Estimation and fault diagnosis for vehicle energy systems

Driven by a desire to achieve reduced carbon emissions and maintenance costs, along with an increase in efficiency and performance, electrification has become a major trend in modern vehicles. This increase in electrification is accompanied by an increase in thermal power dissipated due to electrical inefficiencies. Consequently, temperature regulation becomes a greater challenge for these safety-critical systems. Electrified vehicles consist of systems of systems that operate over a wide span of energy domains and timescales. To ensure their safe, reliable, and efficient performance, a holistic system perspective for estimation is needed. Accurate dynamic state estimation is critical for two main reasons: 1. Thermal management: This dissertation proposes a system perspective state estimation framework for complex multi-domain and multi-timescale dynamical systems. The framework consists of a multilevel hierarchical network of observers with each level having a unique update rate. To account for the significant interactions between subsystems, a novel bidirectional coordination strategy is developed. Sufficient conditions for the stability and convergence of the hierarchical network are derived. Experimental validation is conducted on a testbed representative of a fluid thermal management system of an electrified aircraft. Closed-loop simulation and experimental results confirm a reduction in computational cost compared to a conventional centralized observer and an increase in estimation accuracy compared to a decentralized observer which ignores coupling between subsystems. 2. Fault diagnosis: This dissertation proposes a robust system-perspective fault diagnosis framework for complex energy systems. Fault detection and isolation is derived from a set of structured residuals obtained from a bank of observers. Robustness is achieved by decoupling the unknown disturbances such as modeling error, linearization error, parameter variation, and noise from the residuals. The proposed approach is validated on a testbed representative of a fluid thermal management system of an electrified aircraft. Simulation and experimental results demonstrate successful fault detection and isolation with no false alarms or missed detections.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Mechanical Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2021

Author dc:creator

• Tannous, Pamela Joseph

Contributors dc:contributor

• Alleyne, Andrew
• Beck, Carolyn
• Salapaka, Srinivasa
• Mehta, Prashant

• Estimation
• vehicle energy systems
• hierarchical estimation
• hierarchical control
• fault diagnosis
• model predictive control
• electrified vehicles

Statement dc:rights

• Copyright 2020 Pamela Tannous

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/109411

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/109411