제주대학교 Repository

Boundary Estimation in EIT using Oppositional Biogeography-Based Optimization

Metadata Downloads
Abstract
전기 임피던스 단층촬영법(EIT)은 비침습성 영상 방법으로, 산업 분야뿐만 아니라 의료 분야에서도 활발히 연구되어 왔다. 그러나 EIT를 사용하여 도전율 분포를 복원하기 위한 역문제 알고리즘들의 성능은 종종 차선(sub-optimal)이 된다. 측정 잡음에 대한 EIT의 높은 민감도, 반올림 오차(rounding-off errors), EIT 문제가 갖는 고유의 부정치성 성질 그리고 전역 최소치 대신에 국부 최소치로의 수렴 등등 이런 요소들로 인해 낮은 성능을 보인다. 게다가, 대부분의 역문제 알고리즘들의 성능은 초기치의 선택뿐만 아니라 gradient matrix의 정확한 계산에 크게 좌우한다. 이런 요인들을 고려해 보면, 정확한 해에 도달하기 위한 효과적인 최적화 알고리즘이 요구된다. 본 논문에서는 2D EIT를 사용하여 대상 도메인 내부에서 표적들의 경계면에 대한 모양과 크기 그리고 위치를 추정하기 위해 oppositional biogeography 기반의 최적화(OBBO) 기법을 제시하고 있다. 표적들의 경계면은 truncated 푸리에 급수의 계수로 표현되는데, 이 때 표적들의 도전율 값은 선험적으로 알려져 있다고 가정한다. OBBO 기법은 흉부와 같은 구조 내부에서 정적 조직의 경계면들을 복원하는데 처음으로 적용된다. 그리고 나서 움직이는 심장의 경계면을 추정하는데 동적 OBBO 기법이 적용된다. 그리고 정적뿐만 아니라 동적 경우에 대해, 반복적인 수치적 시뮬레이션과 실험 데이터를 사용하여 알고리즘의 강인성(robustness)을 입증하였다. OBBO 기법을 사용하여 추정된 파라미터들에 대해 폭넓은 통계학적 분석을 하였고 전형적인 mNR 알고리즘과 EKF 알고리즘과 비교하였다. 전형적인 알고리즘들과 비교했을 때 OBBO 기법이 우수한 성능을 보였다. 게다가, OBBO 기법은 측정 잡음과 경계면의 크기와 위치에 대한 초기치에 강인하였고 도전율의 선험적 지식이 정확하지 않을 때도 타당한 해를 제공하였다.
Electrical impedance tomography (EIT) is a non-invasive imaging modality which has been actively studied for its industrial as well as medical applications. However, the performance of the inverse algorithms to reconstruct the conductivity images using EIT is often sub-optimal. Several factors contribute to this poor performance which includes high sensitivity of EIT to the measurement noise, the rounding-off errors, the inherent ill-posed nature of the problem and the convergence to a local minimum instead of the global minimum. Moreover, the performance of many of these inverse algorithms heavily relies on the selection of initial guess as well as the accurate calculation of a gradient matrix. Considering these facts the need of an efficient optimization algorithm to reach the correct solution cannot be overstated. This thesis presents an oppositional biogeography-based optimization (OBBO) technique to estimate the shape, size and location of the region boundaries inside an object domain using 2D EIT. The region boundaries are expressed as coefficients of truncated Fourier series while the conductivities of the regions are assumed to be known a priori. OBBO is first applied to reconstruct the static organ boundaries inside a chest like structure, when a full set of independent measurements is available at the measurement instance. A dynamic version of OBBO is then applied to estimate the non-stationary heart boundaries, assuming that only a partial subset of EIT measurement frame is available at a particular instance. The robustness of the algorithm, for static as well as dynamic cases, has been verified - first through repetitive numerical simulations by adding randomly generated measurement noise to the simulated voltage data, and then with the help of experimental setups. An extensive statistical analysis of the estimated parameters using OBBO and its comparison with the traditional mNR and EKF algorithms is presented. OBBO has shown far superior performance as compared to the traditional algorithms. Furthermore, it has been found that OBBO is robust to measurement noise and the initial guess of the size and location of the boundaries as well as it offers reasonable solution when the a priori knowledge of the conductivity is not very accurate.
Author(s)
에마르라시드
Issued Date
2011
Awarded Date
2011. 8
Type
Dissertation
URI
http://dcoll.jejunu.ac.kr/jsp/common/DcLoOrgPer.jsp?sItemId=000000005577
Alternative Author(s)
Ahmar Rashid
Affiliation
제주대학교
Department
대학원 에너지응용시스템학부 전자공학전공
Advisor
김경연
Table Of Contents
Acknowledgements i
Abbreviations and notations ii
List of figures viii
List of tables xiii
초록 xiii
1. Introduction 1
1.1 Motivation and problem description 1
1.2 Boundary estimation in EIT 1
1.3 Dissertation outline 5
2. Electrical impedance tomography 6
2.1 EIT forward problem 6
2.2 EIT inverse problem 9
2.2.1 The boundary representation 11
3. Evolutionary algorithms 13
3.1 Basics of evolutionary algorithms 14
3.2. Standard evolutionary algorithms 18
3.2.1 Genetic algorithms 18
3.2.2 Evolutionary strategies 19
3.2.3 Evolutionary programming 20
3.2.4 Genetic programming 20
3.3 Advanced evolutionary algorithms 21
3.3.1 Differential evolution algorithm 21
3.3.2 Biogeography based optimization 23
3.3.2.1 BBO algorithm 25
3.4 Hybrid evolutionary algorithms 27
3.4.1 Biogeography-based optimization combined with evolutionary strategies 27
3.4.2 Oppositional-biogeography-based optimization 28
3.4.2.1 OBBO algorithm 29
4. OBBO applied to reconstruct organ boundaries in human thorax using EIT 32
4.1 Biogeography-based optimization applied to EIT 33
4.2 OBBO applied to EIT boundary estimation 35
4.3 Results and discussions 39
4.3.1 Numerical results 39
4.3.2 Experimental results 56
5. Dynamic optimization 65
5.1 The uncertainties in DOPs 65
5.3 Important features of DOPs 66
5.3.1 Severity of change 67
5.3.2 Frequency of change 67
5.3.3 Observability and detectability of change 67
5.3.4 Dynamics of change 67
5.4 Desired characteristics of a DEA suitable to solve a DOP 68
5.5 Evolutionary approaches to solve DOPs 70
5.5.1 Reinitialization 70
5.5.2 Memory-based approaches 70
5.5.3 Multiple population approaches 72
5.5.4 Mutation and self-adaptation 73
5.5.5 Local variation 74
5.5.6 Diversity preserving techniques 74
6. Dynamic OBBO with variable relocation 76
6.1 Variable relocation 76
6.2 Dynamic OBBO algorithm 80
7. Dynamic OBBO applied to dynamic heart boundary estimation using EIT 83
7.1 Results and discussions 83
7.1.1 Numerical results 83
7.1.2 Experimental results 93
Conclusions 99
Summary 101
References 102
Degree
Doctor
Publisher
제주대학교
Citation
에마르라시드. (2011). Boundary Estimation in EIT using Oppositional Biogeography-Based Optimization
Appears in Collections:
Faculty of Applied Energy System > Electronic Engineering
공개 및 라이선스
  • 공개 구분공개
파일 목록

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.