• Proceedings of the KIEE Conference
• /
• 2011.07a
• /
• pp.1958-1959
• /
• 2011

RBFNNs(Radial Basis Function Neural Networks) 모델의 경우 Min-Max, HCM(Hard C-means)클러스터링 그리고 FCM(Fuzzy C-means)클러스터링 중 한가지를 통해 데이터 입자는 로드 규칙을 생성한 후 퍼지 공간을 분할 및 가우시안 함수의 정점을 정의한다. 본 논문은 기존의 방법과는 다르게 Min-Max와 FCM클러스터링을 혼합하여 로드의 규칙을 생성한 후 퍼지 공간을 분할 및 가우시안 함수의 정정을 정의하는 방법으로 사용하고자 한다. PSO최적화 알고리즘을 이용하여 같은조건에서 최적화한 기존의 방법으로 모델링된 RBFNNs와 Min-Max와 FCM 클러스터링을 혼합하여 사용한 방법의 비교를 통하여 어떤 모델의 성능이 더욱 좋은지 비교하고자 한다.

• Proceedings of the KIEE Conference
• /
• 2008.10b
• /
• pp.77-78
• /
• 2008

This paper presents to control speed of induction motors with uncertainties. We use an adaptive backstepping controller with fuzzy neural networks(FNNs) and model reference adaptive system(MRAS) at Indirect vector control method. The adaptive backstepping controller using FNNs can control speed of induction motors even we have a minimum of information. And this controller can be used to approximate most of uncertainties which are derived from unknown motor parameters, load torque such as disturbances. MRAS estimates to rotor resistance and also can find optimal flux to minimize power losses of Induction motor. Indirect vector PI current controller is used to keep rotor flux constant without measuring or estimating the rotor flux. Simulation and experiment results are verified the effectiveness of this proposed approach.

• Proceedings of the KIEE Conference
• /
• 2009.07a
• /
• pp.1903_1904
• /
• 2009

본 논문은 퍼지 RBF 뉴럴네트워크를 이용한 트랙킹 검출 방법을 제시한다. IEC 60112에서 규정한 실험 장치와 방법에 따라 실험을 수행하였다. NI 장비를 사용하여 전류 파형을 측정하고, 측정된 전류 파형으로부터 FFT, 웨이블렛등의 신호처리 기법을 사용하여 12개의 특징점을 추출한다. 추출된 특징점들을 퍼지 RBF 뉴럴네트워크의 입력으로 사용하여 트랙킹 발생 유무를 검출한다. 퍼지 RBF 뉴럴네트워크는 WLSE를 사용하여 학습하고, HFC-PGA를 이용하여 특징점들의 선택, 퍼지 규칙의 수, 후반부 다항식 차수, 퍼지화 계수등을 최적화 하였다.

• Journal of Electrical Engineering and Technology
• /
• v.10 no.4
• /
• pp.1910-1914
• /
• 2015

Human posture recognition is an attractive and challenging topic in computer vision due to its promising applications in the areas of personal health care, environmental awareness, human-computer-interaction and surveillance systems. Human posture recognition in video sequences consists of two stages: the first stage is training and evaluation and the second is deployment. In the first stage, the system is trained and evaluated using datasets of human postures to ‘teach’ the system to classify human postures for any future inputs. When the training and evaluation process is deemed satisfactory as measured by recognition rates, the trained system is then deployed to recognize human postures in any input video sequence. Different classifiers were used in the training such as Multilayer Perceptron Feedforward Neural networks, Self-Organizing Maps, Fuzzy C Means and K Means. Results show that supervised learning classifiers tend to perform better than unsupervised classifiers for the case of human posture recognition.

• Journal of Distribution Science
• /
• v.11 no.6
• /
• pp.13-19
• /
• 2013

Purpose - This study aims to predict the audit reports of listed companies on the Tehran Stock Exchange by using meta-heuristic algorithms. Research design, data, methodology - This applied research aims to predict auditors reports' using meta-heuristic methods (i.e., neural networks, the ANFIS, and a genetic algorithm). The sample includes all firms listed on the Tehran Stock Exchange. The research covers the seven years between 2005 and 2011. Results - The results show that the ANFIS model using fuzzy clustering and a least-squares back propagation algorithm has the best performance among the tested models, with an error rate of 4% for incorrect predictions and 96% for correct predictions. Conclusion - A decision tree was used with ten independent variables and one dependent variable the less important variables were removed, leaving only those variables with the greatest effect on auditor opinion (i.e., net-profit-to-sales ratio, current ratio, quick ratio, inventory turnover, collection period, and debt coverage ratio).

