• Structural Engineering and Mechanics
• /
• v.23 no.4
• /
• pp.409-430
• /
• 2006

The aim of this study is to investigate the reliability of strong motion records processed by causal and acausal Butterworth filters in comparison to the results obtained from a synthetic accelerogram. For this purpose, the fault parallel component of the Bolu record of the Duzce earthquake is modeled with a sum of exponentially damped sinusoidal components. Noise-free velocities and displacements are then obtained by analytically integrating the synthetic acceleration model. The analytical velocity and displacement signals are used as a standard with which to judge the validity of the signals obtained by filtering with causal and acausal filters and numerically integrating the acceleration model. The results show that the acausal filters are clearly preferable to the causal filters due to the fact that the response spectra obtained from the acausal filters match the spectra obtained from the simulated accelerogram better than that obtained by causal filters. The response spectra are independent from the order of the filters and from the method of integration (whether analytical integration after a spline fit to the synthetic accelerogram or the trapezoidal rule). The response spectra are sensitive to the chosen corner frequency of both the causal and the acausal filters and also to the inclusion of the pads. Accurate prediction of the static residual displacement (SRD) is very important for structures traversing faults in the near-fault regions. The greatest adverse effect of the high pass filters is their removal of the SRD. However, the noise-free displacements obtained by double integrating the synthetic accelerogram analytically preserve the SRD. It is thus apparent that conventional high pass filters should not be used for processing near-fault strong-motion records although they can be reliably used for far-fault records if applied acausally. The ground motion parameters such as ARIAS intensity, HUSID plots, Housner spectral intensity and the duration of strong-motion are found to be insensitive to the causality of filters.

• Journal of Korean Society of Transportation
• /
• v.34 no.2
• /
• pp.158-167
• /
• 2016

This study evaluates potential sources of errors in k-NN(k-nearest neighbor) algorithm such as procedures, variables, and input data. Previous research has been thoroughly reviewed for understanding fundamentals of k-NN algorithm that has been widely used for short-term traffic states prediction. The framework of this algorithm commonly includes historical data smoothing, pattern database, similarity measure, k-value, and prediction horizon. The outcomes of this study suggests that: i) historical data smoothing is recommended to reduce random noise of measured traffic data; ii) the historical database should contain traffic state information on both normal and event conditions; and iii) trial and error method can improve the prediction accuracy by better searching for the optimum input time series and k-value. The study results also demonstrates that predicted error increases with the duration of prediction horizon and rapidly changing traffic states.

• The Journal of Bigdata
• /
• v.4 no.2
• /
• pp.1-12
• /
• 2019

As Artificial Intelligence (AI) technology develops, it is applied to various fields such as image, voice, and text. AI has shown fine results in certain areas. Researchers have tried to predict the stock market by utilizing artificial intelligence as well. Predicting the stock market is known as one of the difficult problems since the stock market is affected by various factors such as economy and politics. In the field of AI, there are attempts to predict the ups and downs of stock price by studying stock price patterns using various machine learning techniques. This study suggest a way of predicting stock price patterns based on the Convolutional Neural Network(CNN) among machine learning techniques. CNN uses neural networks to classify images by extracting features from images through convolutional layers. Therefore, this study tries to classify candlestick images made by stock data in order to predict patterns. This study has two objectives. The first one referred as Case 1 is to predict the patterns with the images made by the same-day stock price data. The second one referred as Case 2 is to predict the next day stock price patterns with the images produced by the daily stock price data. In Case 1, data augmentation methods - random modification and Gaussian noise - are applied to generate more training data, and the generated images are put into the model to fit. Given that deep learning requires a large amount of data, this study suggests a method of data augmentation for candlestick images. Also, this study compares the accuracies of the images with Gaussian noise and different classification problems. All data in this study is collected through OpenAPI provided by DaiShin Securities. Case 1 has five different labels depending on patterns. The patterns are up with up closing, up with down closing, down with up closing, down with down closing, and staying. The images in Case 1 are created by removing the last candle(-1candle), the last two candles(-2candles), and the last three candles(-3candles) from 60 minutes, 30 minutes, 10 minutes, and 5 minutes candle charts. 60 minutes candle chart means one candle in the image has 60 minutes of information containing an open price, high price, low price, close price. Case 2 has two labels that are up and down. This study for Case 2 has generated for 60 minutes, 30 minutes, 10 minutes, and 5minutes candle charts without removing any candle. Considering the stock data, moving the candles in the images is suggested, instead of existing data augmentation techniques. How much the candles are moved is defined as the modified value. The average difference of closing prices between candles was 0.0029. Therefore, in this study, 0.003, 0.002, 0.001, 0.00025 are used for the modified value. The number of images was doubled after data augmentation. When it comes to Gaussian Noise, the mean value was 0, and the value of variance was 0.01. For both Case 1 and Case 2, the model is based on VGG-Net16 that has 16 layers. As a result, 10 minutes -1candle showed the best accuracy among 60 minutes, 30 minutes, 10 minutes, 5minutes candle charts. Thus, 10 minutes images were utilized for the rest of the experiment in Case 1. The three candles removed from the images were selected for data augmentation and application of Gaussian noise. 10 minutes -3candle resulted in 79.72% accuracy. The accuracy of the images with 0.00025 modified value and 100% changed candles was 79.92%. Applying Gaussian noise helped the accuracy to be 80.98%. According to the outcomes of Case 2, 60minutes candle charts could predict patterns of tomorrow by 82.60%. To sum up, this study is expected to contribute to further studies on the prediction of stock price patterns using images. This research provides a possible method for data augmentation of stock data.

