• 대한임베디드공학회논문지
• /
• 제14권6호
• /
• pp.287-294
• /
• 2019

With the automation of production lines in the manufacturing industry, the importance of real-time fault diagnosis of facility is increasing. In this paper, we propose a fault diagnosis algorithm of LM (Linear Motion)-guide based on deep learning using vibration signals. Generally, in order to guarantee the performance of the deep learning, it is necessary to have a sufficient amount of data, but in a manufacturing industry, it is often difficult to obtain enough data due to physical and time constraints. To solve this problem, we propose a convolutional neural networks (CNN) model based on transfer learning. In addition, the spectrogram image is input to the CNN to reflect the frequency characteristic of the vibration signals with time. The performance of fault diagnosis according to various load condition and transfer learning method was compared and evaluated by experiments. The results showed that the proposed algorithm exhibited an excellent performance.

• 한국지반공학회지:지반
• /
• 제13권1호
• /
• pp.111-122
• /
• 1997

표면파를 이용하여 지반의 강성을 추정하는 기법인 SASW 시험에서 위상속도(phase volocity)를 결정하기 위해서는 위상각(phase angle)의 전개(unwrapping)가 필수적이다. 포장 구조에서처럼 깊이에 따라 강성의 차이가 현저한 경우는 기존의 위상각 전개방식으조는 정확한 위상속도를 결정하기가 용이하지 않다. 이는 기존의 위상각 전개방식은 주위상각(principal phase angle)에 2n의 정수배를 더하는 것인데, 위상각 스펙트럼(phase spectrum)에서 정수배를 결정하는 데에 어려움이 있기 때문이다. 본 연구에서는 이러한 문제점을 해결하기 위해서, 임펄스 응답 필터 기법(Impulse Response Filtration Technique), 또는 IRF기법이라고 하는 새로운 위상각 분석 기법을 제안하였다. IRF 기법의 원리는 임펄스 응답을 필터 처리함으로써 파군(wave group)을 분리하는 것인데,파군의 분리는 임펄스 응답에 대한 Gabor spectrogram을 분석한 정보를 근거로 한다. Gabor spectrogram은 전파되는 파의 에너지를 주파수-시간 공간에서 나타내는 contour 그림으로서, 파군의 전파 상황을 시각적으로 표현하는 수단이다. 이렇게 필터 처리된 임펄스 응답을 이용하면, 위상각 스펙트럼의 분석을 정확하게 할 수 있으며, 위상각의 전개에 있어서 난해함을 제거할 수 있다. 끝으로, 전쳔적인 포장 구조에 대하여 이론적으로 SASW 시험을 모사하였으며, 그 결과를 이용하여 IRF기법의 효용성을 입증하였다.

• 대한의용생체공학회:의공학회지
• /
• 제7권2호
• /
• pp.139-146
• /
• 1986

본 논문은 tms 32010이라는 디지탈 신호 처리용소자를 사용하여 초음하 펄스 도플러 장치를 구현한 것이다. 도플러 장피란 초음파 신호의 송수신 과정에서 발생하는 도플러 효과를 이용하여 혈류의 속도를 측정하는 장치이다. 한 점에서의 속도를 측정하는 단일채널 도플러 장치에서는 실시간 고속 푸리에 변환기를 구현하여 도플러 주하수의 스펙트럼을 측정함으로서 혈류속도를 측정하며 초음파 빔의 일직선상에서의 여러점을 동시에 측정하는 다중채널 도플러 장치에서는 영점교차검출기를 구현하여 평균주파수를 측정하였다. 자중채널 장치는 직렬처리법을 사용하여 하드웨어를 간단히 하였으며 8점에서의 속도를 측정할 수 있도록 하였다.

