Performance-Driven Surrogate Modeling of High-Frequency Structures

✍ Scribed by Slawomir Koziel, Anna Pietrenko-Dabrowska

• Publisher: Springer
• Year: 2020
• Tongue: English
• Leaves: 411
• Category: Library

✦ Synopsis

This book discusses surrogate modeling of high-frequency structures including antenna and microwave components. The focus is on constrained or performance-driven surrogates. The presented techniques aim at addressing the limitations of conventional modeling methods, pertinent to the issues of dimensionality and parameter ranges that need to be covered by the surrogate to ensure its design utility. Within performance-driven methodologies, mitigation of these problems is achieved through appropriate confinement of the model domain, focused on the regions promising from the point of view of the relevant design objectives. This enables the construction of reliable surrogates at a fraction of cost required by conventional methods, and to accomplish the modeling tasks where other techniques routinely fail. The book provides a broad selection of specific frameworks, extensively illustrated using examples of real-world microwave and antenna structures along with numerous design examples. Furthermore, the book contains introductory material on data-driven and physics-based surrogates. The book will be useful for the readers working in the area of high-frequency electronics, including microwave engineering, antenna design, microwave photonics, magnetism, especially those that utilize electromagnetic (EM) simulation models in their daily routines.

• Covers performance-driven and constrained modeling methods, not available in other books to date;
• Discusses of a wide range of practical case studies including a variety of microwave and antenna structures;
• Includes design applications of the presented modeling frameworks, including single- and multi-objective parametric optimization.

