• Journal of Power Electronics
• /
• v.11 no.5
• /
• pp.751-758
• /
• 2011

This paper presents an effective robust predictive control scheme for the active power filter (APF) using a smith-predictor based current regulator, which show superior features when compared to proportional-integral (PI) controllers in terms of an enhanced closed-loop bandwidth and an improved current tracking accuracy. A moving average filter (MAF) is implemented using a field programmable gate array (FPGA) for signal pre-processing to eliminate the switching ripple contamination. An adaptive linear neural network (ADALINE) is used for individual harmonic estimation to achieve selective compensation purpose. The effectiveness and validity of the devised control algorithm are confirmed by extensive simulation and experimental results.

• Proceedings of the KIEE Conference
• /
• 2001.07d
• /
• pp.2815-2817
• /
• 2001

This study describes the emotion classification using two different feature extraction methods for four-channel EEG signals. One of the methods is linear prediction analysis based on AR model. Another method is cross-correlation coefficients on frequencies of ${\theta}$, ${\alpha}$, ${\beta}$ bands. Using the linear predictor coefficients and the cross-correlation coefficients of frequencies, the emotion classification test for four emotions, such as anger, sad, joy, and relaxation is performed with a neural network. Comparing the results of two methods, it seems that the linear predictor coefficients produce the better results than the cross-correlation coefficients of frequencies for-emotion classification.

• KIISE Transactions on Computing Practices
• /
• v.22 no.3
• /
• pp.127-132
• /
• 2016

Several methods have been proposed to apply artificial neural networks to text compression in the past. However, the networks and targets are both limited to the small size due to hardware capability in the past. Modern GPUs have much better calculation capability than CPUs in an order of magnitude now, even though CPUs have become faster. It becomes possible now to train greater and complex neural networks in a shorter time. This paper proposed a method to transform the distribution of original data with a probabilistic neural predictor. Experiments were performed on a feedforward neural network and a recurrent neural network with gated-recurrent units. The recurrent neural network model outperformed feedforward network in compression rate and prediction accuracy.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.156-161
• /
• 2003

It is an interest problem to predict substance distributions in three-dimensional space. Recently, a research field as Geostatistics is advanced. It is a kind of inter- or extrapolation mathematically. Some useful means for the inter- and extrapolation are known, in which slide window method with neural networks is hopeful one. We propose multi-dimensional extrapolation using multi-layer neural networks and the slide-window method. The multi-dimensional extrapolation is not similar to one-dimension. It has plural algorithms. We researched line predictors and local-plain predictors I two-dimensional space. The both predictors are equivalent; however, in multi-dimensional extrapolation, it is very important to find the direction of predictions. Especially, since the slide window method requires information to predict the future in sampling data, if they are not ordered appropriately in the direction, the predictor cannot operate. We tested the extrapolation for typical two-dimensional functions, and found an excellent character of slide-window method based on local-plain. By using the method, we can extrapolate the function until twice-outer regions of the definitions.

• Proceedings of the KIEE Conference
• /
• 2003.07d
• /
• pp.2456-2458
• /
• 2003

Recently, an automatic crane control system is required with high speed and rapid transportation. Therefore, when container is transferred from the initial coordinate to the finial coordinate, the container paths should be built in terms of the least time and without sway. Therefore, we calculated the anti-collision path for avoiding collision in its movement to the finial coordinate in this paper. And we constructed the neural network predictive two degree of freedom PID (NNPPID) controller to control the precise navigation. The proposed Predictive control system is composed of the neural network predictor, two degree of freedom PID(TDOFPID) controller, neural network self-tuner which yields parameters of TDOFPID. We analyzed crane system through simulation, and proved excellency of control performance over the conventional controllers.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.993-996
• /
• 1993

This paper addresses the structure and its associated learning algorithms of a feedforward multi-layered connectionist network, which has distributed learning abilities, for realizing the basic elements and functions of a traditional fuzzy logic controller. The proposed neural-network-based fuzzy logic control system (NN-FLCS) can be contrasted with the traditional fuzzy logic control system in their network structure and learning ability. An on-line supervised structure/parameter learning algorithm dynamic learning algorithm can find proper fuzzy logic rules, membership functions, and the size of output fuzzy partitions simultaneously. Next, a Reinforcement Neural-Network-Based Fuzzy Logic Control System (RNN-FLCS) is proposed which consists of two closely integrated Neural-Network-Based Fuzzy Logic Controllers (NN-FLCS) for solving various reinforcement learning problems in fuzzy logic systems. One NN-FLC functions as a fuzzy predictor and the other as a fuzzy controller. As ociated with the proposed RNN-FLCS is the reinforcement structure/parameter learning algorithm which dynamically determines the proper network size, connections, and parameters of the RNN-FLCS through an external reinforcement signal. Furthermore, learning can proceed even in the period without any external reinforcement feedback.

• Proceedings of the KIEE Conference
• /
• 2002.11c
• /
• pp.7-10
• /
• 2002

This paper describes a method of comfortableness evaluation using 4-channel EEGs. The proposed method uses the linear predictor coefficients as EEG feature parameters and neural network as comfortableness pattern classifier. For subject independent system, multi-templates are stored and the most similar template can be selected. Changing the temperature and humidity conditions, 4-channel EEG signals for 10 subjects are collected. As a result, the developed algorithm showed about 66.7% performance in the comfortableness evaluation.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1994.10a
• /
• pp.15-19
• /
• 1994

The application of neural network for fault dection of cutting force in turning was introduced. This monitoring system consist of a RBF predicton model and a ART-1 pattern classifier. RBF prediction model predict a cutting force signal. Prediction error of predictor is used for a input vector of ART-1 pattern classifier. Prediction error could be successfully performed to fault signal monitoring of ART-1 pattern classifier.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.10a
• /
• pp.516-521
• /
• 1998

본 논문에서는 비선형 시스템 해석 문제에 널리 이용되고 있는 퍼지 시스템(Fuzzy System)과 신경 회로망(MlPNN)의 성능을 평가하기 위해 비선형 예측기를 구성하였고 두 예측기를 비선형 시계열(Time Series) 예측 문제에 적용하여 두 예측기의 성능을 비교 분석하였다. 예측 실험을 위한 데이터로 Mackey-Glass와 Lorenz 시계열을 사용하였다.

• Smart Structures and Systems
• /
• v.16 no.6
• /
• pp.1133-1145
• /
• 2015

A new neural network (NN) predictive controller (NNPC) algorithm has been developed and tested in the computer simulation of active control of a nonlinear structure. In the present method an NN is used as a predictor. This NN has been trained to predict the future response of the structure to determine the control forces. These control forces are calculated by minimizing the difference between the predicted and desired responses via a numerical minimization algorithm. Since the NNPC is very time consuming and not suitable for real-time control, it is then used to train an NN controller. To consider the effectiveness of the controller on probability of damage, fragility curves are generated. The approach is validated by using simulated response of a 3 story nonlinear benchmark building excited by several historical earthquake records. The simulation results are then compared with a linear quadratic Gaussian (LQG) active controller. The results indicate that the proposed algorithm is completely effective in relative displacement reduction.