• Proceedings of the KIEE Conference
• /
• 2006.07d
• /
• pp.1868-1870
• /
• 2006

Neural network is used in many fields of control systems, but input-output patterns of a control system are not easy to be obtained and by using as single feedback neural network controller. And also it is difficult to get a satisfied performance when the changes of rapid load and disturbance are applied. To resolve those problems, this paper proposes a new algorithm which is the neural network controller. The new algorithm uses the neural network instead of activation function to control object at the output node. Therefore, control object is composed of neural network controller unifying activation function, and it supplies the error back propagation path to calculate the error at the output node. As a result, the input-output pattern problem of the controller which is resigned by the simple structure of neural network is solved, and real-time learning can be possible in general back propagation algorithm. Application of the new algorithm of neural network controller gives excellent performance for initial and tracking response and it shows the robust performance for rapid load change and disturbance. The proposed control algorithm is implemented on a high speed DSP, TMS320C32, for the speed of 3-phase induction motor. Enhanced performance is shown in the test of the speed control.

• Journal of Institute of Control, Robotics and Systems
• /
• v.17 no.8
• /
• pp.731-738
• /
• 2011

In this paper, we introduce an advanced architecture of K-Means clustering-based polynomial Radial Basis Function Neural Networks (p-RBFNNs) designed with the aid of SSOA (Space Search Optimization Algorithm) and develop a comprehensive design methodology supporting their construction. In order to design the optimized p-RBFNNs, a center value of each receptive field is determined by running the K-Means clustering algorithm and then the center value and the width of the corresponding receptive field are optimized through SSOA. The connections (weights) of the proposed p-RBFNNs are of functional character and are realized by considering three types of polynomials. In addition, a WLSE (Weighted Least Square Estimation) is used to estimate the coefficients of polynomials (serving as functional connections of the network) of each node from output node. Therefore, a local learning capability and an interpretability of the proposed model are improved. The proposed model is illustrated with the use of nonlinear function, NOx called Machine Learning dataset. A comparative analysis reveals that the proposed model exhibits higher accuracy and superb predictive capability in comparison to some previous models available in the literature.

• The Journal of Korean Association of Computer Education
• /
• v.9 no.5
• /
• pp.77-84
• /
• 2006

In this paper, we propose an adaptive image watermarking algorithm in DWT domain by using HVS(human Visual system) and SOM(Self-Organizing Map) among neural networks. HVS can be described in terms of two properties of HVS: brightness and texture sensitivity. The SOM is used to obtain the local characteristics of image, Therefore, the suitable strength and length of embedded watermark is determined by using HVS and SOM. The experimental results show that proposed method provides a suitable strength and length of watermark and has good perceptual invisibility and robustness for various attacks.

• Proceedings of the KSME Conference
• /
• 2000.04a
• /
• pp.505-510
• /
• 2000

Vehicle performances such as fuel consumption and catalyst-out emissions are affected by a driving pattern, which is defined as a driving cycle with the grade in this study. We developed an algorithm to recognize a current driving pattern by using a neural network. And this algorithm can be used in adapting the driving control strategy to the recognized driving pattern. First, we classified the general driving patterns into 6 representative driving patterns, which are composed of 3 urban driving patterns, 2 suburban driving patterns and 1 expressway driving pattern. A total of 24 parameters such as average cycle velocity, positive acceleration kinetic energy, relative duration spent at stop, average acceleration and average grade are chosen to characterize the driving patterns. Second, we used a neural network (especially the Hamming network) to decide which representative driving pattern is closest to the current driving pattern by comparing the inner products between them. And before calculating inner product, each element of the current and representative driving patterns is transformed into 1 and -1 array as to 4 levels. In the end, we simulated the driving pattern recognition algorithm in a temporary pattern composed of 6 representative driving patterns and, verified the reliable recognition performance.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.6 no.6
• /
• pp.873-878
• /
• 2002

Stator current spectrum by the fast Fourier transform (FFT) of current signals has been widely used for fault detection in induction motors. In this paper, we propose efficient rotor fault detection of Induction motors using stator current spectrum monitoring. The proposed method utilizes the mean absolute difference (MAD) between a Predetermined reference vector and a feature vector extracted from the stator current spectrum. Our proposed approach requires a smaller amount of computations when compared to fault detection algorithms based on neural networks, since it uses simple MAD criterion to detect rotor faults related broken rotor bars. Experimental results show that our proposed method can successively detect the rotor fault of the induction motor.

