Performance Evaluation and State-of-Health (SOH) Monitoring of Electric Motor Drive Used for Electric Vehicle Propulsion

Date of Award

3-21-2023

Publication Type

Dissertation

Degree Name

Ph.D.

Department

Electrical and Computer Engineering

Keywords

Condition monitoring, Motor drive, Online remaining lifetime, Packaging fault identification, State-of-health, Thermal network model

Supervisor

N.Kar

Supervisor

LL.Iyer

Rights

info:eu-repo/semantics/embargoedAccess

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Abstract

The move towards electrified transportation has revived several key attributes of an electromechanical drive system (EMDS) regarding reliability, cost, and power density. One of the most critical components of an EMDS is the inverter, accounting for almost 50% of the drive cost. With the advent of semiconductor technology, it has witnessed evolutions in various aspects such as circuit topology, control schemes, and reliability assessment. However, accelerated consolidation of the electric vehicle (EV) as a mainstream automotive technology still emphasizes more aggressive inverter designs with predictive monitoring for the next generation of EVs. This work presented in this dissertation, hence, focuses on modeling an advanced inverter with improved efficiency and reliability.

First, the state-of-the-art electric motor drive topologies are analyzed using drive cycle-based vehicle dynamics model for understanding steady-state and transient performance characteristics. A unique co-simulation method has been developed for this analysis to incorporate harmonics interferences due to drive switching transient and manufacturing constraints. The developed co-simulation method analyzes conventionally used inverter topologies with immensely used silicon (Si) and advanced wide bandgap (WBG) based devices comparing efficiency, power density, cost, and weight. A comprehensive analytical model of an automotive inverter has been developed with electrical and thermal characteristics. Correspondingly, a finite element-based machine model is developed to considering parameter variation such as winding resistance, inductance, and flux linkage due to design constrains. Further improvement is achieved in developing machine model with online parameter estimation using non-invasive evolutionary algorithm. The analysis shows a great performance boost with WBG devices in a conventional inverter topology and improved power density. Nevertheless, the packaging technology of the WBG devices is very expensive due to the complexity in heat distribution, smaller form factor and limited knowledge on matrix formation of semiconductor chips as compared to high power density Si based packaging technology. Additionally, the comparative analysis shows that this technology is yet to be modified for advanced voltages. Thus, a novel hybrid multi-level inverter and nine-switch inverters are proposed with a combination of Si-based and WBG based devices. The designed inverter model forms a unique combination of available WBG and Si-based devices to operate at high voltage condition with reduced cost. Also, the proposed inverter topology uses extended bandgap and high electro mobility towards reduced switching loss, which leads to increase in lifetime. However, some other factors such as temperature dissipation have a significant effect on the lifetime of an inverter depending on the design configuration and need to be incorporated for long time performance analysis.

Therefore, in this dissertation a novel online state-of-health monitoring approach has been developed to identify the lifetime of an inverter power module towards improved design configuration. The approach contains an advanced power loss calculation scheme of the drive switching and conduction states considering steady-state and transient switching stages with the influence of parasitic elements. Subsequently, a novel thermal network model is developed for accurate temperature tracking towards accurate heat distribution through the power module layers. This model enhances the accuracy of the temperature tracking with a three-dimensional thermal network model considering semiconductor device cross-coupling area. In addition, an iterative process has been developed for constant estimation of thermal impedances considering material degradation.

To identify material degradation, a novel condition monitoring model has been developed with packaging material properties and geometric information. The model contains a modified cycle counting method for number of temperature cycles identification in an online condition. Considering the number of temperature cycles, the total stress is calculated, incorporating the power modules internal geometry, and temperature coefficient mismatch. Consequently, novel bond-wire fault detection and solder fatigue identification and classification models have been proposed using semiconductor electrical and thermal parameters. The developed model analyzes the material degradation and its impact on the semiconductor electrical and thermal characteristics to determine the fault location. Additionally, the developed model observes the parameter changes with the fault propagation. Further recommendations for model improvement, and future work is summarized towards the end of the dissertation.

Link to Full Text

Share

COinS