• 대한의용생체공학회:의공학회지
• /
• 제15권3호
• /
• pp.305-316
• /
• 1994

All physical data in the real world are nonstationary signals that have the time varying statistical characteristics. Although few algorithms suitable to process the nonstationary signals have ever been suggested, these are treated the nonstationary signals under the assumption that the nonstationary signal is a piece-wise stationary signal. Recently, statistical analysis algorithms for the nonstationary signal have concentrated so much interest. In this paper, nonstationary EMG signals are mapped onto the orthogonal wavelet transform domain so that the eigenvalue spread of its autocorrelation matrix could be more smaller than that in the time domain. Then the model in the wavelet transform domain and an algorithm to estimate the model parameters are suggested. Also, an test signal generated by a white gaussian noise and the EMG signal are identified, and the algorithm performance is considered in the sense of the mean square error and the evaluation parameters.

• 대한의용생체공학회:의공학회지
• /
• 제16권3호
• /
• pp.317-324
• /
• 1995

A need exists for the accurate identification of time series models having time varying parameters, as is important in the case of real time identification of nonstationary EMG signal. Thls paper describes real time identification and muscle fatigue monitoring method of nonstationary EMG signal. The method is composed of the efficient identifier which estimates the autoregressive parameters of nonstationary EMG signal model, and its real time implementation by using T805 parallel processing computer. The method is verified through experiment with real EMG signals which are obtained from surface electrode. As a result, the proposed method provides a new approach for real time Implementation of muscle fatigue monitoring and the execution time is 0.894ms/sample for 1024Hz EMG signal.

• 대한의용생체공학회:의공학회지
• /
• 제15권2호
• /
• pp.167-174
• /
• 1994

In this paper, a DSP and microcomputer-based EMG controlled functional electrical stimulation (FES) system, for restoring walking of paraplegics at the patients' own command, is presented. The above-lesion EMG is a time-varying nonstationary signal and its autoregressive (AR) parameters are identified by the nonstationary identification algorithm using a DSP chip. The identified AR parameters are used for the cloassification of the function and the control of the movement. The below-lesion response-EMG signal is used as a measure of muscle fatigue. This FES system is designed to measure muscle fatigue and control the stimulation intensity according to the amplitude of the response-EMG signal. While the automatic electrical intensity control is obtained by identifying the movement, the proposed FES system is suitable for the automatic control of paraplegic walking.

• 대한의용생체공학회:의공학회지
• /
• 제16권4호
• /
• pp.507-516
• /
• 1995

근전도 신호(electromyogram)의 시변 비정상(time varying nonstationary) 특성은 신호의 정확한 모델링 및 인식에 제약 조건으로 받아들여 졌다. 특히, 최근 들어 장애자들을 위한 보철제어분야에서 근전도 신호를 이용한 기능적 전기 자극을 위한 FES(funcitonal electrical stimulation) 시스템에 있어 근전도 신호의 파라메터 인식은 중요한 요소로서 작용한다. 그러나, 근전도 신호는 자세의 변화 및 근육 피로도 등의 요인에 의해서 시변 비정상 특성을 띠고 있기 때문에 시간에 따라 변하는 인식 파라메터를 정확하게 인식할 수 있는 새로운 알고리즘의 개발과 실시간 처리가 가능한 컴퓨터 하드웨어의 설계가 요구된다. 따라서, 본 논문에서는 시평면의 근전도 신호를 이산 여현 변환(discrete cosine transform)을 이용하여 변환 평면으로 옮긴 다음 상태 방정식(state space equation)을 써서 변환 평면상에서의 AR(autoregressive) 모델을 세우고 주어진 근전도 신호에 대해 모델 파라메터를 추정하였으며, 제안한 알고리즘은 실시간 처리를 위하여 2개의 독립적인 중앙 연산 처리 장치를 갖춘 INMOS사의 IMS T-805 병렬 처리 컴퓨터를 이용하여 동시 다발적인 연산을 수행함으로서 알고리즘의 연산 효율을 높였다. 제안된 알고리즘의 타당성을 검증하기 위해 모델의 추정 오차에 영향을 미치는 입력 자기상관 행렬(input correlation matrix)의 condition number의 변화 및 평균자승오차(mean square error)를 구하여 기존의 SLS(sequential least square) 알고리즘과 비교하였다.