• Journal of Welding and Joining
• /
• v.17 no.5
• /
• pp.40-46
• /
• 1999

The sensors employed in the robotic are welding system must detect the changes in weld characteristics and produce the output that is in some way related to the change being detected. Such adaptive systems, which synchronise the robot arm and eyes using a primitive brain will form the basis for the development of robotic GMA(Gas Metal Arc) welding which increasingly higher levels of artificial intelligence. The objective of this paper is to realize the mapping characteristics of bead height through learning. After learning, the neural estimation can estimate the bead height desired from the learning mapping characteristic. The design parameters of the neural network estimator(the number of hidden layers and the number of nodes in a layer) are chosen from an estimation error analysis. A series of bead of bead-on-plate GMA welding experiments was carried out in order to verify the performance of the neural network estimator. The experimental results show that the proposed neural network estimator can predict the bead height with reasonable accuracy and guarantee the uniform weld quality.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.11
• /
• pp.3130-3136
• /
• 2009

This paper deals with the optimal on-line real time voltage regulation methods in power distribution systems interconnected with the Distributed Generation(DG) systems. In order to deliver suitable voltage to as many customers as possible, the optimal sending voltage should be decided by the effective voltage regulation method by using artificial neural networks to consider the rapid load variation and random operation characteristics of DG systems. The results from a case study show that the proposed method can be a practical tool for the voltage regulation in distribution systems including many DG systems.

• Journal of Institute of Control, Robotics and Systems
• /
• v.12 no.6
• /
• pp.539-546
• /
• 2006

The Air Quality Sensor(AQS), located near the fresh air inlet, serves to reduce the amount of pollution entering the vehicle cabin through the HVAC(heating, ventilating, and air conditioning) system by sending a signal to close the fresh air inlet door/ventilation flap when the vehicle enters a high pollution area. The sensor module which includes two independent sensing elements for responding to diesel and gasoline exhaust gases, and temperature sensor and humidity sensor was designed for intelligent AQS in automobile. With this sensor module, AVR microcontroller was designed with back propagation neural network to a powerful gas/vapor pattern recognition when the motor vehicles pass a pollution area. Momentum back propagation algorithm was used in this study instead of normal backpropagation to reduce the teaming time of neural network. The signal from neural network was modified to control the inlet of automobile and display the result or alarm the situation in this study. One chip microcontroller, ATmega 128L(ATmega Ltd., USA) was used for the control and display. And our developed system can intelligently reduce the malfunction of AQS from the dampness of air or dense fog with the backpropagation neural network and the input sensor module with four sensing elements such as reducing gas sensing element, oxidizing gas sensing element, temperature sensing element and humidity sensing element.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.45 no.1
• /
• pp.113-122
• /
• 1996

As integrity of semiconductor device is increased, accurate and efficient modeling and recipe generation of semiconductor fabrication procsses are necessary. Among the major semiconductor manufacturing processes, dry etc- hing process using gas plasma and accelerated ion is widely used. The process involves a variety of the chemical and physical effects of gas and accelerated ions. Despite the increased popularity, the complex internal characteristics made efficient modeling difficult. Because of difficulty to determine the control input for the desired output, the recipe generation depends largely on experiences of the experts with several trial and error presently. In this paper, the optimal control of the etching is carried out in the following two phases. First, the optimal neural network models for etching process are developed with genetic algorithm utilizing the input and output data obtained by experiments. In the second phase, search for optimal control inputs in performed by means of using the optimal neural network developed together with genetic algorithm. The results of study indicate that the predictive capabilities of the neural network models are superior to that of the statistical models which have been widely utilized in the semiconductor factory lines. Search for optimal control inputs using genetic algorithm is proved to be efficient by experiments. (author). refs., figs., tabs.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.61 no.1
• /
• pp.135-142
• /
• 2012

In this paper, we proposed a new architecture called radial basis function-based polynomial neural networks classifier that consists of heterogeneous neural networks such as radial basis function neural networks and polynomial neural networks. The underlying architecture of the proposed model equals to polynomial neural networks(PNNs) while polynomial neurons in PNNs are composed of Fuzzy-c means-based radial basis function neural networks(FCM-based RBFNNs) instead of the conventional polynomial function. We consider PNNs to find the optimal local models and use RBFNNs to cover the high dimensionality problems. Also, in the hidden layer of RBFNNs, FCM algorithm is used to produce some clusters based on the similarity of given dataset. The proposed model depends on some parameters such as the number of input variables in PNNs, the number of clusters and fuzzification coefficient in FCM and polynomial type in RBFNNs. A multiobjective particle swarm optimization using crowding distance (MoPSO-CD) is exploited in order to carry out both structural and parametric optimization of the proposed networks. MoPSO is introduced for not only the performance of model but also complexity and interpretability. The usefulness of the proposed model as a classifier is evaluated with the aid of some benchmark datasets such as iris and liver.