CAREER: A New Model Order Reduction Framework for the Control of Advanced Propulsion and Energy Storage Systems for Electrified Vehicles

点击翻译按钮获取中文摘要

This Faculty Early Career Development (CAREER) project will investigate a novel framework that systematically transfers the accuracy and fidelity of physics-based models into low-order models suitable for control design of automotive propulsion systems. Automotive propulsion technologies have become considerably complex as manufacturers strive to improve fuel economy, emissions, safety and the driver experience. Such complexity requires the adoption of advanced control and estimation algorithms. The project will provide the necessary tools to control engineers who need to employ high-fidelity, physics-based models, albeit with considerably reduced complexity. The research will be directly applied to the estimation of thermal imbalances in Li-ion battery packs for high-performance electric vehicles, and to the estimation and control of surge dynamics in downsized boosted engines, two key technologies for current and future automotive powertrains. The automotive industry will participate to the research by providing insights and facilitating technology transfer. Research, education and outreach will be integrated through the Buckeye Current electric motorcycle team at The Ohio State University, inspired by e-mobility, e-racing and sustainable transportation. Integration will be also pursued through the enrichment of existing courses, creation of web-based teaching tools and student internship programs, and by collaborating with the Teaching Engineering to K-8 (TEK8) programs to attract young talents from underrepresented groups.The state of the art for model-based control design relies almost exclusively on low-fidelity models based on empirical approaches and on approximations of the physical system, which ultimately leads a loss of accuracy. This research introduces a transformative framework that analytically generates control-oriented models from the conservation laws for thermal, fluid and chemical systems in nonlinear Partial Differential Equations (PDEs) form. A projection-based Model Order Reduction (MOR) will be applied to generate Reduced Order Models (ROM) whose parameters are directly derived from physical constants through analytical relations, dramatically reducing the need for calibration. The projection will enforce specific properties "a priori" in the resulting ROM, namely stability, observability and controllability. Case studies will be conducted on the estimation of thermal imbalances in Li-ion battery packs for high-performance electric vehicles, and the estimation and control of surge dynamics in downsized boosted engines.