Files in this item



application/pdfKuai_Yingying.pdf (17MB)
(no description provided)PDF


Title:Comprehensive optimization and practical design of power electronic systems under multiple competing performance demands
Author(s):Kuai, Yingying
Director of Research:Chapman, Patrick L.
Doctoral Committee Chair(s):Chapman, Patrick L.
Doctoral Committee Member(s):Krein, Philip T.; Domínguez-García, Alejandro D.; Chen, Deming
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Voltage Regulator
Voltage Sourced Inverter
Model Simplification
Linear Regression
LCL Filter
Optimal Design
Abstract:Balancing trade-offs between several requirements, such as cost, steady-state performance, transient response, and efficiency is a constant theme in power electronic systems design. This project develops a comprehensive optimization approach to design the power stage and feedback controller simultaneously and achieve practical solutions that are ready to implement. The study is presented through the context of voltage regulator (VR) and voltage sourced inverter (VSI) design, both of which exemplify power electronic systems under several competing performance demands. A single-objective, multi-constraint structure ensures that all specifications are treated with equal priority, and that no pass-or-fail specification is unnecessarily optimized. To maximize the probability of reaching optimality and to obtain a systematic synthesis of system response with respect to parameters, it is proposed to consider a maximum number of major parameters as design variables, with a minimal number of preset constants, and to incorporate power circuit and controller model into one comprehensive model. Component parameters are considered discrete and defined by a database. Special challenges to implement the proposed optimization scheme, such as a large and discrete variable space and a model that contains discontinuous and indifferentiable functions, are solved by using stochastic search methods such as genetic algorithms (GA), and model simplification by linear regression. Two optimization schemes are proposed with different allocations of computational complexity, human expertise, and stochastic uncertainty. They provide flexibility for practical designers to optimize a system according to specific requirements and situations. Extensive modeling is performed for both the VR and VSI systems. Model simplification by linear regression is investigated. Detailed guidelines are described with definitions of regression parameters. Regression errors are analyzed and satisfactory accuracy is achieved by optimizing parameters using GA. Case studies for VR and VSI design using the two proposed schemes are discussed. Improved design solutions and shortened optimization processing time are achieved. Informative analysis treating the optimization system together with the database is presented.
Issue Date:2012-09-18
Rights Information:Copyright 2012 Yingying Kuai
Date Available in IDEALS:2012-09-18
Date Deposited:2012-08

This item appears in the following Collection(s)

Item Statistics