Distributed Model Predictive Control with Application to 48V Diesel Mild Hybrid Powertrains
Author | : Yuxing Liu |
Publisher | : |
Total Pages | : |
Release | : 2019 |
ISBN-10 | : OCLC:1158287667 |
ISBN-13 | : |
Rating | : 4/5 ( Downloads) |
Download or read book Distributed Model Predictive Control with Application to 48V Diesel Mild Hybrid Powertrains written by Yuxing Liu and published by . This book was released on 2019 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: 48V mild hybrid technology along with electrification of auxiliary loads is a promising solution to enhance fuel economy and reduce tailpipe emissions. However, the increased complexity of advanced electrified powertrains brings also significant challenges in the control design and calibration process. Conventional methods based on decentralized or hierarchical control architectures inevitably ignore the interactions among subsystems, and hence cannot achieve system-wide optimal performance. Meanwhile, developing and implementing centralized control architectures are practically intractable, due to the presence of multiple control inputs, different optimization objectives, and reconfigurable system structures. This dissertation aims at developing a novel Distributed Model Predictive Control (MPC) framework, tailored for a 48V Diesel mild hybrid powertrain, coupled with an electrically driven booster (E-Booster) and an electrically heated catalyst (EHC). The proposed methodology exploits the benefits of a distributed control system consisting of interconnected, local optimal controllers that approach system-wide optimal performance by cooperation, and also exhibit a flexible system structure to accommodate actuator on/off operations. In specific, this dissertation addresses two essential control problems in the field of electrified Diesel powertrains. First, a low-level engine air path control is designed for reference tracking, covering both turbocharging and electrical boosting modes. A nonlinear distributed MPC is developed, which is able to achieve the system-wide optimal performance and closed-loop stability, while rendering the E-Booster module plug-and-play. This approach is extended to a Lyapunov-based distributed MPC, where a nonlinear control law is embedded in local controllers to ensure the closed-loop stability with no communication. Then, a high-level supervisory control is designed for system-level energy management of a hybrid electric vehicle with EHC, during cold-start and normal operations. An online implementable distributed supervisory control is designed, rooted in the solution structure of Hybrid Minimum Principle and costate approximations. The distributed control is essentially an extension of the well-known Equivalent Consumption Minimization Strategy (ECMS), so the EHC module is compatible with any prior ECMS controller. The distributed control in normal operations is extended to an MPC, which incorporates vehicle velocity predictions to enhance fuel economy and reduce calibration efforts.