• 한국정보과학회논문지:소프트웨어및응용
• /
• 제30권3_4호
• /
• pp.326-338
• /
• 2003

신경회로망 설계 및 모델선택의 목표는 최적의 구조를 가지는 일반화 성능이 우수한 네트워크를 구성하는 것이다. 하지만 학습데이타에는 노이즈(noise)가 존재하고, 그 수도 충분하지 않기 때문에 최종적으로 표현하고자 하는 진확률 분포와 학습 데이타에 의해 표현되는 경험확률분포(empirical probability density) 사이에는 차이가 발생한다. 이러한 차이 때문에 신경회로망을 학습데이타에 대하여 과다하게 적합(fitting)시키면, 학습데이타만의 확률분포를 잘 추정하도록 매개변수들이 조정되어 버리고, 진확률 분포로부터 멀어지게 된다. 이러한 현상을 과다학습이라고 하며, 과다학습된 신경회로망은 학습데이타에 대한 근사는 우수하지만, 새로운 데이타에 대한 예측은 떨어지게 된다. 또한 신경회로망의 복잡도가 증가 할수록 더 많은 매개변수들이 노이즈에 쉽게 적합되어 과다학습 현상은 더욱 심화된다. 본 논문에서는 통계적인 관점을 바탕으로 신경회로망의 일반화 성능을 향상시키는 신경회로 망의 설계 및 모델 선택의 통합적인 프로세스를 제안하고자 한다. 먼저 학습의 과정에서 적응적 정규화가 있는 자연기울기 학습을 통해 수렴속도의 향상과 동시에 과다학습을 방지하여 진확률 분포에 가까운 신경회로망을 얻는다. 이렇게 얻어진 신경회로망에 자연 프루닝(natural pruning) 방법을 적용하여 서로 다른 크기의 후보 신경회로망 모델을 얻는다. 이러한 학습과 복잡도 최적화의 통합 프로세스를 통하여 얻은 후보 모델들 중에서 최적의 모델을 베이시안 정보기준에 의해 선택함으로써 일반화 성능이 우수한 최적의 모델을 구성하는 방법을 제안한다. 또한 벤치마크 문제를 이용한 컴퓨터 시뮬레이션을 통하여, 제안하는 학습 및 모델 선택의 통합프로세스의 일반화 성능과 구조 최적화 성능의 우수성을 검증한다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.1296-1301
• /
• 2004

Fuzzy logic, neural network, fuzzy-neural network play an important as the key technology of linguistic modeling for intelligent control and decision making in complex systems. The fuzzy -neural network (FNN) learning represents one of the most effective algorithms to build such linguistic models. This paper proposes learning approach of fuzzy-neural network by immune algorithm. The proposed learning model is presented in an immune based fuzzy-neural network (FNN) form which can handle linguistic knowledge by immune algorithm. The learning algorithm of an immune based FNN is composed of two phases. The first phase used to find the initial membership functions of the fuzzy neural network model. In the second phase, a new immune algorithm based optimization is proposed for tuning of membership functions and structure of the proposed model.

• 대한전기학회논문지:전력기술부문A
• /
• 제48권2호
• /
• pp.97-106
• /
• 1999

This paper proposes the structure and the algorithm of the Fuzzy-Neural Controller(FNNC) which is able to adapt itself to unknown plant and the change of circumstances at the Fuzzy Logic Controller(FLC) with the Neural Network. This Learning Fuzzy Logic Controller is made up of Fuzzy Logic controller in charge of a main role and Neural Network of an adaptation in variable circumstances. This construct optimal fuzzy controller applied to the 2-area load frequency control of power system, and then it would examine fitness about parameter variation of plant or variation of circumstances. And it proposes the optimal Scale factor method wsint three preformance functions( E, , U) of system dynamics of load frequency control with error back-propagation learning algorithm. Applying the controller to the model of load frequency control, it is shown that the FNNC method has better rapidity for load disturbance, reduces load frequency maximum deviation and tie line power flow deviation and minimizes reaching and settling time compared to the Optimal Fuzzy Logic Controller(OFLC) and the Optimal Control for optimzation of performance index in past control techniques.

• 경영과학
• /
• 제11권1호
• /
• pp.187-196
• /
• 1994

Time-Space-Time(T-S-T) switching network is modeled as a graceful degrading system. Call blocking probability is defined as a measure of performance. Several performance related measures are suggested under the presence of failure. An optimization model is proposed, which determines optimal values of system parameters of the switching network.

• 조명전기설비학회논문지
• /
• 제29권2호
• /
• pp.104-112
• /
• 2015

This paper deals with an methodology for acquiring optimal order of rational function model in FDNE(frequency dependent network equivalents) with GA(genetic Algorithm). In order to analyze the modern power system with huge complexity, an practical and efficient equivalent model is needed which represents the system's characteristics of transient phenomenon. this paper shows developing a z domain rational function model which have the resultant coefficient from proposed GA simulation. To demonstrate this methodology, some simulations are performed with practical power system of NZ which applied with fault condition and nonlinear converter load.