• Geophysics and Geophysical Exploration
• /
• v.25 no.3
• /
• pp.140-161
• /
• 2022

Seismic data with missing traces are often obtained regularly or irregularly due to environmental and economic constraints in their acquisition. Accordingly, seismic data interpolation is an essential step in seismic data processing. Recently, research activity on machine learning-based seismic data interpolation has been flourishing. In particular, convolutional neural network (CNN) and generative adversarial network (GAN), which are widely used algorithms for super-resolution problem solving in the image processing field, are also used for seismic data interpolation. In this study, CNN-based algorithm, U-Net and GAN-based algorithm, and conditional Wasserstein GAN (cWGAN) were used as seismic data interpolation methods. The results and performances of the methods were evaluated thoroughly to find an optimal interpolation method, which reconstructs with high accuracy missing seismic data. The work process for model training and performance evaluation was divided into two cases (i.e., Cases I and II). In Case I, we trained the model using only the regularly sampled data with 50% missing traces. We evaluated the model performance by applying the trained model to a total of six different test datasets, which consisted of a combination of regular, irregular, and sampling ratios. In Case II, six different models were generated using the training datasets sampled in the same way as the six test datasets. The models were applied to the same test datasets used in Case I to compare the results. We found that cWGAN showed better prediction performance than U-Net with higher PSNR and SSIM. However, cWGAN generated additional noise to the prediction results; thus, an ensemble technique was performed to remove the noise and improve the accuracy. The cWGAN ensemble model removed successfully the noise and showed improved PSNR and SSIM compared with existing individual models.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.46 no.3
• /
• pp.139-147
• /
• 2023

The metal bush assembling process is a process of inserting and compressing a metal bush that serves to reduce the occurrence of noise and stable compression in the rotating section. In the metal bush assembly process, the head diameter defect and placement defect of the metal bush occur due to metal bush omission, non-pressing, and poor press-fitting. Among these causes of defects, it is intended to prevent defects due to omission of the metal bush by using signals from sensors attached to the facility. In particular, a metal bush omission is predicted through various data mining techniques using left load cell value, right load cell value, current, and voltage as independent variables. In the case of metal bush omission defect, it is difficult to get defect data, resulting in data imbalance. Data imbalance refers to a case where there is a large difference in the number of data belonging to each class, which can be a problem when performing classification prediction. In order to solve the problem caused by data imbalance, oversampling and composite sampling techniques were applied in this study. In addition, simulated annealing was applied for optimization of parameters related to sampling and hyper-parameters of data mining techniques used for bush omission prediction. In this study, the metal bush omission was predicted using the actual data of M manufacturing company, and the classification performance was examined. All applied techniques showed excellent results, and in particular, the proposed methods, the method of mixing Random Forest and SA, and the method of mixing MLP and SA, showed better results.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2009.10a
• /
• pp.797-803
• /
• 2009

The accurate prediction of dynamic characteristics of high speed rotors, such as gas turbines, is important to avoid the possibility of operating the machinery near the critical speeds or unstable speed regions. However, the dynamic analysis methods and softwares for gas turbines have been developed in the process of producing many gas turbines by manufacturers and most of them have seldom been disclosed to the public. Recently, commercial FEM softwares, such as SAMCEF, ANSYS and NASTRAN, started supporting some rotordynamics analysis modules based on 3-D finite elements. In this paper, the dynamic analysis method using commercial S/W, especially ANSYS, is attempted for the small-size gas turbine engine rotor, and the analysis capability and limitations of its rotordyamics module are evaluated for further improvement of the module. As the preliminary procedure, the rotordyamic analysis capability of ANSYS was tested and evaluated with the reference models of the well-known dynamics. The limitations in application of the rotordynamics module were then identified. Under the current capability and limitations of ANSYS, it is shown that Lee diagram, a new frequency-speed diagram enhanced with the concept of $H{\infty}$ in rotating machinery, can be indirectly obtained from FRFs computed from harmonic response analysis of ANSYS. Finally, it is demonstrated based on the modeling and analysis method developed in the process of the S/W verification that the conventional Campbell diagram, Lee diagram, mode shapes and critical speeds of the small-size gas turbine engine rotor can be computed using the ANSYS rotordynamics module.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.20 no.1
• /
• pp.36-44
• /
• 2010

