Cyber-physical merged learning for online optimisation of multi-mode hybrid vehicles with diverse time-scale objectives

Zhang, Cetengfei ORCID: 0000-0001-8710-1872 (2024). Cyber-physical merged learning for online optimisation of multi-mode hybrid vehicles with diverse time-scale objectives. University of Birmingham. Ph.D.

Preview

Zhang2024PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (23MB) | Preview

Abstract

A comprehensive investigation of the energy management and optimisation techniques used in multi-mode plug-in hybrid electric vehicles (PHEVs) is presented in this thesis. It focuses on using artificial intelligence and sophisticated control algorithms to improve multi-mode PHEV performance and efficiency. With the goal of increasing fuel efficiency and product life cycle, the research combines cutting-edge techniques for cyber-physical adaptive control schemes, real-time energy management, and battery state estimation.
This thesis first presents the development of a dedicated adaptive particle swarm optimisation (DAPSO) for offline optimisation based on a digital twin boost of the efficiency and dependability of PHEVs’ energy management system (EMS) control. In terms of fuel efficiency and battery state-of-charge (SoC) maintenance, the DAPSO is superior in performance compared to traditional algorithms.
Then, regarding the electrification of multi-mode PHEVs, an intelligent digital model of a battery is developed with challenging circumstances for an automotive battery to obtain real-time status estimation. In order to improve functionality and reliability in real-time applications, this model uses both deep neural networks (NN) and Gaussian process regression with automatic relevance determination (ARD-GPR) approaches based on a modified equivalent circuit battery model (ECM) for handling health indicators throughout the charging and discharging processes.
This thesis finally develops a cuboid equivalent consumption minimisation strategy (CECMS) for multi-mode PHEVs. The C-ECMS provides improved performance in online optimal energy management with multiple objectives in diverse time scales. In real-world driving conditions, this strategy improves overall performance and economy by establishing a compromise between battery health, fuel efficiency, and SoC control accuracy.
The results address both the scientific and practical elements of automotive technology, providing insightful information for the development of more efficient and environmentally friendly multi-mode PHEVs.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):