分会场

高效清洁燃烧

摘要

This paper studies a new concept of cyber-physical real-time optimization, wherein a mechanism of digital twin assisted parallel learning (DTPL) is proposed for improving robustness and energy efficiency of a fuzzy logic (FL-) based engine-powered hybrid propulsion control system. This mechanism allows parallel learning between a real supervisory controller and its digital twin during real-world driving. Once the virtual controller takes the lead, the new parameters of membership functions will be synchronized to the real controller at the same time. Followed by analysing the configuration of the hybrid propulsion model and its FL-based control system, the algorithm of chaos-enhanced accelerated particle swarm optimization is applied for parallel learning of the membership function. Based on fuel-prioritized cost functions, conditions for controller parameter synchronization are designed for a better fuel economy. the DTPL-driven control system is validated by a hardware-in-the-loop test. The results demonstrate that the DTPL-driven control system significantly reduces fuel consumption, up to 15% from charge sustaining and charge depleting based, and up to 12% from conventional FL-based systems.

关键词

Digital twin;fuzzy logic control;hybrid electric vehicles;online energy management;parallel computing;particle swarm optimization

电子U盘全文仅限大会已缴费参会代表下载。

您还没有登录，请您先 点击这里登录