• 한국지진공학회논문집
• /
• 제28권4호
• /
• pp.183-191
• /
• 2024

Accurate seismic vulnerability assessment requires high quality and large amounts of ground motion data. Ground motion data generated from time series contains not only the seismic waves but also the background noise. Therefore, it is crucial to determine the high-pass cut-off frequency to reduce the background noise. Traditional methods for determining the high-pass filter frequency are based on human inspection, such as comparing the noise and the signal Fourier Amplitude Spectrum (FAS), f² trend line fitting, and inspection of the displacement curve after filtering. However, these methods are subject to human error and unsuitable for automating the process. This study used a deep learning approach to determine the high-pass filter frequency. We used the Mel-spectrogram for feature extraction and mixup technique to overcome the lack of data. We selected convolutional neural network (CNN) models such as ResNet, DenseNet, and EfficientNet for transfer learning. Additionally, we chose ViT and DeiT for transformer-based models. The results showed that ResNet had the highest performance with R² (the coefficient of determination) at 0.977 and the lowest mean absolute error (MAE) and RMSE (root mean square error) at 0.006 and 0.074, respectively. When applied to a seismic event and compared to the traditional methods, the determination of the high-pass filter frequency through the deep learning method showed a difference of 0.1 Hz, which demonstrates that it can be used as a replacement for traditional methods. We anticipate that this study will pave the way for automating ground motion processing, which could be applied to the system to handle large amounts of data efficiently.

• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2007년도 한국음성과학회 공동학술대회 발표논문집
• /
• pp.143-145
• /
• 2007

Muscle groups that are located in and around the vocal tract can produce audible changes in frequency and/or intensity of the voice. Vocal vibrato is a characteristic feature in the singing of performers trained in the western classical tradition and vibrato is generally considered to result from modulation in frequency amplitude and timbre. Vocal tremor is also characterized by periodic fluctuations in the voice frequency or intensity and vocal tremor is symptom of a neurological disease as Spasmodic dysphonia , Parkinson's disease. Vocal vibrato and Vocal tremor may have many of the same origins and mechanisms in the voice production systems. The purpose of this study is to find acostic character of Korean traditional song Pansori singer's vibrato and Spasmodic dysphonia patient's vocal tremor. twelve Pansori singers and seven Spasmodic dysponia patients participated to this study. Power spectrum and Real time Spectrogram are used to analyze the acoustic characteristics of Pansori singing and Spasmodic dysphonia patient's voice The results are as follows; First, vowel formant differences between Pansori singing and Spasmodic dysphonia patient's voice are higher F1, F3. Second, The vibrato rate show differences between Pansori singing and Spasmodic dysphonia patients;$4^{\sim}6/sec$ and $5{\sim}6/sec$ Vibrato rate of pitch is 5.7 Hz ${\sim}$ 42.4 Hz for Pansori singing , 3.8 Hz ${\sim}$ 27.9 Hz for Spasmodic dysphonia patients ;Vibrato rate of intensity range is 0.07 dB ${\sim}$ 8.26 dB for Pansori singing and 0.07 dB ${\sim}$ 4.81 dB for Spasmodic dysphonia patients

• 음성과학
• /
• 제9권3호
• /
• pp.3-16
• /
• 2002

We examined some problems in speaker identification procedures: transformation of acoustic parameters into auditory scales, invalid measurement values, and comparability of spectral energy values across the frequency range. To resolve those problems, we analyzed the acoustic spectral energy of three Korean numbers produced by ten female students from narrow-band spectrograms at 19 proportional time points of each voiced segment. Then, cells of the first five spectral matrices were averaged to form a matrix model for each speaker. The correlation coefficients and sum of the absolute amplitude difference in each pair of the spectral models of the ten subjects were obtained. Also, some individual matrix models were compared to those of the same subject or the other subject with a similar spectral model. Results showed that in numbers '2' and '9' subjects could not be clearly distinguished from the others but in number '4' it shed some possibility of setting threshold values for speaker identification if we employed the coefficients and the sum of absolute difference. Further studies would be desirable on various combinations of the range of long-term average spectra and the degree of signal pre-emphasis.

• 대한임베디드공학회논문지
• /
• 제15권3호
• /
• pp.119-127
• /
• 2020

The detection and recognition of abnormal driving becomes crucial for achieving safety in Intelligent Transportation Systems (ITS). This paper presents a feature extraction method based on spectral data to train a neural network model for driving behavior recognition. The proposed method uses a two stage signal processing approach to derive time-saving and efficient feature vectors. For the first stage, the feature vector set is obtained by calculating variances from each frequency bin containing the power spectrum data. The feature set is further reduced in the second stage where an intersection method is used to select more significant features that are finally applied for training a neural network model. A stream of live signals are fed to the trained model which recognizes the abnormal driving behaviors. The driving behaviors considered in this study are weaving, sudden braking and normal driving. The effectiveness of the proposed method is demonstrated by comparing with existing methods, which are Particle Swarm Optimization (PSO) and Convolution Neural Network (CNN). The experiments show that the proposed approach achieves satisfactory results with less computational complexity.

