• 한국정밀공학회지
• /
• 제14권7호
• /
• pp.29-38
• /
• 1997

A closed-loop system modeling of an active/semiactive suspension system has been accomplished through an artificial neural network. A 7DOF full model as a system's equation of motion has been derived and an output feedback linear quadratic regulator has been designed for control purpose. A training set of a sample data has been obtained through a computer simulation. A 7DOF full model with LQR controller simulated under several road conditions such as sinusoidal bumps and rectangular bumps. A general multilayer perceptron neural network is used for dynamic modeling and target outputs are fedback to the a layer. A backpropagation method is used as a training algorithm. Model validation of new dataset have been shown through computer simulations.

• Journal of Welding and Joining
• /
• 제32권2호
• /
• pp.43-47
• /
• 2014

Famous artificial neural network (ANN) is applied to predict proper process window of arc welding. Target weldment is variously combined lap joint fillet welding of automotive steel plates. ANN's system variable such as number of hidden layers, perceptrons and transfer function are carefully selected through case by case test. Input variables are welding condition and steel plate combination, for example, welding machine type, shield gas composition, current, speed and strength, thickness of base material. The number of each input variable referred in welding experiment is counted and provided to make it possible to presume the qualitative precision and limit of prediction. One of experimental process windows is excluded for predictability estimation and the rest are applied for neural network training. As expected from basic ANN theory, experimental condition composed of frequently referred input variables showed relatively more precise prediction while rarely referred set showed poorer result. As conclusion, application of ANN to arc welding process window derivation showed comparatively practical feasibility while it still needs more training for higher precision.

• 한국산업융합학회 논문집
• /
• 제22권2호
• /
• pp.87-93
• /
• 2019

This research deals with posture optimization for humanoid robot against external forces using genetic algorithm and neural network. When the robot takes a motion to push an object, the torque of each joint is generated by reaction force at the palm. This study aims to optimize the posture of the humanoid robot that will change this torque. This study finds an optimized posture using a genetic algorithm such that torques are evenly distributed over the all joints. Then, a number of different optimized postures are generated from various the reaction forces at the palm. The data is to be used as training data of MLP(Multi-Layer Perceptron) neural network with BP(Back Propagation) learning algorithm. Humanoid robot can find the optimal posture at different reaction forces in real time using the trained neural network include non-training data.

• 융합신호처리학회논문지
• /
• 제1권1호
• /
• pp.1-6
• /
• 2000

본 논문은 HMM(Hidden Markov Model)을 이 용하여 인식을 수행할 경우의 오류를 최소화 할 수 있는 후처리 과정으로 신경망을 결합시켜 HMM 단독으로 사용하였을 때 보다 높은 인식률을 얻을 수 있는 HMM과 신경망의 하이브리드 시스템을 제안한다 HMM을 이용하여 학습한 후 학습에 참여하지 않은 데이터를 인식하였을 때 오인식 데이터를 정인식으로 인식하도록 HMM의 출력으로 얻은 각 출력확률을 후처리에 사용될 신경망의 학습용으로 사용하여 신경망을 학습하여 HMM과 신경망을 결합한 하이브리드 시스템을 만든다 이와 같은 HMM과 신경망을 결합한 하이브리드 모델을 사용하여 단독 숫자음에서 실험한 결과 HMM 단독으로 사용하였을 때 보다 MLP에서는 약 $4.5\%$ RBFN에서는 약 $2\%$의 인식률 향상이 있었다. 기존의 하이브리드 시스템이 갖는 많은 학습시간이 소요되는 문제점과 실시간 음성인식시스템을 구현할 패의 학습데이터의 부족으로 인한 인식률 저하를 해결할 수 있는 방법임을 확인할 수 있었다

• Computers and Concrete
• /
• 제25권6호
• /
• pp.551-563
• /
• 2020

This paper investigates the artificial neural network (ANN) to predict the dimensionless parameters for the maximum contact pressures and contact areas of a contact problem. Firstly, the problem is formulated and solved theoretically by using Theory of Elasticity and Integral Transform Technique. Secondly, the contact problem has been extended based on the ANN. The multilayer perceptron (MLP) with three-layer was used to calculate the contact distances. External load, distance between the two quarter planes, layer heights and material properties were created by giving examples of different values were used at the training and test stages of ANN. Program code was rewritten in C++. Different types of network structures were used in the training process. The accuracy of the trained neural networks for the case was tested using 173 new data which were generated via theoretical solutions so as to determine the best network model. As a result, minimum deviation value (difference between theoretical and C++ ANN results) of was obtained for the network model. Theoretical results were compared with artificial neural network results and well agreements between them were achieved.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1997년도 한국자동제어학술회의논문집; 한국전력공사 서울연수원; 17-18 Oct. 1997
• /
• pp.1557-1560
• /
• 1997

In this paper, we have built an experimental magnetic levitation system for a possible use of control education. We have give a mathermatical model of the nonlinear system and have shown the stability region of the linearized system when it is controlled by a PD controller. We also proposed a neural network control system which uses a neural network as a feedforward controller thgether with a conventional feedback PF controller. We have generated a desired output trajectory, which was designed for the benefit of the generalization of the neural network controller, and trained the desired output trajectory, which was desigend for the benefit of the generalization of the neural netowrk controller, and trained a neural network controller with the data of the actual input and the output of the system obtained by applying the desired output trajectroy. A good tracking performance was observed for both the desired trajectiories used and not used for the neural network training.

• 전기학회논문지
• /
• 제58권12호
• /
• pp.2532-2537
• /
• 2009

For defect characterization in steam generator tubes in nuclear power plant, artificial neural network has been extensively used to classify defect types. In this paper, we study the effectiveness of Bagging for improving the performance of neural network for the classification of tube defects. Bagging is a method that combines outputs of many neural networks that were trained separately with different training data set. By varying the number of neurons in the hidden layer, we carry out computer simulations in order to compare the classification performance of bagging neural network and single neural network. From the experiments, we found that the performance of bagging neural network is superior to the average performance of single neural network in most cases.

• 대한전자공학회논문지SD
• /
• 제45권1호
• /
• pp.20-27
• /
• 2008

본 논문에서는 비선형 패턴 분류를 위해 FPGA 칩에 신경회로망을 구현하였다. 병렬처리 연산을 위해 순방향 신경회로망이 구현 되었다. 신경망의 학습을 off-line으로 한 다음에 가중치 값들을 저장하여 사용한다. 예로서, AND와 XOR 논리의 패턴 구분이 수행된다. 실험결과를 통해 FPGA에 구현된 신경회로망이 잘 작동하는 것을 검증하였다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2001년도 ICCAS
• /
• pp.96.2-96
• /
• 2001

This paper presents a control method for the experimental mobile vehicle. By merging the advantages of neural network, adaptive and fuzzy control, neural network-based adaptive fuzzy control is proposed. It can deal with a large amount of training data by neural network, from these data producing more accurate fuzzy rules by adaptive control, and then controlling the object by fuzzy control. This is not the simple combination of the three methods, but merging them into one control system Experiments and some future considerations are given.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 1995년도 추계학술발표회논문집
• /
• pp.96-100
• /
• 1995

In this paper an induction motor control using fuzzy controller and neural network adptive observer is studied. The proposed observer which comprises neural network flux observer which comprises neural network flux observer and neural network torque observer is trained to learn the flux dynamics and torque dynamics and subjected to further on-line training by means of a backpropagation algorithem. Therefore it has been shown that the robust control of induction motor neglects the rotor time constant variations