• 대한인간공학회지
• /
• 제30권6호
• /
• pp.689-694
• /
• 2011

Objective: The aim of this study is to evaluate the fluctuation of signal amplitude during repetitive dynamic contraction based on surface electromyography(EMG). Background: The most previous studies were considered isometric muscle contraction and they were difference to smoothing window length by moving average filter. In practical, the human movement is dynamic state. Dynamic EMG signal which indicated as the nonstationary pattern should be analyzed differently compared with the static EMG signal. Method: Ten male subjects participated in this experiment, and EMG signal was recorded by biceps brachii, anterior/posterior deltoid, and upper/lower trapezius muscles. The subject was performed to repetitive right horizontal lifting task during ten cycles. This study was considered three independent variables(muscle, amplitude processing technique, and smoothing window length) as the within-subject experimental design. This study was estimated muscular activation by means of the linear envelope technique(LE). The dependent variable was set coefficient of variation(CV) of LE for each cycle. Results: The ANOVA results showed that the main and interaction effects between the amplitude processing technique and smoothing window length were significant difference. The CV value of peak LE was higher than mean LE. According to increase the smoothing window length, this study shows that the CV trend of peak LE was decreased. However, the CV of mean LE was analyzed constant fluctuation trend regardless of the smoothing window length. Conclusion: Based on these results, we expected that using the mean LE and 300ms window length increased reproducibility and signal noise ratio during repetitive dynamic muscle contraction. Application: These results can be used to provide fundamental information for repetitive dynamic EMG signal processing.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.842-847
• /
• 2005

This paper proposes a novel real-time EMG pattern recognition for the control of a multifunction myoelectric hand from four channel EMG signals. To cope with the nonstationary signal property of the EMG, features are extracted by wavelet packet transform. For dimensionality reduction and nonlinear mapping of the features, we also propose a linear-nonlinear feature projection composed of PCA and SOFM. The dimensionality reduction by PCA simplifies the structure of the classifier, and reduces processing time for the pattern recognition. The nonlinear mapping by SOFM transforms the PCA-reduced features to a new feature space with high class separability. Finally a multilayer neural network is employed as the pattern classifier. We implement a real-time control system for a multifunction virtual hand. From experimental results, we show that all processes, including virtual hand control, are completed within 125 msec, and the proposed method is applicable to real-time myoelectric hand control without an operation time delay.

• 전자공학회논문지SC
• /
• 제42권2호
• /
• pp.39-48
• /
• 2005

본 논문에서는 다기능 근전의수를 제어하기 위해 전완에서 취득한 4 채널의 근전도로부터 9 가지 동작을 인식하는 새로운 방법을 제안한다. 비정상 신호특성을 가진 근전도를 해석하기 위해서 시간-주파수 영역에서 표현되는 특징벡터를 웨이블렛 패킷변환을 통해 추출한다. 높은 차원을 가지는 시간-주파수 특징벡터에 대하여 차원축소와 비선형변환을 수행하기 위해 PCA와 SOFM으로 구성된 특징투영 방법을 제안한다. PCA를 이용한 차원축소는 패턴분류기의 구조를 단순화하고 패턴인식을 위한 계산시간을 단축할 수 있다. SOFM을 이용한 비선형변환은 PCA에 의해 차원이 축소된 특징벡터를 새로운 공간으로 투영함으로써 클래스 분리도를 향상시킨다. 마지막으로 각 동작은 패턴분류기인 다층 신경회로망에 의해 인식된다. 실험 결과로부터 제안한 방법이 높은 인식률을 보임과 동시에 연속적인 패턴인식을 위한 실시간 구현이 가능함을 보인다.