The accurate prediction of dynamic characteristics of high speed rotors, such as gas turbines, is important to avoid the possibility of operating the machinery near the critical speeds or unstable speed regions. However, the dynamic analysis methods and softwares for gas turbines have been developed in the process of producing many gas turbines by manufacturers and most of them have seldom been disclosed to the public. Recently, commercial FEM softwares, such as SAMCEF, ANSYS and NASTRAN, started supporting some rotordynamics analysis modules based on 3-D finite elements. In this paper, the dynamic analysis method using commercial S/W, especially ANSYS, is attempted for the small-size gas turbine engine rotor, and the analysis capability and limitations of its rotordyamics module are evaluated for further improvement of the module. As the preliminary procedure, the rotordyamic analysis capability of ANSYS was tested and evaluated with the reference models of the well-known dynamics. The limitations in application of the rotordynamics module were then identified. Under the current capability and limitations of ANSYS, it is shown that Lee diagram, a new frequency-speed diagram enhanced with the concept of $H{\infty}$ in rotating machinery, can be indirectly obtained from FRFs computed from harmonic response analysis of ANSYS. Finally, it is demonstrated based on the modeling and analysis method developed in the process of the S/W verification that the conventional Campbell diagram, Lee diagram, mode shapes and critical speeds of the small-size gas turbine engine rotor can be computed using the ANSYS rotordynamics module.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.15 no.12 s.105
• /
• pp.1389-1397
• /
• 2005

Since blast-induced vibration may cause serious problem to the rock mass as well as the nearby structures, the prediction of blast-induced nitration and the stability evaluation must be performed before blasting activities. Dynamic analysis has been increased recently in order to analyze the effect of the blast-Induced vibration. Most of the previous studies, however, were based on the continuum analysis unable to consider rock joints which significantly affect the wave propagation and attenuation characteristics. They also adopted pressure corves estimated tv theoretical or empirical equations as input detonation load, thus there were very difficult to reflect the characteristics of propagating media. In this study, therefore, we suggested a dynamic distinct element analysis technique which uses velocity waveform obtained from a test blast as an input detonation load. A distinct element program, UDEC was used to consider the effect of rock joints. In order to verify the validity of proposed method, the test blast was simulated. The predicted results from the proposed method showed a good agreement with the measured vibration data from the test blast. Through the dynamic numerical modelling on the planned road tunnel and slope, we evaluated the effect of blast-induced nitration and the stability of rock slope.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.24 no.4
• /
• pp.447-453
• /
• 2016

The reduction technologies of exhaust gas from both the off-road engine and on-road vehicles are important. It is possible to apply various combustion technologies with engines after the application of a treatment technology to this field. In this study, main injection timing, pilot injection timing, and exhaust gas recirculation (EGR) rate were selected as the experimental parameters whose effects on the emission of exhaust gases and on the fuel consumption characteristics were to be determined. In the experiment, the emission of nitrogen oxide (NOx) and Smoke, and the Torque at the same fuel consumption level, were measured. The experimental data were analyzed using the Taguchi method with an L9 orthogonal array. Additionally, analysis of variation (ANOVA) was used to confirm the influence of each parameter. Consequently, the level of each parameter was selected based on the signal-to-noise ratio data (main injection timing, 3; pilot injection timing, 3; EGR rate, 2), and the results of the Taguchi prediction were verified experimentally (error: NOx, 10.3 %; Smoke, 6.6 %; brake-specific fuel consumption (BSFC), 0.6 %).

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.12
• /
• pp.1772-1781
• /
• 2017

In this study, the design methodology for alleviating the overfitting problem of Polynomial Neural Networks(PNN) is realized with the aid of two kinds techniques such as L2 regularization and Sum of Squared Coefficients (SSC). The PNN is widely used as a kind of mathematical modeling methods such as the identification of linear system by input/output data and the regression analysis modeling method for prediction problem. PNN is an algorithm that obtains preferred network structure by generating consecutive layers as well as nodes by using a multivariate polynomial subexpression. It has much fewer nodes and more flexible adaptability than existing neural network algorithms. However, such algorithms lead to overfitting problems due to noise sensitivity as well as excessive trainning while generation of successive network layers. To alleviate such overfitting problem and also effectively design its ensuing deep network structure, two techniques are introduced. That is we use the two techniques of both SSC(Sum of Squared Coefficients) and $L_2$ regularization for consecutive generation of each layer's nodes as well as each layer in order to construct the deep PNN structure. The technique of $L_2$ regularization is used for the minimum coefficient estimation by adding penalty term to cost function. $L_2$ regularization is a kind of representative methods of reducing the influence of noise by flattening the solution space and also lessening coefficient size. The technique for the SSC is implemented for the minimization of Sum of Squared Coefficients of polynomial instead of using the square of errors. In the sequel, the overfitting problem of the deep PNN structure is stabilized by the proposed method. This study leads to the possibility of deep network structure design as well as big data processing and also the superiority of the network performance through experiments is shown.