Development of Efficient Aircraft Design Process Using Decision-Making and Uncertainty Modeling

나잉슈미얏 Konkuk University, Graduate School of Konkuk Unive 2014 국내박사

Design knowledge and experience are the bases to carry out aircraft design tasks due to the high complexity and integration of the tasks during this phase. When carrying out the same task, different designers may need individual strategies to fulfill their own demands. Thus, efficient aircraft design process considering knowledge-based aircraft configuration decision-making and uncertainty modeling for possibility based design optimization is developed in this research. The advantage of database application is used in the development of aircraft configuration decision-making modeling process. The decision-making modeling process is performed based on fighter/attack aircraft database which is systematically constructed using public domain sources and approximation techniques. Decision-making models are developed by means of data mining as well as approximation models based on the developed database to facilitate initial aircraft configuration decisions. The developed decision-making models are applied to derive initial configuration of F-CK-1 aircraft. Optimization process is carried out using derived initial design. As a consequence of application of the developed process, optimum design is got with less iteration steps by performing the optimization process using derived initial configuration than the optimization process with random selection of initial configuration. Effectiveness, usability and simplicity of the process can be seen by demonstrating the developed process as described in case study. Deterministic optimum designs, which are frequently pushed to the limits of design constraint boundaries, leaving little or no room for tolerances (uncertainty) in modeling, simulation uncertainties, and manufacturing imperfections. Deterministic optimum designs that are obtained without consideration of uncertainty may result in unreliable designs. Thus, aircraft design process which is implemented considering uncertainty by possibilistically is considered in this research. Possibility Base Design Optimization (PBDO) is performed by modeling each source of uncertainty using membership function. In PBDO process constraints are evaluated possibilistically using Performance Measure Approach (PMA) that improves numerical efficiency and stability. In this research, uncertainties from accuracy of analysis methods are taken into account in optimization process. These uncertainty intervals are derived by comparing analysis outputs and known data. After that membership functions are constructed based on uncertain intervals. Two different type of membership functions are applied in this research. Several case studies are performed to demonstrate efficiency of proposed process. 항공기 설계에 있어 설계 지식과 경험은 여러 분야를 통합해야 하는 높은 복잡성 때문에 수행되어야 하는 기초이다. 같은 작업을 수행할 시, 각 설계자는 설계 요구를 충족시키기 위한 자신만의 전략이 필요할 수 있다. 그러므로 불확정성 기반의 설계 최적화를 위한 모델링과 지식기반(knowledge-based) 형상 결정을 고려한 효과적인 항공기 설계 프로세스가 이번 연구에서 개발되었다. 데이터베이스 어플리케이션의 장점이 항공기 형상 결정 모델링 프로세스 개발에 적용되었다. 의사 결정 모델 프로세스는 공개된 데이터와 근사화 기법을 이용하여 체계적으로 구축된 전투기 및 공격기에 대한 데이터베이스를 기반으로 수행되었다. 의사 결정 모델은 초기 항공기 형상을 결정하는데 용이하기 위해 개발된 데이터베이스 기반의 근사모델 및 데이터마이닝의 일환으로 개발되었다. 개발된 의사 결정 모델은 F-CK-1 항공기 초기 형상을 구현하기 위해 적용된다. 최적화 프로세스는 구현된 초기 형상을 사용하여 수행되었다. 개발 프로세스의 적용 결과로써, 초기 형상을 무작위로 선택하는 것보다 구현된 초기형상을 사용하는 것이 최적화를 통한 반복 해석 횟수가 보다 적음을 알 수 있다. 다른 경쟁 프로세스와의 비교를 통해 개발 프로세스의 효율성, 유용성 및 공정의 단순화의 척도를 알 수 있다. 결정론적 최적 설계는 설계 가능 공간에서 제약 영역의 경계부분에 위치하기 때문에 모델링의 불확정성에 의한 허용 오차, 시뮬레이션 불확정성, 제조 결함 등으로 인해 요구되는 설계 공간을 허용하지 않는다. 불확정성을 고려하지 않는 결정론적 최적 설계는 신뢰할 수 없는 결과물을 내놓기도 한다. 따라서 본 연구에서는 불확정성을 고려한 항공기 설계 프로세스가 고려되었다. 가능성 기반의 설계 최적화는 (Possibility Based Design Optimization) 퍼지 넘버의 소속함수를 사용하여 불확정성 요인을 각각 모델링 하여 수행된다. 가능성 기반 설계 최적화 프로세스에서는 제약조건이 수치적 효율과 안정성을 향상시키는 성능측정법(Performance Measure Approach)를 사용하여 가망성을 토대로 평가한다. 본 연구에서는 해석 방법의 정확도를 통해 불확실성을 최적화 과정에서 고려하였다. 변수의 불확실도를 나타내는 간격은 해석 결과와 기준 데이터를 비교하여 구할 수 있다. 이후 이 간격을 기준으로 소속함수를 만들어 낼 수 있다. 본 연구에서는 두 개의 서로 다른 소속함수가 적용되었다. 또한 제한된 프로세스의 효율성을 설명하기 위하여 여러 가지 사례 연구가 진행되었다.

