• 한국항공우주학회지
• /
• 제41권1호
• /
• pp.48-53
• /
• 2013

다항식 반응면 모델은 실제의 물리적, 수치적 실험을 대체하는 근사모델로 여러 공학분야에서 사용되고 있다. 일반적으로 반응면 구성에 필요한 실험점 수를 줄이기 위하여 낮은 차수의 다항식을 사용하므로, 심한 비선형성이 동반되는 현상에 대한 모델링에는 한계가 있다. 본 연구에서는 다항식의 차수를 증가시키는 방법 및 다항식을 구성하는 최적의 기저함수를 선정하는 방법을 통해 다항식 반응면의 모델링 능력을 확장할 수 있는 방법을 개발하였다. 최적 기저함수의 선정에는 유전 알고리즘을 적용하였으며, 1 변수 및 2변수 함수와 풍동시험 데이터에 대한 모델링 사례를 통해 개발된 방법이 비선형성이 심한 현상을 모델링하는데 적용될 수 있음을 확인하였다.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 1998년도 춘계학술대회논문집; 용평리조트 타워콘도, 21-22 May 1998
• /
• pp.444-449
• /
• 1998

A modal transducer in two-dimensional structure can be implemented by varying the distributed transducer's gain spatially. In this paper, a method based on finite element method is developed for optimizing spatial gain distribution of PVDF transducer to create the modal transducer for specific modes. Using this concept, one can design the modal transducer in two-dimensional structure having arbitrary geometry and boundary conditions. As a practical means for implementing this continuous gain distribution without repoling die PVDF film, the gain distribution is approximated by optimizing gain-weights of interface circuit. The whole spatial area of the PVDF film is divided into several electrode segments and the signals from each segment are properly weighted and summed by interface circuit. This corresponds to the approximation of a continuous function using discrete values. The electrode partition is optimized using the genetic algorithm. Gain-weights are optimized using the simplex search method. A modal sensor for first to fourth modes of aluminum plate is designed using PVDF film with gain-weighted interface circuit. Various lamination angles of PVDF film are taken into consideration to utilize the anisotropy of the PVDF film. Performance of the optimized' PVDF sensor is demonstrated by numerical simulations..

• 소음진동
• /
• 제8권4호
• /
• pp.632-642
• /
• 1998

A method based on finite element discretization is developed for optimizing the polarization profile of PVDF film to create the modal transducer for specific modes. Using this concept, one can design the modal transducer in two-dimensional structure having arbitrary geometry and boundary conditions. As a practical means for implementing this polarization profile without repoling the PVDF film the polarization profile is approximated by optimizing electrode patterns, lamination angles, and poling directions of the multi-layered PVDF transducer. This corresponds to the approximation of a continuous function using discrete values. The electrode pattern of each PVDF layer is optimized by deciding the electrode of each finite element to be used or not. Genetic algorithm, suitable for discrete problems, is used as an optimization scheme. For the optimization of each layers lamination angle, the continuous lamination angle is encoded into discrete value using binary 5 bit string. For the experimental demonstration, a modal sensor for first and second modes of cantilevered composite plate is designed using two layers of PVDF films. The actuator is designed based on the criterion of minimizing the system energy in the control modes under a given initial condition. Experimental results show that the signals from residual modes are successfully reduced using the optimized multi-layered PVDF sensor. Using discrete LQG control law, the modal peaks of first and second modes are reduced in the amount of 12 dB and 4 dB, resepctively.

• Steel and Composite Structures
• /
• 제52권1호
• /
• pp.89-107
• /
• 2024

This article focuses on the static and dynamic analysis and optimization of an anisogrid lattice plate subjected to axial compressive load with simply supported boundary conditions. The lattice plate includes diagonal and transverse ribs and is modeled as an orthotropic plate with effective stiffness properties. The study employs the first-order shear deformation theory and the Ritz method with a Legendre approximation function. In the realm of optimization, the Non-dominated Sorting Genetic Algorithm-II is utilized as an evolutionary multi-objective algorithm to optimize. The research findings are validated through finite element analysis. Notably, this study addresses the less-explored areas of optimizing the geometric parameters of the plate by maximizing the buckling load and natural frequency while minimizing mass. Furthermore, this study attempts to fill the gap related to the analysis of the post-buckling behavior of lattice plates, which has been conspicuously overlooked in previous research. This has been accomplished by conducting nonlinear analyses and scrutinizing post-buckling diagrams of this type of lattice structure. The efficacy of the continuous methods for analyzing the natural frequency, buckling, and post-buckling of these lattice plates demonstrates that while a degree of accuracy is compromised, it provides a significant amount of computational efficiency.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권1호
• /
• pp.40-49
• /
• 2004

