No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

<p> This thesis presents a strategy for documenting real-time control systems, and the work products that result from its application to the development of an alternate fuel injection controller. In doing so, this document contributes technically to the areas of automotive control, and control systems documentation. The strategy was not developed independently of the control system, but in a manner which reflects its size and complexity.</p> <p> The controller is used to generate and transmit appropriately timed and sized pulses to an alternative fuel injector array, and switch auxiliary devices including a fuel heater, and an injector lock-off. Such a controller, when used to inject natural gas or propane into a gasoline burning engine, provides a reduction in both engine operating costs and harmful engine emissions.</p> <p> The controller stores a fuel map that relates the amount of energy released by the combustion of petroleum to that released by the combustion of an alternate fuel, over a range of varying environmental conditions. The fuel map is used to calculate the length of alternate injection pulses. These maps have been designed by, and are the property of Cosimo's Garage Ltd. and thus will not appear in this document.</p> <p> At present, nearly all large engine car conversion technology is more rigid than the solution provided here. Conversion costs are often prohibitive and problems requiring professional service are frequent. Should the controller described here, help to curb conversion costs and reduce the need for frequent service as is expected, the controller will be a viable candidate for production and sale.</p> / Thesis / Master of Applied Science (MASc)

This thesis presents a simulation model for a Control Area Network (CAN) software stack, the Dual FIFO CAN (DFC) stack, and a method for identifying and incorporating the details of the host environment (hardware setup, operating system, etc.) into the implementation of the simulation model in order to achieve a high level of fidelity. The method enable the simulation model to produce more realistic simulation results that are close to real-life experiments of the target system compared to existing commercial and academic simulation tools, which mostly ignore the system details The simulation model is implemented based on the specification documents of the DFC stack as well as knowledge gained from real-life experiments about the DFC stack and its host environment, a dual-core Electric Control Unit (ECU) hardware test bench that runs a Real-Time Operating System (RTOS). Like the actual DFC stack, the simulation model offers features such as dual non-preemptive FIFO transmit queues and TX buffers, and reserved slots in the queues for higher-priority messages. By using the method introduced in this research, the simulation model also offers options, once enabled and configured with proper parameters, for simulating a host environment that has effects on the behaviors of the modeled CAN stack. And these features are not fully available in existing commercial and academic simulation tools. The model provides internal calibration values of the DFC stack as configurable parameters to the user, making it easy to customize the simulation. Configurable calibration values includes the total number of slots in the transmit FIFO queues, number of reserved slots in the queues, transmit-rate thresholds that decide to which transmit queue a message is routed and whether a message is eligible to enter the reserved slots of the queues, and together they determine the queuing behaviors of the DFC stack. The options for simulating a host environment (an ECU on a CAN network in a modern vehicle, for instance) is capable of recreating the timing effects (delays, jitters or other effects due to the processing load, physical limitation and internal implementation) of the target host environment on the simulation results. Both deterministic (constant values, etc.) and/or statistical (probability distributions, etc.) models can be used to configure each single timing effect from the simulated host environment. The simulation model is also automated to transmit a set of customized transmit message (configurable message ID, DLC, period and internal transmission priority) and process simulation results according to the purpose of the simulation (statistical analysis, plots of data, etc). These features make it possible for the simulation model to be used not only to simulate various customized simulation scenarios, but also for different purposes in various stages of the development process, for instance, a pre-experiment simulation run before a test bench experiment to test the correctness of the calibrations and predict the possible outcomes of the experiment, or, simulations for confirmation purposes in order validate the test bench data after the test experiment. The model is compatible with typical modeling, simulation and development environments as it is implemented in MATLAB SimEvents environment, which works with third-party CAN development tools such as Vector CANoe. It is also designed to work with the high-fidelity model of the Vector CAN protocol stack from Whinton (2016). / Thesis / Master of Applied Science (MASc)

Due to the rapid and prosperous development of information technology, data centers are widely used in every aspect of social life, such as industry, economy or even our daily life. This work considers the idea of developing a data-driven model based model predictive control (MPC) to regulate temperature for a class of single-rack data centers (DCs). An auto-regressive exogenous (ARX) model is identified for our DC system using partial least square (PLS) to predict the behavior of multi-inputs-single-output (MISO) thermal system. Then an MPC controller is designed to control the temperature inside IT rack based on the identified ARX model. Moreover, fuzzy c-means (FCM) is employed to cluster the measured data set. Based on the clustered data sets, PLS is adopted to identify multiple locally linear ARX models which will be combined by appropriate weights in order to capture the nonlinear behavior of the highly-nonlinear thermal system inside the IT rack. The effectiveness of the proposed method is illustrated through experiments on our single-rack DC and it is also compared with proportional-integral (PI) control. / Thesis / Master of Applied Science (MASc)

Natural user interfaces, including verbal vocal interactions like speech processing, are ubiquitous and commonly used in both industry and academic settings. However, this field is limited by the speech and language components. Non-verbal vocal interaction (NVVI) provides further opportunities for people to use their vocals as an input modality. Despite the many possibilities of NVVI input modalities, such as whistling, humming, and tongue clicking, the field is niche and literature is few and far between. This work attempts to address these gaps, as well as the small sample sizes of performed studies of prior work. The problem definition is defined as to perform a large-scale study exploring a pitch-based NVVI modality that uses a relative pitch interaction technique to offer a continuous mode of one-dimensional interaction. A user study is outlined and performed via an ecosystem comprising of Amazon Mechanical Turk for recruitment and study access, a modularized study website, and a secure server that stores the study results, tasks users with controlling a slider with the NVVI technique by humming and whistling, in addition to using the computer mouse to perform these tasks as a baseline. In total, 72 participants' results are considered for analysis. Results show that the pitch based NVVI technique used in this study does not follow Fitts' Law, is not as performant as the computer mouse, humming is a more performant modality with the NVVI technique than whistling, and that participants experienced a significantly higher task workload using the NVVI technique than the computer mouse. Using the results of this study and from reviewed literature, an NVVI framework is developed and implemented as a contribution of this work. / Master of Science