Automotive Partnership Canada
Development of Optimal Electric Drive Trains for On-Road Vehicles
This is a collaborative R&D project between McGill University and the Canadian industry partners Linamar, TM4, and Infolytica. The goal of the proposed research is to develop a generic, high-performance electrical powertrain for electric vehicles that offers higher performance in a smaller package and at lower cost than today’s electrical powertrains.The focus of our current work is on the accurate prediction of iron losses, robust and multi-objective optimization for electrical machines design.
More information about the project can be found here.
Finite element method is a powerful tool to solve partial differential equations. It brings unknown field quantities into a system of linear equations as "Ax=b", in which A is a sparse matrix. The Conjugate Gradient (CG) method is one of the most popular iterative methods used for solving such a large and sparse system. The dominant computation cost lies in the sparse matrix-vector multiplication (SMVM) at each iteration stage of the CG method. Currently, Field-Programmable Gate Arrays, multi-core processors, and Graphic Processing Units offer competitive computing advantage and dictates new SMVM schemes to exploits their parallelism. We are developing algorithms to format large matrices, partition computation, and equally load the processors, The aim is to achieve linear increase of computational speed as the number of cores/processing units increases irrespective of the size of the problem. These algorithms would lie in the core of some CG solvers to solve large finite element problems.
Microwave breast cancer detection techniques have been proposed as a complementary technology to the standard x-ray mammography. They offer the potential advantages of low cost, comfortable scans, and they do not require the ionizing radiation that mammography does. Microwave-based breast imaging systems operate based on the inherent contrast in the dielectric properties between healthy and malignant tissues.
The focus of our current work is on a clinical prototype for microwave breast screening via multistatic radar using time-domain measurements. We have also designed anatomically and electrically realistic breast models (phantoms) with which we can test the system under various scenarios. Our system has been tested thoroughly with the phantoms, and we are now optimizing the measurements through patient testing in a clinical setting.