Robust Methods for Aeropropulsive Design Optimization

Yildirim, Anil ProQuest Dissertations & Theses University of Mich 2021 해외박사(DDOD)

Design and integration of advanced propulsion systems play a critical role in the development of environmentally sustainable aircraft. Even though these advanced technologies, such as boundary layer ingestion, offer significant aeropropulsive benefits, their design is challenging due to the tightly coupled nature of these systems, as well as the lack of previous experience with their design. Aeropropulsive design optimization offers a promising solution to these design problems where coupled models are used to maximize the aeropropulsive benefits of these propulsion systems. Despite its advantages, the use of aeropropulsive design optimization has been limited due to the shortcomings in robustness of existing design optimization approaches. In this work, we address several key shortcomings in existing design optimization approaches and introduce robust methods for aeropropulsive design optimization. Some of these developments target the shortcomings in CFD-based design optimization such as geometric parameterization and CFD solver robustness, while the rest of the developments focus on coupled aeropropulsive model development and optimization. Using all of these developments, we performed a series of coupled aeropropulsive design optimizations to quantify the benefit of boundary layer ingestion to the STARC-ABL concept. This first of a kind design study was enabled by the robustness of the overall aeropropulsive design framework we developed herein and will be useful for guiding future design studies of boundary layer ingesting propulsion systems. The robust methods presented in this thesis are crucial for future CFD-based design optimization problems, including aeropropulsive design optimization. These advancements bring us closer to using the ever-growing power of scientific computing in the process of designing future environmentally sustainable aircraft.

Development of rotor structural design optimization framework for compound rotorcraft with a lift offset

임재훈 서울대학교 대학원 2014 국내박사

Design optimization for the structural construction of a rotor system in a compound rotorcraft with a lift offset was conducted by applying a two-level procedure. In this optimization process, enhancements were made regarding the high-speed forward flight capability, vibration characteristics, and rotor blade weight of a compound rotorcraft. In this type of rotorcraft, the combination of a lift offset, a rigid rotor blade, and a variable rotor speed enables the achievement of satisfactory vibration characteristics and high forward speed performance. Optimized structural properties of the blade and chordwise geometries for each blade radial station were obtained by the present upper-level optimization scheme. By applying the proposed lower-level optimization scheme, an optimized cross-sectional configuration of a composite rotor blade was determined. For validation purposes, the design of an existing stiff-in-plane rotor in XH-59A compound rotorcraft was optimized by applying the present procedure. As a result, it was found that most of the objective function values were significantly improved. The influence of the optimization results on the conceptual design was evaluated. It is expected that the present procedure will be applicable for the next-phase advanced compound rotorcraft, which will target further high forward speed capability with low vibration characteristics. 본 논문에서는 Lift offset 로터와 보조추진장치가 장착된 복합형 회전익기의 로터 최적설계 프레임워크를 2 단계의 형태로 개발하였다. 개발된 최적화 프레임워크에 의해 고속전진비행에서의 복합형 회전익기의 성능, 진동, 중량이 향상되었다. lift offset 복합형 회전익기는 lift offset, 강성이 높은 로터 블레이드, 로터 회전속도 변화 등의 다양한 요소가 진동과 성능에 미치기 영향 때문에, 진동과 성능 모두를 만족시키는 설계를 도출하는 것이 난이하다. 상위 최적설계에서는 로터 블레이드의 최적화된 물성치와 시위길이가 도출된다. 상위 최적설계 프레임워크에서 도출된 시위길이, 질량 및 강성 값을 가지는 복합재 블레이드 단면의 설계가 하위 최적설계에서 수행되었다. XH-59A 복합형 회전익기의 자료를 이용하여 개발된 최적설계 프레임워크를 이용한 최적설계를 수행하였다. 그 결과 대부분의 목적함수가 상당히 향상되었음을 확인하였다. 로터 최적설계 결과가 회전익기 개념설계 결과에 미치는 영향을 알아보기 위해 lift offset 복합형 회전익기의 개념설계가 수행 가능한 개념설계 프레임워크가 개발되었다. 이를 이용하여 로터 최적설계 결과가 개념설계에 미치는 영향을 확인하여 본 결과 향상된 로터 성능 및 중량으로 인해 총 중량이 감소한 회전익기가 설계됨을 확인하였다.

