based on the above-stated facts, this research proposes a model-based risk assessment technique for evaluating the trustworthiness of autonomous cyber-physical production systems. The proposed technique focuses on the identification and assessment of trust-related risks originating from the dynamic behavior of the machine learning component in autonomous cyber-physical production systems. For this, we use existing standards and techniques proposed for risk assessment in cyber-physical production systems as common ground to facilitate better implementation of trustworthiness in autonomous cyber-physical production systems. The proposed technique is aimed at overcoming the structural and behavioral limitations reported in existing model-based risk assessment techniques when dealing with autonomous cyber-physical production systems.

While the existing thermomechanical constitutive models can predict the behavior of SMA-actuated systems in most cases, in this study, we have shown that there are certain situations in which these models are not able to predict the behavior of SMAs. To this end, a rotary SMA-actuated robotic arm is modeled using the existing constitutive models. The model is verified against the experimental results to document that under certain conditions, the model is not able to predict the behavior of the SMA-actuated manipulator. Such cases most often occur when the temperature and stress of the SMA wire change simultaneously. The constitutive model discrepancy is also studied experimentally using a dead-weight that is actuated by an SMA wire. Subsequently, an enhanced phenomenological model is developed. The enhanced model is able to predict the behavior of SMAs under complex thermomechanical loadings. For the SMA-actuated robotic arm, several control methods are designed through simulations. A position-based PID controller is designed first, and it is found that this controller cannot perform well for all the desired angular positions(set-points). A Variable Structure Control (VSC) based on the angular position and velocity is presented that has a relatively better erformance for all the set-points. To improve the erformance of the VSC, in terms of the steady state error, an Extended Kalman Filter is designed and used to modify the VSC design. The modified VSC is based on the angular position and angular velocity of the actuator and the estimated temperature of the SMA wire. Furthermore, a Sliding Mode Controller is designed based on the stress of the SMA wire. Finally, a model-based Backstepping Controller is designed for the SMA-actuated arm. This model-bsed controller allows designing the controller parameters based on the parameters of the system. Additionally, the stability of the controller is studied. Using the Lyapunov stability analysis, it is shown that the model-based Backstepping Controller is able to asymptotically stabilize the system. / Ph. D.

Throughout the history of software development, advances have been made that improve the ability of developers to create systems by enabling them to work closer to their application domain. These advances have given programmers higher level abstractions with which to reason about problems. A separation of concerns between logic and implementation allows for reuse of components, portability between implementation platforms, and higher productivity. Parallels can be drawn between the challenges that the field of reconfigurable computing (RC) is facing today and what the field of software engineering has gone through in the past. Most RC work is done in low level hardware description languages (HDLs) at the circuit level. A large productivity gap exists between the ability of RC developers and the potential of the technology. The small number of RC experts is not enough to meet the demands for RC applications. Model-based engineering principles provide a way to reason about RC devices at a higher level, allowing for greater productivity, reuse, and portability. Higher level abstractions allow developers to deal with larger and more complex systems. A modeling environment has been developed to aid users in creating models, storing, reusing and generating hardware implementation code for their system. This environment serves as a starting point to apply model-based techniques to the field of RC to tighten the productivity gap. Future work can build on this model-based framework to take advantage of the unique features of reconfigurable devices, optimize their performance, and further open the field to a wider audience. / Master of Science

Formally-based design and implementation techniques for complex safety-critical embedded systems are required not only to handle the complexity, but also to provide correctness guarantees. Traditional design approaches struggle to cope with complexity, and they generally require extensive testing to guarantee correctness. As the designs get larger and more complex, traditional approaches face many limitations. An alternate design approach is to adopt a "correct-by-construction" paradigm and synthesize the desired hardware and software from the high-level descriptions expressed using one of the many formal modeling languages. Since these languages are equipped with formal semantics, formally-based tools can be employed for various analysis. In this dissertation, we adopt one such formal modeling language - MRICDF (Multi-Rate Instantaneous Channel-connected Data Flow). MRICDF is a graphical, declarative, polychronous modeling language, with a formalism that allows the modeler to easily describe multi-clocked systems without the necessity of global clock. Unnecessary synchronizations among concurrent computation entities can be avoided using a polychronous language such as MRICDF. We have explored a Boolean theory-based techniques for synthesizing multi-threaded/concurrent code and extended the technique to improve the performance of synthesized multi-threaded code. We also explored synthesizing ASIPs (Application Specific Instruction Set Processors) from MRICDF models. Further, we have developed formal techniques to identify constructive causality in polychronous models. We have also developed SMT (Satisfiablity Modulo Theory)-based techniques to identify dimensional inconsistencies and to perform value-range analysis of polychronous models. / Ph. D.

Functional Electrical Stimulation (FES) is a technique that applies electrical currents to nervous tissue in order to actively induce muscle contraction. Recent research has shown that FES provides a promising treatment to restore functional tasks due to paralysis caused by spinal cord injury, head injury, and stroke, to mention a few. Therefore, the overarching goal of this research work is to develop FES controllers to enable patients with movement-disorder to control their limbs in a desired manner and, in particular, to aid Parkinson's patients to suppress hand tremor. In our effort to develop strategies for muscle stimulation control, we first implement a model-based control technique assuming that all the states are measurable. The Hill-type muscle model coupled with a simplified 2DoF model of the arm is used to study the performance of our proposed adaptive sliding mode controller for simulation purpose. However, in the more practical situations, human limb dynamics are extremely complicate and it is inadequate to use model based controllers, especially considering there are still technical limitations that allow in vivo measurements of muscle activity. To tackle these challenges, we have developed output feedback adaptive control approaches for a class of unknown multi-input multi-output systems. Such control strategies are first developed for linear systems, and then extended to the nonlinear case. The proposed controllers, supported by experimental results, require minimum knowledge of the system dynamics and avoid many restrictive assumptions typically found in the literature. Therefore, we expect that the results introduced in this dissertation can provide a solution for a wide class of nonlinear uncertain systems, with focus on practical issues such as partial state measurement and the presence of mismatched uncertainties. / Ph. D.