• 대한전기학회논문지:전력기술부문A
• /
• 제48권10호
• /
• pp.1231-1238
• /
• 1999

This paper presents an approach for the optimal heating load Identification using Diagonal Recurrent Neural Networks(DRNN). In this paper, the DRNN captures the dynamic nature of a system and since it is not fully connected, training is much faster than a fully connected recurrent neural network. The architecture of DRNN is a modified model of the fully connected recurrent neural network with one hidden layer. The hidden layer is comprised of self-recurrent neurons, each feeding its output only into itself. In this study, A dynamic backpropagation (DBP) with delta-bar-delta learning method is used to train an optimal heating load identifier. Delta-bar-delta learning method is an empirical method to adapt the learning rate gradually during the training period in order to improve accuracy in a short time. The simulation results based on experimental data show that the proposed model is superior to the other methods in most cases, in regard of not only learning speed but also identification accuracy.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1997년도 추계학술대회 논문집
• /
• pp.454-457
• /
• 1997

The purpose of this study is to develop an optimal model, using existing models, that is able to estimate the amount of spatter utilizing artificial neural network in the short circuit transfer mode of gas metal arc (GMA) welding. The amount of spatter generated during welding can become a barometer which represents the process stability of metal transfer in GMA welding, and it depends on some factors which constitute a periodic waveforms of welding current and arc voltage in short circuit GMA welding. So, the 12 factors, which could express the characteristics for the waveforms, and the amount of spatter are used as input and output variables of the neural network, respectively. Two neural network models to estimate the amount of spatter are proposed: A neural network model, where arc extinction is not considered, and a combined neural network model where it is considered. In order to reduce the calculation time it take to produce an output, the input vector and hidden layers for each model are optimized using the correlation coefficients between each factor and the amount of spattcr. The est~mation performance of each optimized model to the amount of spatter IS assessed and compared to the est~mation performance of the model proposed by Kang. Also, through the evaluation for the estimation performance of each optimized model, it is shown that the combined neural network model can almost perfectly predict the amount of spatter.

• Structural Monitoring and Maintenance
• /
• 제9권3호
• /
• pp.289-303
• /
• 2022

In this study, the impact of assigned pixel labels on the accuracy of crack image identification of steel structures is examined by using an atrous separable convolution neural network (ASCNN). Firstly, images containing fatigue cracks collected from steel structures are classified into four datasets by assigning different pixel labels based on image features. Secondly, the DeepLab v3+ algorithm is used to determine optimal parameters of the ASCNN model by maximizing the average mean-intersection-over-union (mIoU) metric of the datasets. Thirdly, the ASCNN model is trained for various image sizes and hyper-parameters, such as the learning rule, learning rate, and epoch. The optimal parameters of the ASCNN model are determined based on the average mIoU metric. Finally, the trained ASCNN model is evaluated by using 10% untrained images. The result shows that the ASCNN model can segment cracks and other objects in the captured images with an average mIoU of 0.716.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2010년도 춘계학술대회 논문집
• /
• pp.1222-1230
• /
• 2010

This study is about developing a transit assignment model on railway network. Current transit assignment models are mainly focused on road or urban transportation so that these models, for example, transit assignment model based on optimal strategy generates unrealistic transit assignment. Especially, since the advent of KTX, more passengers are using the transfer route containing KTX but most transit assignment models have a shortcoming that transfer is not considered or is overestimated. We present a new stochastic transit assignment model based on LOGIT considering transfer resistance.

• 대한교통학회지
• /
• 제19권1호
• /
• pp.89-99
• /
• 2001

기존의 고정수요(Fixed Demand)를 전제로 한 가로망 설계 모형에서는 가로망의 구조나 용량이 개선되더라도 장래 기·종점 통행수요는 변하지 않는다고 가정한다. 이는 단기적인 가로망 설계에서는 성립할 수 있지만, 현실적으로 기·종점 통행수요는 네트워크 서비스수준에 따라 변화하므로 고정수요를 전제한 장기적인 가로망 설계문제에서는 그 타당성을 잃어버린다 그러므로 장래 최적 가로망 설계는 현실적 여건과 교통특성상 기·종점 통행 수요가 모형 내부에서 결정되는 내생변수로 처리하는 가변수요(Variable Demand)를 반영한 가로망 설계 문제로 모형을 구축하는 것이 바람직하다. 이러한 맥락에서 본 논문은 가변수요를 갖는 가로망 설계문제에 대한 이중계층 모형을 구축한 다음, 가로망내의 특성치가 변화하였을 때 그 파급영향을 먼저 파악하고 현 가로망 개선에서 가장 먼저 고려해야 할 링크를 찾아내기 위해 민감도 분석을 수행하였고, 민감도 분석과 연관되어 전체 시스템 효과척도를 최적화할 수 있는 대안적인 알고리즘을 제시하고 적용하여 구축된 모형으로 그 유효성을 검증하였고, 기존 고정수요 가로망 설계기법에 내재된 한계점을 극복하고자 하였다.