Multidisciplinary design optimization of aircraft wing structures with aeroelastic and aeroservoelastic constraints

정상영 University of Oklahoma 1999 해외박사

Design procedures for aircraft wing structures with control surfaces are presented using multidisciplinary design optimization. Several disciplines such as stress analysis, structural vibration, aerodynamics, and controls are considered simultaneously and combined for design optimization. Vibration data and aerodynamic data including those in the transonic regime are calculated by existing codes. Flutter analyses are performed using those data. A flutter suppression method is studied using control laws in the closed-loop flutter equation. For the design optimization, optimization techniques such as approximation, design variable linking, temporary constraint deletion, and optimality criteria are used. Sensitivity derivatives of stresses and displacements for static loads, natural frequency, flutter characteristics, and control characteristics with respect to design variables are calculated for an approximate optimization. The objective function is the structural weight. The design variables are the section properties of the structural elements and the control gain factors. Existing multidisciplinary optimization codes (ASTROS and MSC/NASTRAN) are used to perform single and multiple constraint optimizations of fully built up finite element wing structures. Three benchmark wing models are developed and/or modified for this purpose. The models are tested extensively.

Isogeometric configuration design optimization of built-up structures in generalized curvilinear coordinates

이승욱 서울대학교 대학원 2017 국내박사

In the thesis, an isogeometric configuration design optimization method for built-up and shell structures based on generalized curvilinear coordinate (GCC) is developed.We derive the isogeometric configuration sensitivity of the Mindlin plates by using the material derivative and adjoint approaches. This is utilized in the configuration design optimization that includes a variation of design components in its shape and orientation. Due to the non-interpolatory property of the Non-Uniform Rational B-Spline(NURBS) basis functions, a mismatch of patches in the built-up structures could occur during the isogeometric design optimization, which can be easily resolved using transformed basis functions. Also, isogeometric configuration sensitivity of shell structures is derived with separating shape and orientation effect. For the shell structures, the design is generally affected by the coupled effect of shape and orientation variations. But, at each material point, exact rotational transformation can be calculated using GCC system and isogeometric approach. An orientation variation is identified as rotational transformation of body-fixed local curvilinear coordinate system. Configuration design sensitivity of shell structure is verified by comparing finite difference sensitivity. And, configuration design optimization is performed for built-up and shell structures The built-up structure is made by combining various elements such as plate, beam and shell. When optimizing the built-up structure, configuration design sensitivity is necessary because shape and orientation variations simultaneously happen. Moreover, in theisogeometric analysis (IGA), the control points play the role of design variables so that no more design parameterization is necessary. Hence, the IGA-based one is suitable for the configuration optimization of the built-up structures By the IGA, the NURBS basis function in computer aided design (CAD) system is directly utilized in the response analysis, which enables the seamless incorporation of higher continuity and exact geometry such as curvature and normal vector into the computational framework. IGA provides a more accurate sensitivity of complex geometries including higher order geometric effects such as normal and curvature information. The impact of exact curvature in the bending problem of Mindlin plates on the configuration design sensitivity is demonstrated through numerical examples. The obtained design sensitivity is further utilized in the configuration design optimization of built-up structures. Configuration design sensitivity analysis (DSA) for shell structure based on GCC system is formulated using direct differentiation method (DDM). In the design sensitivity of the curved structure such as curved beam and shell, it is difficult to separate shape and orientation contributions. They affect design variation at the same time. We divide shape and orientation effects through exact transformation between two local curvilinear coordinate in the original design and perturbed design. It can be possible to calculate accurate sensitivity although amount of design perturbation is large.

