Underwater robotic vehicles are used in a variety of environments that would be dangerous for humans. For these vehicles to be successful, they need to be tolerant of a variety of internal and external faults. To be resilient to internal faults, the system must be capable of determining the source of faulty behavior. However many different faults within a robotic vehicle can create identical faulty behavior, which makes the vehicles impossible to diagnose using conventional methods. I propose a novel active diagnosis method for differentiating between faults that would otherwise have identical behavior. I apply this method to a communication system and a power distribution system in a robotic vehicle and show that active diagnosis is successful in diagnosing partially observable faults. An example of an external fault is inter-robot communication in underwater robotics. The primary communication method for underwater vehicles is acoustic communication which relies heavily on line-of-sight tracking and range. This can cause severe packet loss between agents when a vehicle is operating around obstacles. I propose novel path-planning methods for an Autonomous Underwater Vehicle (AUV) that ferries messages between agents. I applied this method to a custom underwater simulator and illustrate how it can be used to preserve at least twice as many packets sent between agents than would be obtained using conventional methods.

A mixture model approach for clustering longitudinal data is introduced. The approach, which is based on mixtures of multivariate power exponential distributions, allows for varying tail-weight and peakedness in data. In the longitudinal setting, this corresponds to more or less concentration around the most central time course in a component. The models utilize a modified Cholesky decomposition of the component scale matrices and the associated maximum likelihood estimators are derived via a generalized expectation-maximization algorithm. / Thesis / Master of Science (MSc)

Model-based clustering is the use of finite mixture models to identify underlying group structures in data. Estimating parameters for mixture models is notoriously difficult, with the expectation-maximization (EM) algorithm being the predominant method. An alternative approach is the evolutionary algorithm (EA) which emulates natural selection on a population of candidate solutions. By leveraging a fitness function and genetic operators like crossover and mutation, EAs offer a distinct way to search the likelihood surface. EAs have been developed for model-based clustering in the multivariate setting; however, there is a growing interest in matrix-variate distributions for three-way data applications. In this context, we propose an EA for finite mixtures of matrix-variate distributions. / Thesis / Master of Science (MSc)

The incorporation of detailed chemistry models in internal combustion engine simulations is becoming mandatory as new combustion strategies and lower global emissions limits are setting the path towards a more efficient engine cycle simulation tool. In this report, the computational capability of the stochastic-based Kinetics SRM engine suite by CMCL Innovations is evaluated in depth. With the main objectives of this research to create a multi-physics co-simulation framework and improve the traditional engine modelling approach of individual simulation of engine system parts, the Kinetics SRM code was coupled with a GT-SUITE engine model to fill in the gap of accurate emissions predictions from one-dimensional simulation tools. The system was validated against testing points collected from the AJ133 V8 5L GDI engine running on the NEDC. The Kinetics SRM model is further advanced through a sensitivity analysis for the “unknown” internal parameters of the chemistry tool. A set of new parameters’ values has been established that gives the best overall trade-off between prediction accuracy and computational time. However, it still showed high percentage errors in modelling the emissions and it was discovered that the specific software package currently cannot simulate directed injection events. This is the first time a Kinetics SRM/GT-SUITE coupled code is employed to model a full 8-cylinder GDI SI engine. The approach showed some limitations regarding the Kinetics SRM and that in many cases is limited to trend analysis. The coupled engine – combustion emissions model is then linked with an exhaust aftertreatment system model in MATLAB Simulink, creating a multi-physics model-based co-simulation framework of engine performance, combustion characterisation, in-cylinder emissions formation and aftertreatment efficiency.

For distributed memory architectures, communication cost is a significant source of overhead in parallel scientific applications. Many proposed communication optimizations duplicate the behavior of well-written hand-tuned parallel code. Because of continuous changes in architectural components, these types of low-level optimizations are not always effective. This thesis seeks to develop a high-level optimization of work replication in which computations are replicated to minimize communications. There exist performance trade-offs between computation cost and communication cost because of work replication. Due to these trade-offs, it is required to determine which computations should be replicated to improve overall performance. This research presents the development of a model-based approach with heuristics to automatically determine the computations to replicate. Using a computational and communication model, the execution time is predicted to make replication decisions.

No description available.

No description available.

No description available.

Cluster analysis identifies homogeneous groups that are relevant within a population. In model-based clustering, group membership is estimated using a parametric finite mixture model, commonly the mathematically tractable Gaussian mixture model. One-way clustering methods can be restrictive in cases where there are suspected relationships between the variables in each component, leading to the idea of biclustering, which refers to clustering both observations and variables simultaneously. When the relationships between the variables are known, biclustering becomes one-way supervised. To this end, this thesis focuses on a novel one-way supervised biclustering family based on the Gaussian mixture model. In cases where biclustering may be overestimating the number of components in the data, a model averaging technique utilizing Occam's window is applied to produce better clustering results. Automatic outlier detection is introduced into the biclustering family using mixtures of contaminated Gaussian mixture models. Algorithms for model-fitting and parameter estimation are presented for the techniques described in this thesis, and simulation and real data studies are used to assess their performance. / Thesis / Doctor of Philosophy (PhD)

This thesis presents the \emph{Boundary Diagram Tool}, a tool for change impact analysis of large Simulink designs of embedded systems. The Boundary Diagram Tool extends the Reach/Coreach Tool, an existing tool for model slicing within a single Simulink model, to trace the impact of model changes through multiple Simulink models and to network interfaces of an automotive controller. While the change impact analysis results can be viewed directly within the Simulink models, the tool also uses various block diagrams to represent the impact analysis results with different levels of abstraction, motivated by industrial needs. In order to effectively present the complex impact analysis results, various techniques for visual representation of large graphs are employed. Furthermore, the Reach/Coreach Tool as an underlying model slicing engine was significantly improved. The Boundary Diagram Tool is currently being integrated into the software development process of a large automotive OEM (Original Equipment Manufacturer). It provides support during several phases of the change management process: change request analysis and evaluation, as well as the implementation, verification and integration of software changes. The tool also aids impact analyses required for compliance with functional safety standards such as ISO 26262. / Thesis / Master of Applied Science (MASc)