Search Results

Laser cladding is a method of material deposition through which a powdered or wire feedstock material is melted and consolidated by use of a laser to coat part of a substrate. Determining the parameters to fabricate the desired clad bead geometry for various configurations is problematic as it involves a significant investment of raw materials and time resources, and is challenging to develop a predictive model. The goal of this research is to develop an experimental methodology that minimizes the amount of data to be collected, and to develop a predictive model that is accurate, adaptable, and expandable. To develop the predictive model of the clad bead geometry, an integrated five-step approach is presented. From the experimental data, an artificial neural network model is developed along with multiple regression equations.

Successful manufacture of Carbon Fibre Reinforced Polymers (CFRP) by Long-Fibre Reinforced Thermoplastic (LFT) processes requires knowledge of the effect of numerous processing parameters such as temperature set-points, rotational machinery speeds, and matrix melt flow rates on the resulting material properties after the final compression moulding of the charge is complete. The degree to which the mechanical properties of the resulting material depend on these processing parameters is integral to the design of materials by any process, but the case study presented here highlights the manufacture of CFRP by LFT as a specific example. The material processing trials are part of the research performed by the International Composites Research Centre (ICRC) at the Fraunhofer Project Centre (FPC) located at the University of Western Ontario in London, Ontario, Canada.

Autonomous vehicles are currently a subject of great interest and there is heavy research on creating and improving algorithms for detecting objects in their vicinity. A ROS-based deep learning approach has been developed to detect objects using point cloud data. With encoded raw light detection and ranging (LiDAR) and camera data, several basic statistics such as elevation and density are generated. The system leverages a simple and fast convolutional neural network (CNN) solution for object identification and localization classification and generation of a bounding box to detect vehicles, pedestrians and cyclists was developed. The system is implemented on an Nvidia Jetson TX2 embedded computing platform, the classification and location of the objects are determined by the neural network. Coordinates and other properties of the object are published on to various ROS topics which are then serviced by visualization and data handling routines.

Neural networks are flexible modeling tools that can be used in conjunction with multi-body dynamics models to better predict nonlinear behaviour of components. This paper focuses on a process that incorporates a neural network model of a nonlinear damping force into a single degree of freedom mass-spring-damper model. Software tools and their interaction are specified. The verification of this process is the focal point of this paper and is a necessary step before further correlation studies can be performed on more complex component representations.

Performance of the ignition system becomes more important than ever, because of the extensively used EGR in modern spark-ignition engines. Future lean burn SI and SACI combustion modes demand even stronger ignition capability for robust ignition control. For spark-based ignition systems, extensive research has been carried out to investigate the discharge characteristics of the ignition process, including discharge current amplitude, discharge duration, spark energy, and plasma stretching. The correlation between the spark stretch and the discharge energy, as well as the impact of discharge current level on this correlation, are important with respect to both ignition performance, and ignition system design. In this paper, a constant volume combustion chamber is applied to study the impact of plasma stretch on the spark energy release process with cross-flow speed from 0 m/s up to 70 m/s.

The development of a collision severity model can serve as an important tool in understanding the requirements for devising countermeasures to improve occupant safety and traffic safety. Collision type, weather conditions, and driver intoxication are some of the factors that may influence motor vehicle collisions. The objective of this study is to use artificial neural networks (ANNs) to identify the major determinants or contributors to fatal collisions based on various driver, vehicle, and environment characteristics obtained from collision data from Transport Canada. The developed model will have the capability to predict similar collision outcomes based on the variables analyzed in this study. A multilayer perceptron (MLP) neural network model with feed-forward back-propagation architecture is used to develop a generalized model for predicting collision severity. The model output, collision severity, is divided into three categories - fatal, injury, and property damage only.