In this paper, we propose a new architecture of Genetic Algorithms(GAs)-based Self-Organizing Polynomial Neural Networks(SOPNN), discuss a comprehensive design methodology and carry out a series of numeric experiments. The conventional SOPNN is based on the extended Group Method of Data Handling(GMDH) method and utilized the polynomial order (viz. linear, quadratic, and modified quadratic) as well as the number of node inputs fixed (selected in advance by designer) at Polynomial Neurons (or nodes) located in each layer through a growth process of the network. Moreover it does not guarantee that the SOPNN generated through learning has the optimal network architecture. But the proposed GA-based SOPNN enable the architecture to be a structurally more optimized network, and to be much more flexible and preferable neural network than the conventional SOPNN. In order to generate the structurally optimized SOPNN, GA-based design procedure at each stage (layer) of SOPNN leads to the selection of preferred nodes (or PNs) with optimal parameters- such as the number of input variables, input variables, and the order of the polynomial-available within SOPNN. An aggregate performance index with a weighting factor is proposed in order to achieve a sound balance between approximation and generalization (predictive) abilities of the model. A detailed design procedure is discussed in detail. To evaluate the performance of the GA-based SOPNN, the model is experimented with using two time series data (gas furnace and NOx emission process data of gas turbine power plant). A comparative analysis shows that the proposed GA-based SOPNN is model with higher accuracy as well as more superb predictive capability than other intelligent models presented previously.

• International Journal of Control, Automation, and Systems
• /
• 제2권4호
• /
• pp.423-434
• /
• 2004

In this paper, we introduce a new topology of Self-Organizing Polynomial Neural Networks (SOPNN) based on genetically optimized Multi-Layer Perceptron (MLP) and discuss its comprehensive design methodology involving mechanisms of genetic optimization. Let us recall that the design of the 'conventional' SOPNN uses the extended Group Method of Data Handling (GMDH) technique to exploit polynomials as well as to consider a fixed number of input nodes at polynomial neurons (or nodes) located in each layer. However, this design process does not guarantee that the conventional SOPNN generated through learning results in optimal network architecture. The design procedure applied in the construction of each layer of the SOPNN deals with its structural optimization involving the selection of preferred nodes (or PNs) with specific local characteristics (such as the number of input variables, the order of the polynomials, and input variables) and addresses specific aspects of parametric optimization. An aggregate performance index with a weighting factor is proposed in order to achieve a sound balance between the approximation and generalization (predictive) abilities of the model. To evaluate the performance of the GA-based SOPNN, the model is experimented using pH neutralization process data as well as sewage treatment process data. A comparative analysis indicates that the proposed SOPNN is the model having higher accuracy as well as more superb predictive capability than other intelligent models presented previously.reviously.

• 한국지능시스템학회논문지
• /
• 제10권4호
• /
• pp.343-350
• /
• 2000

본 논문에서는, 최적 시스템을 위해서 FNN과 PNN에 기반을 둔 Multi-FPNN(다중 퍼지 다항식 뉴럴네트워크) 모델을 제안한다. 여기서 FNN 구조는 각각의 분리된 입력변수에 의해 분할된 퍼지 입력공간을 사용해서 설게되고, 간략 퍼지추론 방법과 오류 역전파 알고리즘을 이용한다. FNN은 더 좋은 출력성능을 얻기 위해 PNN과 결합한다. GMDH 방법에 기초한 PNN 구조의 각 노드는 1차 및 2차 고계 다항식의 두 형태를 사용하고, 그 노드의 입력의 입력은 2, 3, 4의 세 종류의 다변수 입력을 사용한다. 그리고 다중 FPNN 모델의 구조와 파라미터를 동정하기 위햐 HCM 크러스터링방법과 유전자 알고리즘을 사용한다. 여기서, 시스템을 위해 데이터 전처리 기능을 수행하는 HCM 클러스터링 방법은 입출력 공간분할에 의해 다중 FPNN 구조를 결정하기 위해 사용된다. 모델의 근사화와 일반화 능력 사이에 충분한 군형을 ？기 위해 하중계수를 가진 합성 성능지수(목적함수)를 사용한다. 데이터 개수, 비선형의 정도(입.출력 데이터 분포)에 위존하는 이 합성 목적함수의 하중계수의 선택 및 조절을 통하여 최적의 다중 FPNN모델을 설계하는 것이 유용하고 효과적임을 보인다. 본 연구는 두 개의 대표적 수치예의 도움으로 설명되고, 그 모델의 근사화 및 일만화 능력에 관련된 합성 성능 지수가 평가되고, 도한 토의된다.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제49권7호
• /
• pp.378-388
• /
• 2000