Integrated Support System for Decision-Making in Design Optimization

정민중 University of Tokyo 2003 해외박사

Recently, optimization methods are being used extensively in many engineering fields. In single-objective optimization, a single optimum solution may not give enough information to help engineers perform Decision-Making (DM) in design. That is because it is usually dicult or impossible to formulate an optimization problem to include all factors that in uence the choice of a particular design. On the other hand, multi-objective optimizations yield quite a good deal of Pareto solutions in complex function and parameter spaces. Pareto solutions have several barriers to understanding their characteristics and to apply them directly to practical design. Normally the amount of Pareto solutions is very large. Moreover, any obtained Pareto solution cannot be a representative one solution without additional information. Therefore, if alternative solutions including Parato, quasi- Pareto and feasible solutions are provided to engineers with their engineering information, they can use the information to choose the best overall design. To understand the meaning of many solutions in high dimension, it helps if their complexity can be reduced. Clustering algorithms and reduction of dimensions are often used to deal with this task. Thus one should use sophisticated clustering and dimension reduction algorithms to interpret optimization solutions in many engineering problems. In addition, one has to visualize the result in a manner useful for engineers. However, it is not easy to handle the many algorithms and methods of optimization, clustering, dimension reduction and visualization without any supporting tools. The goal of this research is to develop and implement an Integrated Support System for DM in design optimization. An evolution-based optimizer, evolutionary clustering algorithms and visualization methods were implemented with conventional dimension reduction, clustering algorithms and visualization methods to interpret the information for DM. An enhanced evolution-based optimizer and a clustering algorithm were developed as primary modules of the DM support system. It was proposed that the idea of satisfactory design space and clustering interpretation for DM. Various clustering algorithms and dimension reduction approaches have been used to explore sensitive design parameters for DM. Details for multi-dimensional visualization methods of optimization solutions are discussed. Moreover, features of distance measures extracting and comparing the characteristics of design solutions are presented. The usefulness of the proposed methods has been tested against numerical and analytical examples. The eectiveness of analysis procedures using the methods was demonstrated. The procedures were applied to three challenging applications of design optimization. The design optimization of microelectrostatic actuator for an optical lens was used for the rst application. The analysis of satisfactory design spaces and their clustering interpretation guided the nal decision for the actuator design. Second, a 90-parameter optimization of turbine blade shape optimization showed interpretive approaches using clustering and dimension reduction. Sensitive design parameters of blade shape were gathered up for DM. As the third application, multi-objective optimization of heat pipe shape of artificial satellite was carried out. Key concepts for heat pipe design were explored and extracted using a synchronous visualization method and a hybrid distance measure. The proposed DM support system has been developed focusing into not only maintainability and extensibility of the system, but also compatibility with conventional design simulation such as a finite element analysis.

A New Method of Decomposition Based Multidisciplinary Design Optimization Architecture Using Optimal Design Sensitivity : 최적 민감도를 이용한 분리기반 다분야통합설계최적화 방법에 관한 연구

Jeong, Heeseok Graduate School of Yonsei University 2001 국내박사

Multi-disciplinary Design Optimization of Rotor Blades using Data-driven Aero-Structure Performance Modeling