• Journal of Ocean Engineering and Technology
• /
• v.12 no.3 s.29
• /
• pp.103-111
• /
• 1998

This paper describes the initial hull form design process which generate a hull form using a neurofuzzy modeling. Neurofuzzy system is to combine the merits of fuzzy inference system and neural networks. Therefore it has structured knowledge representations as well as adaptive capacities. Initial hull form design stage is the process which generate an adoptable hull form from the limited design information and multi-decidions condidering correlations with design factors. It can be assidted efficiently by neurofuzzy system. This paper suggests two methods of an initial hull form generation using the neurofuzzy modeling and B-spline theory. and examines the usefulness of suggested method through its application examples.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.04a
• /
• pp.303-306
• /
• 2004

기존의 퍼지 규칙에 기반을 둔 퍼지 다항식 뉴론(FPN)들로 구성된 SOFPNN은 데이터 수가 적고 비선형 요소가 많은 시스템에 대한 체계적이고 효율적인 최적 모델 을 구축할 수 있었으며 각 층 노드의 선택 입력을 변화시킴으로써 네트워크 구조 전체의 적응능력을 향상 시켰다. 유전자 알고리즘을 이용하여 자기구성 퍼지 다항식 뉴럴 네트워크의 입력변수의 수와 이에 해당되는 입력변수 그리고 규칙 후반부 다항식의 차수를 탐색하여 최적 의 자기구성 퍼지 다항식 뉴럴 네트워크를 구축한다. 그러나, SOFPNN의 기본 뉴론인 퍼지 규칙 기반 다항식 뉴론의 경우 입력변수가 많아질수록 규칙수가 기하급수적으로 증가한다는 단점을 가지고 있으나 본 노문에서 제안한 퍼지 집합 기반 다항식 뉴론(FSPN)의 규칙수는 입력 변수들이 서로 독립적이므로 규칙의 증가가 퍼지 규칙 기반 다항식 뉴런보다는 적다는 장점을 가지고 있다. 이러한 특성을 기반으로 기존의 SOFPNN의 노드에 퍼지 규칙 기반 다항식 뉴런 대신에 퍼지 집합 기반 다항식 뉴런을 적용한 SOFPNN을 제안하여 기존의 SOFPNN과 성능을 비교하였다. 최적의 자기 구성 퍼지 집합기반 다항식 뉴럴 네트워크를 구축하기 위하여 SOFPNN에서처럼 유전자 알고리즘을 이용하여 네트워크의 입력변수의 수와 이에 해당되는 입력변수 그리고 규칙 후반부 다항식의 차수를 탐색하였다.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.84.5-84
• /
• 2001

The objective of this paper is to develop a powered wheelchair controller based on EMG for users with high-level spinal cord injury. EMG is very naturally measured when the user Indicating a certain direction, and the force information which will be used for the speed of wheelchair is easily extracted from EMG. Furthermore, the emergency situation based on EMG will be checked relatively ease. We classified the pre-defined motions such as rest case, forward movement, left movement, and right movement by Fuzzy Min-Max Neural Networks (FMMNN). This classification results shows the feasibility of EMG as an input interface for powered wheelchair. To make the system low cost and small size, we developed EMG AMP and its controller ...

• Journal of Sensor Science and Technology
• /
• v.20 no.3
• /
• pp.151-155
• /
• 2011

A variety of pattern recognition algorithms including neural networks may be applicable to the identification of odors. In this paper, an identification technique for an electronic odor sensing system applicable to wound state monitoring is presented. The performance of the radial basis function(RBF) network is highly dependent on the choice of centers and widths in basis function. For the fine tuning of centers and widths, those parameters are initialized by an ill-conditioned genetic fuzzy c-means algorithm, and the distribution of input patterns in the very first stage, the stochastic gradient(SG), is adapted. The adaptive RBF network with singular value decomposition(SVD), which provides additional adaptation capabilities to the RBF network, is used to process data from array-based gas sensors for early detection of wound infection in burn patients. The primary results indicate that infected patients can be distinguished from uninfected patients.

• Proceedings of the KIEE Conference
• /
• 2008.07a
• /
• pp.1505-1506
• /
• 2008

In this paper, a robust adaptive backstepping control is developed for efficiency optimization of induction motors with uncertainties. The proposed control scheme consists of efficiency flux control(EFC) using a sliding mode adaptive flux observer and robust speed control(RSC) using a function approximation for mechanical uncertainties. In EFC, it is important to find the flux reference to minimize power losses of induction motors. Therefore, we proposed the optimal flux reference using the electrical power loss function. The sliding mode flux observer is designed to estimate rotor fluxes and variation of inverse rotor time constant. In RSC, the unknown function approximation technique employs nonlinear disturbance observer(NDO) using fuzzy neural networks(FNNs). The proposed controller guarantees both speed tracking and flux tracking. Simulation results are presented to illustrate the effectiveness of the approaches proposed.