In this paper, we propose Fuzzy Polynomial Neural Networks(FPNN) based on Polynomial Neural Networks(PNN) and Fuzzy Neural Networks(FNN) for model identification of complex and nonlinear systems. The proposed FPNN is generated from the mutually combined structure of both FNN and PNN. The one and the other are considered as the premise part and consequence part of FPNN structure respectively. As the consequence part of FPNN, PNN is based on Group Method of Data Handling(GMDH) method and its structure is similar to Neural Networks. But the structure of PNN is not fixed like in conventional Neural Networks and self-organizing networks that can be generated. FPNN is available effectively for multi-input variables and high-order polynomial according to the combination of FNN with PNN. Accordingly it is possible to consider the nonlinearity characteristics of process and to get better output performance with superb predictive ability. As the premise part of FPNN, FNN uses both the simplified fuzzy inference as fuzzy inference method and error back-propagation algorithm as learning rule. The parameters such as parameters of membership functions, learning rates and momentum coefficients are adjusted using genetic algorithms. And we use two kinds of FNN structure according to the division method of fuzzy space of input variables. One is basic FNN structure and uses fuzzy input space divided by each separated input variable, the other is modified FNN structure and uses fuzzy input space divided by mutually combined input variables. In order to evaluate the performance of proposed models, we use the nonlinear function and traffic route choice process. The results show that the proposed FPNN can produce the model with higher accuracy and more robustness than any other method presented previously. And also performance index related to the approximation and prediction capabilities of model is evaluated and discussed.

• 한국지능시스템학회논문지
• /
• 제21권3호
• /
• pp.401-406
• /
• 2011

복잡해진 최적화문제를 전통적인 방법보다 효율적으로 해결하기위해 유전알고리즘이나 개미군집화, 하모니서치알고리즘과 같은 다양한 메타휴리스틱이 개발되었다. 그 중에서 하모니 서치알고리즘이 다른 메타휴리스틱알고리즘보다 좋은 결과를 보이고 있다. 하모니 서치 알고리즘은 음악을 작곡할 때 아름다운 소리를 내는 하모니를 찾는 과정을 모방했다. 성능은 하모니 메모리에서 선택하는 비율인 HMCR값과 하모니 메모리에서 선택된 값의 조정 비율을 결정하는 PAR값에 따라 달라지는 것으로 알려져 있다. 다르게 말하면 두 변수의 기반이 되는 하모니 메모리의 사용방법의 문제로 볼 수 있다. 본 논문은 설정한 기간 동안 더 좋은 최적해를 찾지 못할 경우 하모니 메모리의 일부를 좋은 하모니로 구성되게 수정하는 방법을 제안했다. 테스트 함수를 이용한 검증 실험결과에서 하모니 메모리를 수정할 경우 정확도 변화가 적어 신뢰성 있는 정확도를 보였으며, Iteration이 짧더라도 최적값에 근접한 값을 찾았다.

• 지능정보연구
• /
• 제16권4호
• /
• pp.131-145
• /
• 2010

ERP, SCM 등과 같은 기업용 정보 시스템을 활용함에 있어, 고객의 문의에 따라 제품 판매 가능 유무와 가능일자를 계산하여 통보해 주는 지능형 ATP 시스템은 전산 정보를 활용하여 고객 만족도를 최대화할 수 있는 유용한 기능이라고 할 수 있다. 그렇지만 공급 사슬 환경에서 ATP 시스템을 적용하려고 할 경우, 고객이 문의해 온 Retailer에게 납품 가능한 모든 분배센터(Distribution Center)와 공장(Plant)의 미래 시점의 재고량 변화와 운송 능력 등을 모두 고려하여야 하므로 계산량이 방대한 NP-Complete 문제가 된다. 따라서 시스템 사용자가 빠른 시간 내에 해를 구하여 고객에게 결과를 알려 줄 수 있는 ATP 시스템의 개발은 공급 사슬 관리를 효과적으로 활용하기 위하여 반드시 필요한 일이라고 할 수 있다. 본 논문에서는 동적 생산 함수의 개념을 이용하여 비 정수 타임 랙을 고려하여 ATP 시스템을 모델링하고, 해당 수리 모형으로부터 효율적으로 해를 얻기 위하여 유전 알고리듬을 개발하였다. 비 정수 타임 랙을 활용한 ATP 시스템은 비 정수 타임 랙을 올림이나 내림을 통하여 정수화 시킨 후 모형 수립하는 기존의 방법보다 정교하게 현실을 반영할 수 있고, ATP 시스템을 위한 유전 알고리듬의 진화 시스템은 문제크기가 작은 것에서부터 큰 것까지 최적해에 매우 근사한 값을 매우 빠른 시간 내에 풀 수 있음을 알 수 있었다.