• 조명전기설비학회논문지
• /
• 제13권4호
• /
• pp.109-119
• /
• 1999

스위치드 리럭턴스 전동기(SRM)는 상대적으로 낮은 가격, 간단하고 견고한 구조, 제어의 용이성과 고효율을 가지기 때문에 가변속 구동에서 점점 응용범위가 확대되고 있다. 본 논문에서 신경망이론은 퍼지-신경망 제어기의 소속함수와 퍼지규칙을 결정하는데 사용하였으며, 신경망 에뮬레이터는 SRM의 전방향 동특성을 모사하는데 사용하였다. 에뮬레이터의 역전파 오차는 퍼지-신경망 제어기의 소속함수와 퍼지규칙을 개선하는 경로를 제공한다. 32비트 DSP(TNS329C31)는 고속연산과 퍼지-신경망 제어 알고리즘을 실현하는데 사용하였다. 시뮬레이션과 실험결과는 부하변화의 경우 제안된 제어방법이 속도응답에서 종래의 방법보다 우수하였다.

• Journal of Multimedia Information System
• /
• 제2권1호
• /
• pp.179-186
• /
• 2015

This paper describes the application of a simulated annealing to a random signal-based learning. The simulated annealing is used to generate the reinforcement signal which is used in the random signal-based learning. Random signal-based learning is similar to the reinforcement learning of neural network. It is poor at hill-climbing, whereas simulated annealing has an ability of probabilistic hill-climbing. Therefore, hybridizing a random signal-based learning with the simulated annealing can produce better performance than before. The validity of the proposed algorithm is confirmed by applying it to two different examples. One is finding the minimum of the nonlinear function. And the other is the optimization of fuzzy control rules using inverted pendulum.

• 융합신호처리학회논문지
• /
• 제5권2호
• /
• pp.132-137
• /
• 2004

본 논문에서는 다층 신경회로망의 구조를 가지며, 백프로퍼게이션 학습 알고리즘을 이용한 적응신호처리 시스템을 구현하였다. 최소자승 알고리즘을 이용한 적응 잡음 제거기는 기준 신호와 잡음과의 상관도에 영향을 많이 받고, 정보 신호가 잡음에 비하여 상대적으로 작은 경우에 한계를 보이고 있다. 이와 같은 잡음에 대하여 본 논문에서 제안된 시스템은 좋은 성능을 보인다. 또한, 은닉층의 수와 노드 수를 다르게 구성했을 경우에 시스템의 출력에 미치는 결과에 대하여 분석하였다. 제안된 적응 신호처리 시스템의 장점을 알아보기 위하여 성능 평가의 기준이 되는 최소자승 알고리즘을 이용한 시스템과 비교하였다.

• Journal of Welding and Joining
• /
• 제16권3호
• /
• pp.102-110
• /
• 1998

The horizontal fillet welding is prevalently used in heavy and ship building industries to fabricate the large scale structures. A deep understanding of the horizontal fillet welding process is restricted, because the phenomena occurring in welding are very complex and highly non-linear characteristics. To achieve the satisfactory weld bead geometry in robot welding system, the seam tracking algorithm should be reliable. The number of seam tracker was developed for arc welding automation by now. Among these seam tracker, the arc sensor is prevalently used in industrial robot welding system because of its low cost and flexibility. However, the accuracy of arc sensor would be decreased due to the electrical noise and metal transfer. In this study, the signal processing algorithm based on the neural network was implemented to enhance the reliability of measured welding current signals. Moreover, the seam tracking algorithm in conjunction with the signal processing algorithm was implemented to trace the center of weld line. It was revealed that the neural network could be effectively used to predict the welding current signal at the end of weaving.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2000년도 추계학술대회 논문집 학회본부 D
• /
• pp.769-771
• /
• 2000

In this paper, pattern recognition scheme for EMG signal using artificial neural network is proposed. For manipulating ability, the movements of human arm are classified into several categories EMG signals of appropriate muscles are collected during arm movement. Patterns of EMG signals of each movement are recognized as follows: 1) The features of each EMG signal are extracted. 2) With these features, the neural network is trained by using feedforward error back-propagation (FFEBP) algorithm. The results show that the arm movements can be classified with EMG signals at high accuracy.

• 전자공학회논문지A
• /
• 제33A권7호
• /
• pp.44-54
• /
• 1996

In DS/SS systems it is necessary to synchronize the locally generated despreading signal with the received spreading signal to demodulate the received signal. The synch process between the two signals is usually accomplished in two steps : first acquisition then tracking. In this paper, an acquisition system aided by the neural network is proposed for the rapid and exact acquisition in DS/SS. the neural netowrk is composed o fthree-layered perpecptrons and trained by the backpropagation algorithm. The performance of the proposed system is analyzed and compared with ones of conventional systems using the sequential estimation technique under an additive while gaussian noisy channel. In all of th econsidered simulations, the proposed system outperforms conventional systems.

• 제어로봇시스템학회논문지
• /
• 제4권6호
• /
• pp.765-771
• /
• 1998

An artificial neural network(NN) technique is developed for hardware redundant sensor validation. Since the measurement space is a continuous space with many operating regions, it is difficult to train a NN to correctly detect failure in an accurate measurement system. A conventional backpropagation NN is modified to include an additional preprocessing layer that extracts classification features from scalar measurements. This feature extraction means transform the measurement space to parity space. The NN is independent of the state variable being measured, the instrument range, and the signal tolerance. This NN resembles the parity space approach to signal validation, except that analytical parity equations are unneeded and the NN pattern recognition capability is utilized for decision making.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1998년도 하계학술대회 논문집 B
• /
• pp.572-574
• /
• 1998

This paper describes design method of control system with a pre-compensator using a neural network to compensate a error signal between a reference' signal and system response. The neural network which is used here is the mixed structure and it's algorithm is a back propagation that modify coupling coefficients. Applying this method to the position control system using DC servo motor as a driver, we verify the usefulness of this method with simulation.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1994년도 추계학술대회 논문집 학회본부
• /
• pp.397-399
• /
• 1994

In this paper a feature-based automatic neural network signal classifier is presented, where five neural network algorithms such as MLP, RBF, LVQ2, MLP-Tree and LVQ-Tree are combined in parallel to classifiy various signals from their features, based on the majority vote method. To demonstrate the performance and applicability of the proposed signal classifier, some test results for the classification of synthetic waveforms and power disturbances are provided.

• 한국정밀공학회지
• /
• 제12권10호
• /
• pp.149-156
• /
• 1995

This paper presents a method of detecting chatter vibration in end-mill process. The detecting system consists of an adaptive signal processing scheme which uses an autore- gressive time-series model and a neural network is proposed and is verified its effectiveness by using acceleration and cutting force signals recorded during slotting in end-mill operations. Expeerimental results indicate that the proposed system provides excellent detection when chatter is occured within the ranges of cutting conditions considered in this study and an effectiveness of the integration of signals is confirmed.