이다운 서울대학교 대학원 2024 국내박사

Multi-disciplinary Design Optimization of Rotor Blades using Data-driven Aero-Structure Performance Modeling Dawoon Lee Department of Aerospace Engineering The Graduate School Seoul National University Vertical take-off and landing aircraft technology has being matured to achieve clean-sky and safe travel, however there are growing demands for enhanced mission capabilities in both military and commercial applications. One of the key technologies for the successful development of new types of vertical take-off and landing aircraft is the advancement of the rotor system, which is the main propulsor enabling versatile flight. Rotor blade design technologies have also evolved over the past several decades to achieve high aerodynamic efficiency and a long life-cycle with low vibration, however most of the rotor blade design studies have been divided into two separate disciplines: structural design and aerodynamic design. The structural dynamics and aerodynamics of rotor blades are interdependent as known as aeroelastic phenomena. The structural dynamic response deforms the blade shape, and the resulting change of airloads interacts with the blade structure. Although the blades can be sequentially optimized from a structural and aerodynamic perspective and then combined through an iterative design process, there is still room for further design improvement. There is no doubt that a comprehensive design approach is necessary to address the interdisciplinary nature of rotor blades, however it also presents challenging issues for design optimization. Not only are most of the design metrics calculated by aeroelastic analysis inherently nonlinear, but they are also coupled with excessive design constraints necessary to ensure durability and stability. Identifying the optimal direction of nonlinear objective functions becomes difficult due to the narrow and discontinuous feasible regions in the high-dimensional design space. The efficiency of design optimization becomes poor, and the resulting increase in computational cost limits extensive design exploration that considers both structural and aerodynamic design parameters. Therefore, an efficient multi-disciplinary design method that allows design exploration within an affordable computational cost should be developed to simultaneously improve the structural and aerodynamic performance of rotor blades. High-performance computing enables the generation of extensive data set of two- dimensional (2D) performance of rotor blades: beam properties for various composite cross-sections, aerodynamic coefficient table over a wide range of angles of attack and Mach number for various airfoil shapes. These 2D performance data can be used to construct a low-cost, data-driven performance prediction model by applying the advanced machine learning techniques. The pre-trained performance prediction models facilitate the generation of a three-dimensional numerical model for the aeroelastic analysis of rotor blades. The data-driven performance models are integrated into the design framework using the rotorcraft comprehensive analysis (CA) solver. The integrated design framework was developed for an efficient design optimization by eliminating the time-consuming 2D performance analysis process. Finally, the multi-disciplinary design framework for the composite blade structure and aerodynamic blade shape design was developed using a data-driven aero- structure performance prediction model, and the main results are as follows. First, the composite blade structure design framework was developed and applied to the low vibration blade design problem. To address the design complexity due to the nonlinearity of the aeroelastic responses combined with a narrow feasible region within a high-dimensional design space, an improved surrogate model, Cluster-based Kriging, was employed to