No description available.

No description available.

No description available.

No description available.

No description available.

This thesis presents a damage model developed for the rapid prediction of the vulnerability of a ship concept design to AIREX weapon effects. The model uses simplified physics-based and empirical equations, threat charge size, geometry of the design, and the structure of the design as inputs. The damage volumes are customized to the design being assessed instead using of a single volume defined only by the threat charge size as in previous damage ellipsoid methods. This methodology is validated against a range of charge sizes and a library of notional threats is created. The model uses a randomized hit distribution that is generated using notional threat targeting and the geometry of the design. A Preliminary Arrangement and Vulnerability (PAandV) model is updated with this methodology and used to calculate an Overall Measure of Vulnerability (OMOV) by determining equipment failures and calculating the resulting loss of mission capabilities. A selection of baseline designs from a large design space search in a Concept and Requirements Exploration (CandRE) are assessed using this methodology. / Master of Science

In this work, the thermo-mechanical response and failure of loaded, fire-exposed composite structures was studied. Unique experimental equipment and procedures were developed and experiments were performed to assess the effects of mechanical loading and fire exposure on the service life of composite beams. A series of analytical models was assembled to describe the fire growth and structural response processes for the system used in the experiments. This series of models consists of a fire model (to predict the heat flux to the fire-exposed beam), a thermal response model (to calculate the temperature distribution within the beam due to this heat flux), a stiffness-temperature model (to calculate the loss in stiffness at elevated temperatures), a mechanical response model (to compute the strain distribution within the loaded beam), and a material failure model (to calculate the strain at which the beam is expected to fail). Each of these models is independently validated by comparing predictions with experimental results. The models are then used to predict the times-to-failure for beams over a range of fire and loading conditions. The predicted failure times agree fairly well with experimental results, but it is expected that the agreement could be improved with improvements to the first model in the series - the fire model. / Master of Science

Future frame prediction is a difficult but useful problem to solve in deep learning. The technology can be used to predict future occurrences in a video, anticipate anomalies, and aid autonomous devices in smart decision making. Although there is potential with frame prediction technology, there is still progress that needs to be made with it. As the predicted frame becomes farther away from the last input frame, the image becomes blurry and distorted. This indicates that the model is more uncertain about the motion occurring in the image frame. To reduce model uncertainty shown in predictions, optical flow information from each video was extracted and combined with the video frames. An optical flow-based approach is uncommon in frame prediction and has not been implemented with a fully Convolutional Neural Network (CNN) based architecture. In this work, the change in image quality evaluation metrics and overall image quality is analyzed across 4 different datasets between a state-of-the-art frame prediction model and a modified model that combines optical flow information. The results demonstrate that adding optical flow information improves the model Mean Squared Error (MSE) by 4.11% and its Structural Similarity Index Metric (SSIM) by 0.41% for the Moving MNIST dataset. Optical flow improved the SSIM value of Taxi BJ, KTH, and KITTI by 0.02%, 0.011%, and 1.297% respectively. While there was a consistent improvement in performance, the models still need more improvement in terms of the quality of images predicted in the distant future. / Master of Science / Future frame prediction is a technology that allows computers to predict what future video frames will look like. This can be used to predict future occurrences in a video, anticipate anomalies, and aid autonomous devices in smart decision making. Although there is potential with frame prediction technology, there is still progress that needs to be made with it. As the predicted frame becomes farther away from the last input frame, the image becomes blurry and distorted. This indicates that the model is more uncertain about the motion occurring in the image frame. To reduce model uncertainty shown in predictions, optical flow information from each video was extracted and combined with the video frames. Optical flow is the change in direction and magnitude of a moving object in a video. This type of information is helpful for making frame predictions because it gives the model additional information on how objects are moving to base its predictions on. In this work, the change in image quality evaluation metrics and overall image quality is analyzed across 4 different datasets between a state-of-the-art frame prediction model and a modified model that combines optical flow information. The results demonstrate that adding optical flow information improves the model Mean Squared Error (MSE) by 4.11% and its Structural Similarity Index Metric (SSIM) by 0.41% for the Moving MNIST dataset. Optical flow improved the SSIM value of Taxi BJ, KTH, and KITTI by 0.02%, 0.011%, and 1.297% respectively. While there was a consistent improvement in performance, the models still need more improvement in terms of the quality of images predicted in the distant future.

Wind energy forecasting can facilitate wind energy integration into a power system. In particular, the management of power system security would benefit from forecast information on plausible large, rapid change in wind power generation. Numerical Weather Prediction (NWP) systems are presently the best available tools for wind energy forecasting for projection times between 3 and 48 hours. In this thesis, the types of weather phenomena that cause large, rapid changes in wind power in southeast Australia are classified using observations from three wind farms. The results show that the majority of events are due to horizontal propagation of spatial weather features. A study of NWP systems reveals that they are generally good at forecasting the broad large-scale weather phenomena but may misplace their location relative to the physical world. Errors may result from developing single time-series forecasts from a single NWP grid point, or from a single interpolation of proximate grid points. This thesis presents a new approach that displays NWP wind forecast information from a field of multiple grid points around the wind farm location. Displaying the NWP wind speeds at the multiple grid points directly would potentially be misleading as they each reflect the estimated local surface roughness and terrain at a particular grid point. Thus, a methodology was developed to convert the NWP wind speeds at the multiple grid points to values that reflect surface conditions at the wind farm site. The conversion method is evaluated with encouraging results by visual inspection and by comparing with an NWP ensemble. The multiple grid point information can also be used to improve downscaling results by filtering out data where there is a large chance of a discrepancy between an NWP time-series forecast and observations. The converted wind speeds at multiple grid points can be downscaled to site-equivalent wind speeds and transformed to wind farm power assuming unconstrained wind farm operation at one or more wind farm sites. This provides a visual decision support tool that can help a forecast user assess the possibility of large, rapid changes in wind power from one or more wind farms.

Mobility prediction is defined as guessing the next access point(s) a mobile terminal will join so as to connect to a (wired or wireless) network. Knowing in advance where a terminal is heading for allows taking proactive measures so as to mitigate the impact of handovers and, hence, improve the network QoS. This thesis analyzes this topic from different points of view. It is divided into three parts. The first part evaluates the feasibility of mobility prediction in a real environment. It thus analyzes a mobility trace captured from a real network to measure the intrinsic entropy of the nodes motion and to measure the effectiveness of a simple prediction method. The second part investigates how to perform mobility prediction. Firstly, it examines a generic prediction scheme based on a simple machine learning method; this scheme is evaluated under various conditions. Secondly, it shows how the pieces of information that are most useful for the prediction algorithm can be obtained. The third part studies how knowing the probable next access point of a mobile terminal allows one to improve the QoS of the network considered. We deal with two situations. We first show how the handover blocking rate of a cellular network can be decreased thanks to resource reservation. We then propose a new routing protocol for delay tolerant networks (i.e. an ad hoc network where packets must be delayed in the absence of an end-to-end path) that assumes that the contacts between the nodes can be (imperfectly) predicted.