✦ Table of Contents

Preface
Acknowledgments
Contents
Chapter 1: Introduction
References
Chapter 2: Basics of Data-Driven Surrogate Modeling
2.1 Overview
2.2 Design of Experiments
2.2.1 Factorial Designs
2.2.2 Space-Filling Designs
2.2.3 Sequential Sampling
2.3 Modeling Techniques
2.3.1 Polynomial Regression Models
2.3.2 Radial Basis Functions
2.3.3 Kriging
2.3.4 Support Vector Regression
2.3.5 Artificial Neural Networks
2.3.6 Fuzzy Systems
2.3.7 Polynomial Chaos Expansion
2.3.8 Other Methods
2.4 Model Validation
References
Chapter 3: Physics-Based Surrogate Modeling
3.1 Overview
3.2 Low-Fidelity Modeling
3.2.1 Principal Properties and Techniques
3.2.2 Variable-Resolution and Variable-Accuracy Modeling
3.2.3 Variable-Fidelity Physics Modeling
3.2.4 Low-Fidelity Model Selection
3.3 Physics-Based Surrogates: Basic Concepts
3.4 Response Correction Models
3.4.1 Global Modeling Using Multipoint Space Mapping
3.4.2 Space Mapping with a Function Approximation Layer
3.4.3 Multipoint Output Space Mapping
3.4.4 Surrogate Modeling Using Generalized Shape-Preserving Response Prediction
3.5 Feature-Based Modeling
3.5.1 Feature-Based Modeling for Statistical Analysis
3.5.2 Feature-Based Modeling of Antenna Input Characteristics
3.6 Physics-Based Surrogates for Optimization
3.6.1 Space Mapping
3.6.2 Approximation Model Management Optimization
3.6.3 Manifold Mapping
3.6.4 Shape-Preserving Response Prediction
3.6.5 Adaptively Adjusted Design Specifications
3.6.6 Feature-Based Optimization
3.6.7 Adaptive Response Scaling
References
Chapter 4: Design-Oriented Modeling of High-Frequency Structures
4.1 Data-Driven Modeling by Constrained Sampling
4.1.1 Uniform Versus Constrained Sampling
4.1.2 Modeling Procedure
4.1.3 Illustration Examples
4.1.3.1 Dielectric Resonator Antenna
4.1.3.2 Planar Inverted-F Antenna
4.2 Design-Oriented Constrained Modeling for Operating Frequency and Substrate Parameters
4.2.1 Modeling Procedure
4.2.2 Case Study: Ring Slot Antenna
4.2.3 Application Examples and Experimental Validation
4.3 Constrained Feature-Based Modeling of Compact Microwave Structures
4.3.1 Case Study. RRC and Response Features
4.3.2 Modeling Methodology
4.3.3 Numerical Verification and Application Case Studies
References
Chapter 5: Triangulation-Based Constrained Modeling
5.1 Reference Designs
5.2 Surrogate Model Domain Definition
5.3 Surrogate Model Construction
5.4 Demonstration Case Studies
5.4.1 UWB Monopole Antenna
5.4.2 Uniplanar Dipole Antenna
5.4.3 Miniaturized Microstrip Coupler
5.5 Uniform Sampling Methods for Triangulation-Based Modeling
5.5.1 Uniform Sampling Scheme
5.5.2 Demonstration Examples
References
Chapter 6: Nested Kriging Modeling
6.1 Modeling Methodology
6.1.1 Objective Space: Geometry of Optimum Design Set
6.1.2 Reference Designs and Level I Surrogate
6.1.3 Surrogate Model Domain
6.1.4 Level II Surrogate
6.1.5 Design of Experiments
6.2 Demonstration Examples
6.2.1 Uniplanar Dipole Antenna (Antenna I)
6.2.2 Ring Slot Antenna (Antenna II)
6.2.3 Miniaturized Rat-Race Coupler (RRC)
6.2.4 Impedance Matching Transformer
6.3 Application Case Studies: Design Optimization
6.3.1 Optimization Methodology and Initial Design
6.3.2 Results for Antennas I and II
6.3.3 Design Optimization of RRC and Impedance Transformer
6.4 Improved Design of Experiments for Nested Kriging Surrogate
6.4.1 Modified Design of Experiments Procedure
6.4.2 Demonstration Case Studies
References
Chapter 7: Feature-Based Constrained Modeling
7.1 Modeling Framework: Incorporating Response Features into Nested Kriging Surrogates
7.1.1 Response Features
7.1.2 Level I Surrogate
7.1.3 Surrogate Model Construction
7.1.4 Modeling Framework
7.2 Demonstration Case Study I: Dual-Band Dipole Antenna
7.2.1 Test Case and Problem Statement
7.2.2 Results
7.2.3 Applications
7.3 Demonstration Case Study II: Triple-Band Dipole Antenna
7.3.1 Test Case and Problem Statement
7.3.2 Results
7.3.3 Applications
7.4 Demonstration Case Study III: Compact Microwave Coupler
7.4.1 Test Case and Problem Statement
7.4.2 Results
7.4.3 Applications
References
Chapter 8: Constrained Modeling Using Principal Component Analysis
8.1 Modeling Using Domain Confinement and Principal Components