prevent falling into local optima due to sampling bias and overfitting. Second, the efficient rotor blade airfoil design framework was developed by integrating the reduced-order model for airfoil performance prediction (Airfoil Brain), lifting-line based CA solver, and high-fidelity computational fluid dynamic (CFD) solver. The Airfoil Brain can instantly generate the lifting-line model of a rotor blade consisting of arbitrary airfoils. Also, the rotor aerodynamic model of the CA solver is calibrated by using the high-fidelity CFD data, the design process was named to the ABC2 framework. Finally, the above structural and aerodynamic design frameworks were integrated for multi-disciplinary design optimization of rotor blades. Concurrent aerodynamic and structural design optimization was performed to maximize aerodynamic efficiency and minimize rotor hub vibration. The underlying physics of the improved performance are also investigated from both structural and aerodynamic perspectives, confirming the effectiveness and superiority of the present design framework. The design framework developed in this dissertation can derive a true multi- disciplinary optimal composite blade structure and aerodynamic blade shape by simultaneously optimizing the design parameters for both disciplines and improving the design objectives for both disciplines. It can also effectively reduce the cost required for a sequential iterative process design of structural dynamics and aerodynamics. This multi-disciplinary design framework using data-driven performance models provides not only optimum blade design, but also design sensitivity information. It is expected to provide valuable insights for the preliminary design phase or system-level engineering of new vertical take-off and landing aircraft. 도심항공교통 및 차세대 수직 이착륙기에 대한 개발 수요가 증가하며, 성공적인 비행체 개발을 위한 핵심 기술로 주 추력장치인 로터 블레이드의 진동 저감 및 공력 성능 개선이 요구되고 있다. 구조-공력 상호작용이 뚜렷하고 서로의 성능 지수에 밀접한 연관이 있는 로터 블레이드는 구조-공력 설계변수와 성능 지수를 함께 고려한 다학제간 최적설계가 필수적이다. 그러나 지난 수 세기 동안 구조 설계 연구와 공력 설계 연구는 분리되어 발전해왔으며, 이들을 반복 수행하는 설계전략을 사용하여 시간과 자원이 많이 소요될 뿐 아니라 다학제간 최적 설계안을 도출하는데 어려움이 있었다. 구조-공력 설계변수를 동시에 변경하며 진동 저감과 공력 성능 개선을 동시에 이루는 통합 설계 연구는 과도한 계산비용으로 인해 제한적으로 수행되어왔으며, 이를 효율적으로 해결할 수 있는 새로운 다학제간 최적설계 방법론이 필요하다. 본 연구에서는 로터 블레이드 구조-공력 통합해석자와 데이터 기반 성능모델링 기법을 결합한 효율적인 다학제간 최적설계 방법론을 개발하였다. 구조-공력 통합해석자에서 3 차원 로터 블레이드 형상은 다수의 2 차원 형상의 조합인 탄성 보 모델과 양력선 모델로 생성되며, 이들의 모델링에 필요한 2 차원 구조, 공력 성능 데이터를 얻기 위해서는 과도한 계산비용이 요구된다. 이를 효율적으로 대체하기 위해 데이터 기반 구조, 공력 성능 예측모델을 각각 개발하였으며, 먼저 복합재 로터 블레이드의 단면구조가 변화함에 따라 2 차원 강성 및 관성치를 예측할 수 있는 데이터 모델을 개발하여 고정밀도 유한요소 해석을 수행하지 않고도 탄성 보 모델을 즉시 생성하였다. 또한 블레이드 익형의 형상이 변화함에 따라 2 차원 공력계수를 넓은 받음각과 마하 수 범위에서 예측할 수 있는 성능예측 모델을 개발하였다. 로터 블레이드의 양력선 모델을 추가적인 고정밀도 유동해석을 수행하지 않고 2 차원 성능예측 모델을 이용해 즉시 생성할 수 있도록 하였다. 즉, 로터 통합해석자에서 임의의 3 차원 로터 블레이드 해석 모델을 추가적인 2 차원 구조-공력 성능해석 없이 생성할 수 있기에, 이를 활용해 넓은 설계공간을 효율적으로 탐색할 수 있는 다학제간 최적설계 프레임워크를 개발하였다. 개발된 다학제간 최적설계 프레임워크를 로터 블레이드의 구조설계와 공력설계에 각각 적용하여 그 효율성과 유효성을 검증하였다. 먼저 복합재 로터 블레이드의 단면구조를 변화시키며 로터 허브 진동을 저감시키는 최적설계 연구를 수행하고, 그 결과에 대한 정밀한 설계분석을 바탕으로 진동저감을 유도하는 물리적인 이유를 규명하였다. 또한 로터 블레이드 공력 최적설계 프레임워크를 개발하여 블레이드 플랜폼 형상과 다중 익형의 형상설계를 통해 공력 효율을 극대화하는 최적설계 연구를 수행하였다. 최적설계 결과의 고정밀도 유동해석을 통해 본 프레임워크의 정확도를 검증하였으며, 설계된 블레이드와 개별 익형이 갖는 형상 특성과 공력 성능을 분석하여 그 효용성을 입증하였다. 개별적으로 검증된 두 가지 학제의 최적설계 프레임워크를 결합하여 구조-공력 설계변수를 동시에 변화시키며 진동과 공력 성능지수를 개선하는 통합 설계를 수행하였다. 제자리 비행과 전진비행에서 모두 공력 효율을 높이고 로터 허브 진동을 저감시키는 최적 설계를 통해 모든 성능을 개선시키는 로터 블레이드 구조 및 공력 형상을 도출하였다. 다학제간 최적설계를 통해 얻어진 최적 설계안에 대한 정밀한 설계분석을 수행하여 구조-공력 관점에서 성능이 개선된 물리적인 원인을 규명하였으며, 개발된 프레임워크의 효용성을 확인하였다. 본 학위논문에서 개발한 다학제간 최적설계 프레임워크는 블레이드 내부 복합재 단면구조와 익형 형상을 포함한 블레이드 외형을 동시에 최적화할 수 있으며, 최적설계 과정에서 각 학제에 유리한 물리현상을 충분히 활용하여 구조-공력 최적설계안을 도출함이 검증되었다. 따라서 로터 블레이드의 기본설계 단계에서 소요되는 구조 및 공력 설계의 반복 수행과정을 대체하여 시간 및 자원 효율을 높이는 동시에, 다학제간 최적 설계안을 도출해 더 높은 구조 및 공력 성능개선을 이루었다. 계산 효율적인 다학제간 최적설계 프레임워크는 다양한 개념과 임무조건 하에서 운용되는 새로운 종류의 수직이착륙기 개발에 더욱 큰 효과를 기대할 수 있다.

