Reviewing and designing pre-processing units for RBF networks: initial structure identification and coarse-tuning of free parameters

Kayhan, Gökhan; Özdemir, Ali Ekber; Eminoğlu, İlyas

doi:10.1007/s00521-012-1053-8

Cited by 12 publications

(6 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Furthermore, the center and spread width for each hidden node is determined by using k-means clustering algorithm [55,56]. The RBFNN is a fully connected three-layer FFNN and establishes nonlinearity in the hidden layer neurons.…”

Section: B Radial Basis Function Neural Network (Rbfnn)mentioning

confidence: 99%

See 1 more Smart Citation

Islanding detection approach with negligible non-detection zone based on feature extraction discrete wavelet transform and artificial neural network

Hashemi

Mohammadi

2016

Int. Trans. Electr. Energ. Syst.

View full text Add to dashboard Cite

Summary The paper presents a novel detection method based on feature extraction discrete wavelet transform (DWT) combined with artificial neural network (ANN) for identification of islanding condition in distributed generation (DG) system. Islanding detection methods can be classified into two major categories as active and passive methods. The main disadvantages of the passive methods are determined threshold value and related to their large non‐detection zone. The emphasis of the proposed approach is eliminating the aforementioned drawbacks. The DWT allows revealing various hidden features of the signal. The aim of this paper is to determine the best wavelet basis function in order to identify islanding occurrence with higher accuracy and lower decomposition level. The proposed approach requires the measure rate of change of frequency at the DG's terminal, and various features are extracted from DWT then these features used as input into ANN. Also, the performance of the different structures of ANN such as feed‐forward neural network, radial basis function, and probabilistic neural network are compared for islanding detection purpose. The proposed method is simulated and tested in various operation conditions such as islanding conditions, motor starting, capacitor bank switching, and nonlinear load switching. The test results showed that the proposed method correctly detects the islanding operation and does not mal‐operate in the other situations. Copyright © 2016 John Wiley & Sons, Ltd.

show abstract

Section: B Radial Basis Function Neural Network (Rbfnn)mentioning

confidence: 99%

“…In conventional RBFNN, training approach and the number of hidden nodes are usually defined according to the statistical properties of the input data. Furthermore, the center and spread width for each hidden node is determined by using k-means clustering algorithm [55,56].…”

Section: B Radial Basis Function Neural Network (Rbfnn)mentioning

confidence: 99%

Islanding detection approach with negligible non-detection zone based on feature extraction discrete wavelet transform and artificial neural network

Hashemi

Mohammadi

2016

Int. Trans. Electr. Energ. Syst.

View full text Add to dashboard Cite

show abstract

“…Bulduğu küme sayısı en iyi olmamasına rağmen kullanıcıya bir başlangıç değeri sunar. Bu çalışmada [9]' da sunulan melez yapı denenmiştir. Kohanen ağı ile BCO (Bulanık C Ortalamalar) algoritması melez kullanılmıştır.…”

Section: Melez Yapı İle Müap Kümelemeunclassified

Classification of ALS disease using support vector machines

Kucuk

Eminoğlu

2015

2015 23nd Signal Processing and Communications Applications Conference (SIU)

View full text Add to dashboard Cite

Özetçe-Bu çalışmada motor nöron hastalığının en yaygın çeşidi olan ALS hastalığının teşhisi için DVM (Destek vektör makinesi) algoritması kullanılmıştır. EMG (Elektromiyogram) işaretleri, sınıflandırılmadan önce ön işleme, bölütleme, kümeleme ve öznitelik çıkarma aşamalarından geçirilmiştir. Kümeleme aşamasında hiyerarşik ve melez kümeleme yöntemleri kullanılmıştır. Sonrasında zaman, frekans uzayındaki öznitelik vektörleri ve bunların farklı birleşimleri ile elde edilen çoklu öznitelik vektörleri olmak üzere toplam 11 tane öznitelik vektörü denenmiş ve DVM üzerindeki etkileri incelenmiştir. Sonuç olarak; iki farklı kümeleme yönteminin üstünlüklerinin kullanılan öznitelik vektörlerine bağlı olduğu; çoklu öznitelik vektörlerinin ALS hastalığı teşhisinde yüksek başarım sağladığı ve daha az tutarsızlık sergilediği görülmüştür. Abstract-In this study, SVM (Support Vector Machine) algorithm is used for the diagnosis of ALS which is the most common type of motor neuron disease. Before classification of EMG data with SVM (Support Vector Machine); pre-processing, segmentation, feature extraction and clustering stages of data are completed. In the stage of clustering, hybrid and hierarchical clustering methods are employed. After that, feature vectors in time and frequency domains and their different combinations (a total of 11 feature vectors) are fed to the SVM and the obtained results are observed. It is understood that the advantages of clustering methods dependent on the feature vectors; multiple feature vectors provide high performance in the diagnosis of ALS disease and exhibit much lower discrepancy.

show abstract

“…In the recent literature, many studies have dealt with multi-stage learning algorithms [6,10,12]; clustering algorithms for centroid relocation [4, 6-8, 10, 12, 13]; and optimal width adjustment for the basis functions [2,3,6,11,12,14,17]. To briefly describe a few, a three-stage algorithm, comprising k-means, k-nn and Optimal Steepest Descent (OSD) methods, is proposed in [12].…”

Section: Introductionmentioning

confidence: 99%

System Identification through RBF Neural Networks: Improving Accuracy by a Numerical Approximation Method for the Centroids and Widths Adjustment

Oliveira¹,

Braga²,

Reis³

et al. 2018

Proceedings XIII Brazilian Congress on Computational Inteligence

View full text Add to dashboard Cite

Within the last two decades there has been an increasing need for the development of mathematical models out of observed data captured from a system, a process called empirical modelling or systems identification. Under this circumstance, many techniques and methodologies have been proposed, among them the use of Artificial Neural Networks. It is proposed herein a non-hybrid gradient-based learning algorithm for a Radial Basis Function Neural Network aimed at improving the accuracy of non-linear dynamical system modelling. A single-stage non-hybrid approach is employed for the learning process, where the free parameters of the network -the centroids positioning, the receptive fields width, and the weights -are updated through a supervised method. Accurate identification capability is examined by the use of two non-linear datasets and the performance of the proposed method is compared with traditional techniques. Results demonstrate that nonlinear system identification can be significantly improved with easy-toimplement gradient-based RBF learning strategy.

show abstract

Reviewing and designing pre-processing units for RBF networks: initial structure identification and coarse-tuning of free parameters

Cited by 12 publications

References 35 publications

Islanding detection approach with negligible non-detection zone based on feature extraction discrete wavelet transform and artificial neural network

Islanding detection approach with negligible non-detection zone based on feature extraction discrete wavelet transform and artificial neural network

Classification of ALS disease using support vector machines

System Identification through RBF Neural Networks: Improving Accuracy by a Numerical Approximation Method for the Centroids and Widths Adjustment

Contact Info

Product

Resources

About