In-Soo Lee

Neural network indirect adaptive control with fast learning algorithm

Abstract A fast learning algorithm based on a new cost function and a linearized error signal is proposed. The proposed learning algorithm is applied to indirect adaptive control of nonlinear plants. In the proposed method, we use the identification error and the control error to train the NNI and the NNC, respectively. In addition, we introduce a linearized error signal in order to improve the learning speed. Computer simulation results show that the rate of convergence increases, and that the NNC based on the proposed method is insensitive to variations of the plant parameters.

Fault Diagnosis of Induction Motors Using Discrete Wavelet Transform and Artificial Neural Network

This paper proposes a fault diagnosis method for induction motors based on DWT (Discrete Wavelet Transform) and artificial NN. The proposed algorithm is based on ART2 NN (adaptive resonance theory 2 neural network) with uneven vigilance parameters. Proposed fault diagnosis method consists of data preprocessing part by frequency analysis of vibration signal, and fault classifier for fault isolation by ART2 NN. Especially, the data preprocessing part which converts the sampled signals into the frequency domain by DWT is very important to improve the performance of the fault diagnosis. In this paper both rotor and bearing faults of the induction motors are considered for diagnosis. The experiment results demonstrate the effectiveness of the proposed fault diagnosis method of induction motors.

Real-time lane detection and tracking system using simple filter and Kalman filter

In this paper, we propose a real-time lane detection and tracking algorithm using a simple filter and Kalman filter to develop a Lane Departure Warning System(LDWS) that can be implemented in an embedded system. Our LDWS was realized on the I.MX6Q board, mounted on a test vehicle, and traveled about 1,000km to derive the experimental results. Experimental results show that our LDWS operates at 97% detection rate in the day time and 95% in the night time. The average processing time of the our LDWS is about 15frame per seconds.

Implementation of a fault diagnosis system using neural networks for solar panel

In this paper, we developed a fault diagnosis system using neural networks for solar panel of solar street light. To perform a fault diagnosis of solar panel, we obtained the Open Circuit Voltage according to the duty ratio and it is used as input of Adaptive Resonance Theory 2 Neural Network (ART2 NN) and Multilayer Neural Network (MNN). As a result, we can double-check fault diagnosis for solar panel by using two neural networks through GUI and solar panel fault diagnosis is correctly estimated. So it is expected that the fault diagnosis system we proposed will be applicable to similar systems and devices.

Reversible Data Hiding Scheme Using the Difference Between the Maximum and Minimum Values

Qu et al. proposed a PPVO scheme based reversible data hiding technique in 2015. Their scheme obtains the predicted error value by using context pixels, and uses current pixel, predicted error value and block pixels that except for the current pixel. The PPVO scheme has a complex process because of embedding and extraction methods. In this paper, we propose an improved reversible data hiding scheme by using the difference values between the maximum value and minimum value. In our scheme, the difference values are used for a secret message embedding and extraction. Also, our scheme satisfies the characteristic of reversible data hiding. In experimental results, the embedding capacity and image quality of the proposed scheme are superior in comparison with Qu et al.’s PPVO scheme.