8.1.1 Design Space and Objective Space
8.1.2 Reference Designs: Principal Components
8.1.3 Surrogate Model Domain
8.1.4 Design of Experiments: Surrogate Model Construction
8.1.5 Surrogate Model Optimization
8.2 Demonstration Case Study I: Uniplanar Dipole Antenna
8.2.1 Antenna Structure and Problem Statement
8.2.2 Numerical Results and Benchmarking
8.2.3 Application Examples
8.3 Demonstration Case Study II: Ring Slot Antenna
8.3.1 Antenna Structure and Problem Statement
8.3.2 Numerical Results and Benchmarking
8.3.3 Application Examples
8.4 Demonstration Case Study III: Impedance Matching Transformer
8.4.1 Transformer Structure and Problem Statement
8.4.2 Numerical Results and Benchmarking
8.4.3 Application Examples
8.5 Demonstration Case Study IV: Rat-Race Coupler
8.5.1 Coupler Structure and Problem Statement
8.5.2 Numerical Results and Benchmarking
8.5.3 Application Examples
References
Chapter 9: Variable-Fidelity Performance-Driven Modeling
9.1 Variable-Fidelity Performance-Driven Modeling: Procedure Overview
9.2 Variable-Fidelity Performance-Driven Modeling Using Co-kriging
9.2.1 Co-kriging
9.2.2 Illustration Case Study: Impedance Matching Transformer
9.3 Variable-Fidelity Performance-Driven Modeling Using Two-Level Gaussian Process Regression
9.3.1 Gaussian Process Regression (GPR) Basics
9.3.2 Two-Stage GPR Modeling
9.3.2.1 Two-Stage GPR: First Stage
9.3.2.2 Two-Stage GPR: Second Stage
9.3.3 Case Studies
9.4 Variable-Fidelity Performance-Driven Modeling Using Space Mapping
9.4.1 Model Correction Using Space Mapping
9.4.2 Demonstration Case Study: Dual-Band Microstrip Dipole Antenna
9.5 Summary
References
Chapter 10: Constrained Modeling for Efficient Multi-objective Optimization
10.1 Multi-objective Design: Problem Formulation and Solution Methods
10.2 Multi-objective Optimization Using Surrogate Models
10.3 Multi-objective Optimization Using Triangulation-Based Modeling
10.3.1 Surrogate Model Domain
10.3.2 Demonstration Example and Results
10.4 Multi-objective Optimization Using Nested Kriging
10.4.1 Nested Kriging Modeling: Brief Recollection
10.4.2 Nested Kriging Modeling for Multi-Objective Design
10.4.3 Application Case Study I: Planar Yagi Antenna
10.4.4 Application Case Study II: Wideband Monopole Antenna
10.4.5 Application Case Study III: Impedance Matching Transformer
References
Chapter 11: Warm-Start Design Optimization
11.1 Accelerated Optimization Using Design Database
11.1.1 Design Database: Initial Design by Database Geometry Exploration
11.1.2 Optimization Procedure
11.1.3 Case Study I: Planar Yagi Antenna
11.1.4 Case Study II: Compact Rat-Race Coupler
11.2 Warm-Start Optimization Using Kriging Surrogates
11.2.1 Database Designs and Kriging Surrogates
11.2.2 Optimization Procedure I: TR Gradient Search
11.2.3 Optimization Procedure II: Iterative Correction Scheme
References
Chapter 12: Inverse Surrogates for Accelerated Simulation-Driven Design
12.1 Fast Dimension Scaling Using Inverse Surrogates
12.1.1 Problem Formulation
12.1.2 Inverse Model Construction
12.1.3 Scaling Procedure
12.1.4 Case Study I: Miniaturized Rat-Race Coupler
12.1.5 Case Study II: Bandwidth-Enhanced Patch Antenna
12.1.6 Case Study III: Scaling of Band-Notch UWB Antenna
12.2 Scaling for Multiple Operating Conditions
12.2.1 Problem Formulation
12.2.2 Inverse Model Construction
12.2.3 Scaling Procedure
12.2.4 Iterative Correction Scheme
12.2.5 Case Study I: Dual-Band Miniaturized Coupler
12.2.6 Case Study II: Dual-Band Antenna-Re-design for Substrate Parameters
12.2.7 Case Study III: Triple-Band Dipole Antenna
12.2.8 Case Study IV: Four-Objective Scaling of Compact RRC
12.3 Advanced Correction Schemes
12.3.1 Fast Re-design of Compact Couplers: Corrected Scaling for Power Split Control
12.3.1.1 Post-scaling Power Split Correction
12.3.1.2 Case Study and Numerical Results
12.3.1.3 Experimental Validation
12.3.2 Optimization-Based Forward Power Split Correction
12.3.2.1 Basic Scaling Procedure
12.3.2.2 Post-scaling Correction Using Forward Sensitivity Model
12.3.2.3 Case Study and Numerical Results
12.3.2.4 Experimental Validation
References
Chapter 13: Summary and Conclusion
Index