Reliability based design optimization of vehicle multimedia encapsulating module

박정민 Graduate School, Yonsei University 2004 국내석사

확정론적 방법을 사용한 최적 설계 방법은 통계학적 최적 설계 방법보다 덜 복잡하고 빠르게 수렴하는 장점이 있으나 설계 변수나 시스템 파라미터에 존재할 수 있는 불확정성으로 인해 발생할 수 있는 목적함수나 구속 조건의 변동을 고려하지 못한다는 단점으로 인하여 최적설계 결과에 대한 신뢰도의 적용성이 다소 떨어질 수 있다. 즉 통계학적 최적 설계 방법은 설계 변수와 시스템 파라미터의 불확정성이 목적함수와 제한 조건에 미치는 영향을 고려한다는 점에서 확정론적 최적 설계와 차이가 있다. 제품의 수명을 향상시키기 위해서는 다양한 사용 및 환경 조건하에서 평균 수명이 일관성 있게 길어야 한다. 제품의 수명은 설계 변수의 허용 오차나 제품의 사용 및 환경 조건에 영향을 받는다. 그러나, 생산자가 제품의 사용 및 환경 조건을 직접적으로 통제할 수 없을 뿐만 아니라 제품에 사용되는 부품은 허용 오차를 가지므로 제품 수명은 필연적으로 어떠한 산포를 가지게 된다. 신뢰성 기반 최적 설계는 구조 설계에 있어서 설계단계에서부터 신뢰성과 최적화를 동시에 고려하는 것에 그 목적이 있다. 본 논문에서는 무선 인터넷, GPS, 무선결제 단말기, 모바일 오피스 등 각각의 단말기 어플리케이션 기술들로 인하여 정보 검색, 자동차 네비게이션, 무선 전자 상거래, 문서 작성 등과 같은 사무를 볼 수 있게 환경을 구축하는 기구를 다룬다. 즉, 설계단계에서부터 신뢰성을 고려하여 최적설계를 수행하는 신뢰성 기반 최적 설계를 이용하여 차량 내에서 무선 인터넷을 즐길 수 있는 차량용 멀티미디어 탑재기구의 최적화된 설계 변수를 얻고자 한다. 일계이차 모멘트 방법에 의해 신뢰성을 평가하였고 성능 평가 함수를 근사적인 명시 함수, 즉 한계상태식으로 표현하고, 한계상태식이 불명확할 때 효율적으로 사용할 수 있는 실험계획법과 반응표면법을 사용하였다. Deterministic optimization methods have advantages simpler and more rapid process than optimization which contains stochastic character. But deterministic optimization methods couldn’t apply fluctuations of objective function and constraints well because of uncertainties from design variables and system parameters. It is different from deterministic optimization that stochastic optimization considers uncertainties of design variables and system parameters which have effects on objective function and constraints. Average life span of products should be long consistently without failure under varies usage and environment conditions. The life span of products is influenced by permitted error of design variables and environmental conditions on products. But designers couldn’t control the environments directly, and product parts have been permitted error definitely. Therefore life span of products has a statistic distribution inevitably. The object of reliability based design optimization is to combine reliability considerations and optimization procedure in a single framework for structural design from design step. This paper deals equipment which can create documents, search information, do car navigation system, and make wireless electronic commerce using each terminal application technology like wireless internet, GPS, wireless settlement, and mobile office. In this paper, I tried to design multimedia equipment loaded in vehicles, based on the reliability based design optimization considering reliability from design step and achieving optimal design. FOSM method is uses to construct limit state function, and DOE and RSM is utilized for exploring the design space and for constructing the regression models to facilitate the effective solution of optimization problems.

