• 센서학회지
• /
• 제31권1호
• /
• pp.16-23
• /
• 2022

Cardiac disease is the most common cause of death worldwide. Therefore, detection and classification of electrocardiogram (ECG) signals are crucial to extend life expectancy. In this study, we aimed to implement an artificial intelligence signal recognition system in field programmable gate array (FPGA), which can recognize patterns of bio-signals such as ECG in edge devices that require batteries. Despite the increment in classification accuracy, deep learning models require exorbitant computational resources and power, which makes the mapping of deep neural networks slow and implementation on wearable devices challenging. To overcome these limitations, spiking neural networks (SNNs) have been applied. SNNs are biologically inspired, event-driven neural networks that compute and transfer information using discrete spikes, which require fewer operations and less complex hardware resources. Thus, they are more energy-efficient compared to other artificial neural networks algorithms.

• Wind and Structures
• /
• 제38권1호
• /
• pp.75-91
• /
• 2024

Reliable wind signal reconstruction can be beneficial to the operational safety of long-span bridges. Non-Gaussian characteristics of wind signals make the reconstruction process challenging. In this paper, non-Gaussian wind signals are converted into a combined prediction of two kinds of features, actual wind speeds and wind angles of attack. First, two decomposition techniques, empirical mode decomposition (EMD) and variational mode decomposition (VMD), are introduced to decompose wind signals into intrinsic mode functions (IMFs) to reduce the randomness of wind signals. Their principles and applicability are also discussed. Then, four artificial intelligence (AI) algorithms are utilized for wind signal reconstruction by combining the particle swarm optimization (PSO) algorithm with back propagation neural network (BPNN), support vector regression (SVR), long short-term memory (LSTM) and bidirectional long short-term memory (Bi-LSTM), respectively. Measured wind signals from a bridge site in a deep-cutting gorge are taken as experimental subjects. The results showed that the reconstruction error of high-frequency components of EMD is too large. On the contrary, VMD fully extracts the multiscale rules of the signal, reduces the component complexity. The combination of VMD-PSO-Bi-LSTM is demonstrated to be the most effective among all hybrid models.

• Journal of Power Electronics
• /
• 제19권5호
• /
• pp.1224-1234
• /
• 2019

Hybrid rotor position estimation methods that integrate a fundamental model and high frequency (HF) signal injection are widely used for the wide speed-range sensorless control of interior permanent-magnet synchronous machines (IPMSMs). However, the direct transition of two different schemes may lead to system fluctuations or system instability since two estimated rotor positions based on two different schemes are always unequal due to the effects of parameter variations, system delays and inverter nonlinearities. In order to avoid these problems, a seamless transition strategy to define and construct a virtual q-axis inductance is proposed in this paper. With the proposed seamless transition strategy, an estimated rotor position based on a fundamental model is forced to track that based on HF signal injection before the transition by adjusting the constructed virtual q-axis inductance. Meanwhile, considering that the virtual q-axis inductance changes with rotor position estimation errors, a new observer with a two-phase phase-locked loop (TP-PLL) is developed to accurately obtain the virtual q-axis inductance online. Furthermore, IPMSM sensorless control with maximum torque per ampere (MTPA) operations can be tracked automatically by selecting the proper virtual q-axis inductance. Finally, experimental results obtained from an IPMSM demonstrate the feasibility of the proposed seamless transition strategy.

• 한국컴퓨터정보학회논문지
• /
• 제28권11호
• /
• pp.155-160
• /
• 2023

본 논문은 광혈류신호를 이용하여 혈압을 예측하는 방법을 제시한다. 제시한 방법은 먼저, 광혈류신호를 측정한 후, 전처리 과정을 통해 아티펙트를 제거하고 학습을 위한 신호를 얻는다. 그리고 혈압에 영향을 주는 몸무게와 키를 부가 정보로 측정한다. 다음으로, 인공지능 알고리즘을 통해 광혈류신호, 키, 그리고 몸무게를 입력변수로 학습하여 수축기와 이완기 혈압을 추정하도록 시스템을 구축한다. 구축된 시스템은 사전에 입력된 키와, 몸무게, 그리고 측정한 광혈류신호를 가지고 수축기와 이완기 혈압을 예측한다. 제안한 방법은 무구속 방식으로 피검자의 키와 몸무게, 그리고 심장 및 혈관의 상태를 반영하는 광혈류신호를 입력받아 실시간, 연속적으로 혈압 예측이 가능하다. 본 연구에서 제시한 인공지능 기반 혈압예측시스템의 유용성을 확인하기 위해 측정한 혈압과 예측한 혈압의 비교를 통해 결과의 유용성을 확인한다.

• 융합신호처리학회논문지
• /
• 제24권4호
• /
• pp.213-220
• /
• 2023

낙상은 일상의 활동 중에 예기치 않게 발생하여 생활에 많은 어려움을 초래한다. 본 연구는 고위험 직종 종사자들의 낙상 감지를 위한 시스템을 구성하고 자료를 수집하여 예측 모델에 적용함으로써 그 유효성을 검증하는 것을 목적으로 하였다. 이를 위해 가속도센서와 자이로센서를 통해 가속도 신호와 방위각을 산출하여 낙상 여부를 감지하는 웨어러블 기기를 구성하였다. 그리고 연구 참여자들이 이 기기를 복부에 착용하고 정해진 활동을 수행하는 과정에서 낙상과 관련한 동작으로부터 필요한 데이터를 측정하고 기기 내에 존재하는 블루투스 장치를 통해 컴퓨터로 전송하였다. 이렇게 수집된 데이터를 필터링 등을 통해 처리하여 딥러닝 알고리즘들인 1D CNN, LSTM, CNN-LSTM에 근거한 낙상 감지 예측 모델들에 적용하고 그 결과를 평가하였다.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2010년도 춘계학술대회
• /
• pp.512-515
• /
• 2010

