• Journal of Ship and Ocean Technology
• /
• 제4권3호
• /
• pp.1-12
• /
• 2000

A number of numerical methods like Computational Fluid Dynamics(CFD) have been developed to predict the flow fields of a vessel but the present study is developed to infer the wake fields on propeller plane by Statistical Fluid Dynamics(SFD) approach which is emerging as a new technique over a wide range of industrial fields nowadays. Neural network is well known as one prospective representative of the SFD tool and is widely applied even in the engineering fields. Further to its stable and effective system structure, generalization of input training patterns into different classification or categorization in training can offer more systematic treatments of input part and more reliable result. Because neural network has an ability to learn the knowledge through the external information, it is not necessary to use logical programming and it can flexibly handle the incomplete information which is not easy to make a definition clear. Three dimensional stern hull forms and nominal wake values from a model test are structured as processing elements of input and output layer respectively and a neural network is trained by the back-propagation method. The inferred results show similar figures to the experimental wake distribution.

• 한국음향학회지
• /
• 제37권6호
• /
• pp.475-482
• /
• 2018

본 논문에서는 DNN(Deep Neural Network) 기반 오디오 표식을 위한 데이터 증강 방법을 연구한다. 본 시스템에서는 오디오 신호를 멜-스펙트로그램으로 변환하여 오디오 표식을 위한 심층신경망의 입력으로 사용한다. 적은 수의 훈련 데이터를 사용하는 경우 발생하는 문제를 해결하기 위해, 타임 스트레칭, 피치 변화, 동적 영역 압축, 블록 혼합 등의 방법을 사용하여 훈련 데이터를 증강시켰다. 사용된 데이터 증강 기법의 최적 파라미터와 최적 조합을 오디오 표식 시뮬레이션을 통해 확인하였다.

• 정보학연구
• /
• 제8권1호
• /
• pp.51-56
• /
• 2005

The Back propagation algorithm is a very effective supervised training method for multi-layer feed forward neural networks. This paper studies the decision boundary formation based on the Back propagation algorithm. The discriminating powers of several neural network topology are also investigated against five manually created data sets. It is found that neural networks with multiple hidden layer perform better than single hidden layer.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2007년도 심포지엄 논문집 정보 및 제어부문
• /
• pp.3-5
• /
• 2007

This paper presents decision feedback equalizers using a recurrent neural network trained algorithm using extended Kalman filter(EKF) and sigma-point Kalman filter(SPKF). EKF is propagated, analytically through the first-order linearization of the nonlinear system. This can introduce large errors in the true posterior mean and covariance of the Gaussian random variable. The SPKF addresses this problem by using a deterministic sampling approach. The features of the proposed recurrent neural equalizer And we investigate the bit error rate(BER) between EKF and SPKF.

• 비파괴검사학회지
• /
• 제28권3호
• /
• pp.254-262
• /
• 2008

The objective of the proposed study is to produce a tool-condition monitoring (TCM) strategy that will lead to a more efficient and economical drilling tool usage. Drill-wear monitoring is an important attribute in the automatic cutting processes as it can help preventing damages of the tools and workpieces and optimizing the tool usage. This study presents the architectures of a multi-layer feed-forward neural network with back-propagation training algorithm for the monitoring of drill wear. The input features to the neural networks were extracted from the AE signals using the wavelet transform analysis. Training and testing were performed under a moderate range of cutting conditions in the dry drilling of steel plates. The results indicated that the extracted input features from AE signals to the supervised neural networks were effective for drill wear monitoring and the output of the neural networks could be utilized for the tool life management planning.

• 산업기술연구
• /
• 제24권A호
• /
• pp.99-107
• /
• 2004

This paper considers reinforcement learning control with the self-organizing map. Reinforcement learning uses the observable states of objective system and signals from interaction of the system and the environments as input data. For fast learning in neural network training, it is necessary to reduce learning data. In this paper, we use the self-organizing map to parition the observable states. Partitioning states reduces the number of learning data which is used for training neural networks. And neural dynamic programming design method is used for the controller. For evaluating the designed reinforcement learning controller, an inverted pendulum of the cart system is simulated. The designed controller is composed of serial connection of self-organizing map and two Multi-layer Feed-Forward Neural Networks.

• 수산해양기술연구
• /
• 제48권1호
• /
• pp.72-81
• /
• 2012

Using artificial neural network (ANN) technique, auction prices for common mackerel were forecasted with the daily total sale and auction price data at the Busan Cooperative Fish Market before introducing Total Allowable Catch (TAC) system, when catch data had no limit in Korea. Virtual input data produced from actual data were used to improve the accuracy of prediction and the suitable neural network was induced for the prediction. We tested 35 networks to be retained 10, and found good performance network with regression ratio of 0.904 and determination coefficient of 0.695. There were significant variations between training and verification errors in this network. Ideally, it should require more training cases to avoid over-learning, which leads to improve performance and makes the results more reliable. And the precision of prediction was improved when environmental factors including physical and biological variables were added. This network for prediction of price and catch was considered to be applicable for other fishes.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1999년도 제14차 학술회의논문집
• /
• pp.214-217
• /
• 1999

In this paper, genetic algorithm (GA) is implemented to search for the optimal structures (i.e. the kind of neural networks, the number of inputs and hidden neurons) of neural networks which are used approximating a given nonlinear function. Two kinds of neural networks, i.e. the multilayer feedforward [1] and time delay neural networks (TDNN) [2] are involved in this paper. The synapse weights of each neural network in each generation are obtained by associated training algorithms. The simulation results of nonlinear function approximation are given out and some improvements in the future are outlined.

• Journal of Advanced Marine Engineering and Technology
• /
• 제28권1호
• /
• pp.90-97
• /
• 2004

DC motor requires a rotor speed sensor for accurate speed control. The speed sensors such as resolvers and encoders are used as speed detectors. but they increase cost and size of the motor and restrict the industrial drive applications. So in these days. many Papers have reported on the sensorless operation or DC motor(3)-(5). This paper Presents a new sensorless strategy using neural networks(6)-(8). Neural network structure has three layers which are input layer. hidden layer and output layer. The optimal neural network structure was tracked down by trial and error and it was found that 4-16-1 neural network has given suitable results for the instantaneous rotor speed. Also. learning method is very important in neural network. Supervised learning methods(8) are typically used to train the neural network for learning the input/output pattern presented. The back-propagation technique adjusts the neural network weights during training. The rotor speed is gained by weights and four inputs to the neural network. The experimental results were found satisfactory in both the independency on machine parameters and the insensitivity to the load condition.

• Journal of Information Processing Systems
• /
• 제14권6호
• /
• pp.1464-1479
• /
• 2018

In face recognition, sometimes the number of available training samples for single category is insufficient. Therefore, the performances of models trained by convolutional neural network are not ideal. The small sample face recognition algorithm based on novel Siamese network is proposed in this paper, which doesn't need rich samples for training. The algorithm designs and realizes a new Siamese network model, SiameseFacel, which uses pairs of face images as inputs and maps them to target space so that the $L_2$ norm distance in target space can represent the semantic distance in input space. The mapping is represented by the neural network in supervised learning. Moreover, a more lightweight Siamese network model, SiameseFace2, is designed to reduce the network parameters without losing accuracy. We also present a new method to generate training data and expand the number of training samples for single category in AR and labeled faces in the wild (LFW) datasets, which improves the recognition accuracy of the models. Four loss functions are adopted to carry out experiments on AR and LFW datasets. The results show that the contrastive loss function combined with new Siamese network model in this paper can effectively improve the accuracy of face recognition.