Global Variable Fidelity Methodology for Multidisciplinary Tailless UAV Design and Optimization

장막심 건국대학교 대학원 2015 국내박사

Aircraft conceptual design stage usually utilizes simplified analysis methods, empirical, and semi-empirical equations to establish the basic layout of an aircraft. More sophisticated high fidelity analysis methods deliver accurate physics based analysis. They are able to analyze physical phenomena infeasible by low fidelity analysis and provide significantly higher accuracy. However, high fidelity analysis methods often cannot be directly used in aircraft conceptual design due to its high computational cost. Therefore, there is a need for optimization approach that may combine benefits of different levels of analysis methods: high accuracy and low computational cost. This research introduces a global variable fidelity modeling (GVFM) algorithm and its application for single and multidisciplinary design optimization problems. The proposed approach combines well known concepts and methods to produce a new variable fidelity optimization methodology. GVFM combines such methods as: scaling based variable fidelity modeling, design of experiments for initial design space exploration, radial basis functions network for accurate approximation of a scaling model, and trust region management. Different scaling functions were evaluated to determine the most robust and accurate approach for variable fidelity optimization. In addition, integrated analysis code for tailless unmanned aerial vehicle (UAV) conceptual design is introduced. The integrated analysis code combines well known analysis methods, open source, and in-house analysis programs adopted for tailless UAV analysis. Special features of a tailless aircraft aerodynamics, stability and control are taken into account for a conceptual design. Directional stability and control is one of the weakest points of a tailless aircraft. Effect of a split-drag-rudder on aircraft stability is investigated using high fidelity aerodynamic analysis. An automated computational fluid dynamics analysis process is developed and integrated with UAV analysis code to demonstrate the GVFM algorithm. Subsonic and transonic airfoil design, and unmanned combat aerial vehicle conceptual design case studies demonstrate efficiency of the algorithm. All case studies show design improvement with relatively low number of high fidelity function evaluations. 항공기 개념설계 단계에서는 항공기의 기본형상을 확정하기 위해서 간단한 분석이나 경험적, 반 경험적 식들을 항상 사용한다. 더 정교한 고 정밀도 해석방법은 분석에 근거한 정확한 결과를 산출한다. 고 정밀도 해석은 저 정밀도 해석에서 분석할 수 없는 물리적 현상도 분석할 수 있으며, 저 정밀도에 비해 더 높은 정확도를 제공한다. 하지만, 개념설계 단계에서는 해석과정의 많은 계산 시간으로 인해 처음부터 고 정밀도 해석을 사용할 수 없다. 그렇기 때문에 고 정밀도와 저 정밀도의 장점인 높은 정확성과 적은 계산 시간을 결합한 최적화된 접근이 필요하다. 본 논문은 global variable fidelity modeling(GVFM) 알고리즘을 소개하고 단일 또는 종합적인 설계 최적화 문제에 적용한다. 제안된 접근방법은 새로운 다 정밀도 최적화 방법론을 만들어 내기 위해 잘 알려진 개념과 방법을 조합한다. GVFM은 scaling based variable fidelity modeling, 초기 설계 공간 탕색을 위한design of experiments, scaling model 정학한 근사를 위해radial basis functions network를 이용하였으며 그리고 trust region management와 같은 방법들을 결합한다. 다른 스케일링 함수는 다 정밀도 최적화를 위한 가장 robust하고 정확한 접근법을 정하기 위해 평가되었다. 또한, 무미익 무인기 개념설계를 위한 통합 해석 코드가 소개되었다. 통합 해석 코드는 잘 알려진 분석 방법, open source와 무미익 무인기 분석을 위해 사용된 in-house 분석 프로그램으로 조합되어있다. 개념설계를 위해 무미익 항공기의 공력, 조종안정성의 특별한 특징들이 적용되었다. 무미익 항공기의 가장 취약한 점 중 한 가지는 방향안정성이다. 항공기 안정성에 split-drag-rudder의 영향이 고 정밀도 공력해석을 통해 분석되었다. 자동화된 전산유체역학 해석 과정을 개발했고 GVFM 알고리즘을 보여주기 위해 무인기 해석코드와 통합했다. 아음속, 천음속 에어포일 설계와 무인전투기 개념설계 사례연구는 알고리즘의 효율성을 보여준다. 모든 사례연구는 적은 수의 고 정밀도 해석을 통해서도 설계 개선이 보여짐을 확인할 수 있었다.