Data-driven decision analytics framework for complex engineering design applications

点击翻译按钮获取中文摘要

Advances in numerical algorithms and computing architectures made over the last two decades have led to the increasing adoption of computational methods in engineering design practice. This has enabled significant improvements in the performance and quality of complex engineering systems while reducing the time required to bring new products to market. However, in spite of the significant progress that has been made in the field of computation-based engineering design, a number of challenges remain to be overcome in order to develop the next generation of software tools for design of complex engineering systems. The proposed project will address some of the outstanding challenges in the field of data-driven engineering design optimization under uncertainty. The main topics of investigation are: (i) scalable emulator construction algorithms that use Kronecker matrix algebra to accelerate calculations for large-scale high-dimensional datasets, (ii) efficient computational methods for optimum sequential planning of computer experiments and reuse of data for product family design, (iii) scalable compressive sensing algorithms for uncertainty quantification, (ii) emulator-assisted entropy search algorithms for design optimization under uncertainty, and (ii) rapid calculation of decision analytics relevant to engineering design and decision making under uncertainty. Progress in these research areas would enable the application of data-driven methods to complex computational engineering problems that cannot be tackled at present using existing methods. The proposed research program will be carried out in collaboration with Pratt & Whitney Canada who is a global leader in gas turbine engines.