• 대한통합의학회지
• /
• 제6권4호
• /
• pp.191-206
• /
• 2018

Purpose : This study investigated the correlation between speech intelligibility and acoustic measurements of speech sounds produced by the children with speech sound disorders and children without any diagnosed speech sound disorder. Methods : A total of 60 children with and without speech sound disorders were the subjects of this study. Speech samples were obtained by having the subjects? speak meaningful words. Acoustic measurements were analyzed on a spectrogram using the Multi-speech 3700 program. Speech intelligibility was determined according to a listener's perceptual judgment. Results : Children with speech sound disorders had significantly lower speech intelligibility than those without speech sound disorders. The intensity of the vowel /u/, the duration of the vowel /${\omega}$/, and the second formant of the vowel /${\omega}$/ were significantly different between both groups. There was no difference in voice onset time between the groups. There was a correlation between acoustic measurements and speech intelligibility. Conclusion : The results of this study showed that the speech intelligibility of children with speech sound disorders was affected by intensity, word duration, and formant frequency. It is necessary to complement clinical setting results using acoustic measurements in addition to evaluation of speech intelligibility.

• 한국항행학회논문지
• /
• 제26권5호
• /
• pp.365-372
• /
• 2022

관성 측정 센서는 사람 행동 인식 시스템에 주로 사용되는 센서들에 비해 크기가 작고 가벼우며 낮은 비용으로 시스템의 경량화를 달성할 수 있다. 따라서, 본 논문에서는 관성 측정 센서를 이용한 이진 신경망 기반 걸음걸이 패턴 분석 시스템을 제안하고, 연산 가속을 위한 FPGA 기반 가속기 설계 및 구현 결과를 제시한다. 관성 측정 센서를 통해 걸음걸이에 대한 6가지 신호를 측정하고, 단시간 푸리에 변환을 이용하여 스펙트로그램을 추출한다. 높은 정확도를 가지는 경량화 시스템을 갖추기 위해 걸음걸이 패턴 분류에 BNN (binarized neural network) 기반 구조를 사용하였고, 검증 결과 97.5%의 높은 정확도와 메모리 사용량이 합성곱 신경망에 비해 96.7% 감소한 것을 확인하였다. 이진 신경망의 연산 가속을 위해 FPGA를 이용한 하드웨어 가속기 구조로 설계하였다. 제안된 걸음걸이 패턴 분석 시스템은 24,158개의 logic, 14,669개의 register, 13.687 KB의 block memory를 사용하여 구현되어 62.35 MHz의 최대 동작 주파수에서 1.5ms 내에 연산이 완료되어 실시간 동작이 가능함을 확인하였다.

• Smart Structures and Systems
• /
• 제29권1호
• /
• pp.77-91
• /
• 2022

Various monitoring systems have been implemented in civil infrastructure to ensure structural safety and integrity. In long-term monitoring, these systems generate a large amount of data, where anomalies are not unusual and can pose unique challenges for structural health monitoring applications, such as system identification and damage detection. Therefore, developing efficient techniques is quite essential to recognize the anomalies in monitoring data. In this study, several machine learning techniques are explored and implemented to detect and classify various types of data anomalies. A field dataset, which consists of one month long acceleration data obtained from a long-span cable-stayed bridge in China, is employed to examine the machine learning techniques for automated data anomaly detection. These techniques include the statistic-based pattern recognition network, spectrogram-based convolutional neural network, image-based time history convolutional neural network, image-based time-frequency hybrid convolution neural network (GoogLeNet), and proposed ensemble neural network model. The ensemble model deliberately combines different machine learning models to enhance anomaly classification performance. The results show that all these techniques can successfully detect and classify six types of data anomalies (i.e., missing, minor, outlier, square, trend, drift). Moreover, both image-based time history convolutional neural network and GoogLeNet are further investigated for the capability of autonomous online anomaly classification and found to effectively classify anomalies with decent performance. As seen in comparison with accuracy, the proposed ensemble neural network model outperforms the other three machine learning techniques. This study also evaluates the proposed ensemble neural network model to a blind test dataset. As found in the results, this ensemble model is effective for data anomaly detection and applicable for the signal characteristics changing over time.