공간상에 존재하는 전자파를 측정, 분석하여 신호정보(SIGINT; SIGnal INTelligence)를 획득하기 위해서 가장 중요한 것이 전자파의 주파수 정보이다. 특히 레이다 및 미사일에서 방사되는 초고주파 대역의 주파수를 순시 측정하는 방법으로, 지연선로의 위상차를 측정하여 주파수정보를 디지털데이터를 출력하는 부품으로 디지털 주파수 판별기(Digital frequency Discriminator; DFD)가 있다. DFD는 100nSec 이하의 짧은 시간동안에 존재하는 고주파 신호에 대해서도 초고주파신호의 주파수 정보를 실시간으로 측정하여 제공해야 한다. 본 논문에서는 광대역 4 채널의 지연선로와 코릴레이터로 구성된 고주파 입력부와 I/Q신호를 처리하여 주파수 정보를 얻어내는 디지털 처리부 및 정확한 주파수 정보를 얻기 위한 주파수 보정부로 이루어진 DFD의 구현방안을 제안하고 아주 짧은 펄스 형태의 모의 레이다 신호를 입력하여 얻은 시험결과를 토대로 설계의 타당성을 확인한다.

• Investigative Magnetic Resonance Imaging
• /
• 제26권2호
• /
• pp.125-134
• /
• 2022

Purpose: The aim of this study was to investigate the change in signal sensitivity over different acquisition start times and optimize the scanning window to provide the maximal signal sensitivity of [1-¹³C]pyruvate and its metabolic products, lactate and alanine, using spatially localized hyperpolarized 3D ¹³C magnetic resonance spectroscopic imaging (MRSI). Materials and Methods: We acquired 3D ¹³C MRSI data from the brain (n = 3), kidney (n = 3), and liver (n = 3) of rats using a 3T clinical scanner and a custom RF coil after the injection of hyperpolarized [1-¹³C]pyruvate. For each organ, we obtained three consecutive 3D ¹³C MRSI datasets with different acquisition start times per animal from a total of three animals. The mean signal-to-noise ratios (SNRs) of pyruvate, lactate, and alanine were calculated and compared between different acquisition start times. Based on the SNRs of lactate and alanine, we identified the optimal acquisition start timing for each organ. Results: For the brain, the acquisition start time of 18 s provided the highest mean SNR of lactate. At 18 s, however, the lactate signal predominantly originated from not the brain, but the blood vessels; therefore, the acquisition start time of 22 s was recommended for 3D ¹³C MRSI of the rat brain. For the kidney, all three metabolites demonstrated the highest mean SNR at the acquisition start time of 32 s. Similarly, the acquisition start time of 22 s provided the highest SNRs for all three metabolites in the liver. Conclusion: In this study, the acquisition start timing was optimized in an attempt to maximize metabolic signals in hyperpolarized 3D ¹³C MRSI examination with [1-¹³C] pyruvate as a substrate. We investigated the changes in metabolic signal sensitivity in the brain, kidney, and liver of rats to establish the optimal acquisition start time for each organ. We expect the results from this study to be of help in future studies.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 2006년도 정기 학술대회 논문집
• /
• pp.89-96
• /
• 2006

Nondestructive evaluation using surface waves needs an analytical solution for the reference value to compare with experimental data. Finite element analysis is very powerful tool to simulate the wave propagation, but has some defects. It is very expensive and high time-complexity for the required high resolution. For those reasons, it is hard to implement an optimization problem in the actual situation. The developed engine in this paper can substitute for the finite element analysis of surface waves propagation, and it accomplishes the fast analysis possible to be used in optimization. Including this artificial intelligence engine, most of soft computing algorithms can be applied on the special database. The database of surface waves propagation is easily constructed with the results of finite element analysis after reducing the dimensions of data. The principal wavelet-component analysis is an efficient method to simplify the transient wave signal into some representative peaks. At the end, artificial neural network based on the database make it possible to invent the artificial intelligence engine.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1987년도 전기.전자공학 학술대회 논문집(II)
• /
• pp.1583-1586
• /
• 1987

In the study for effective speech control we designed a personal computer system with A/D converter in which the speech signal is transformed by digital data displayed graphically on the moniter and with a D/A converter in which the digital data is transformed into speech signal which people can hear. We analyzed the character of the speech signal produced by the system. We designed the adaptive noise cancel algorithm so that noise and Interference are cancelled whenever the speech signal is recognized by the computer system. This is a basic system for artificial Intelligence.

• International Journal of Control, Automation, and Systems
• /
• 제6권1호
• /
• pp.54-61
• /
• 2008

This paper presents a method for reducing modelling error in the linear motion system with voicecoil actuator (VCA). A model of linear motion system composed of a mechanism and control was prepared to verify the proposed method. In modeling of the system, the damping coefficient obtained experimentally is applied to the model in order to consider the effect of the viscous friction for the moving part in VCA. The response velocity of VCA for duty ratio of PWM signal was analyzed in the time domain. Consequently, the relation between velocity and duty ratio was obtained. The result from the experiment showed an error of 9% when compared with that of simulation. In order to reduce the modeling error, impedance variation according to input frequency was analyzed, and equivalent impedance with multi-frequency was applied to the control part. As a result, the modeling error decreased to 5%.