Correction of nominal point of circuit's parameters based on artificial dynamic (optimizing) neural network

• Autorzy: Józefowicz K. , Rybarczyk A.
• Konferencja: Computer Applications in Electrical Engineering 2009 [Poznan, 20-22 April, 2009]
• Języki publikacji: EN

Abstrakty

EN

In this paper the authors propose using dynamic (optimizing) artificial neural networks as well as interactive environment MATLAB to correct nominal parameters of active filters of higher ranks with a simultaneous process of optimization of a circuit designed [8, 9]. The MATLAB's [3, 15] capabilities of fast constructing and testing algorithms of artificial neural networks as much as a verification of statistical properties of the project acquired were employed. In the paper it was shown how the Hopfield dynamic artificial neural networks aid in designing electronic circuits with a correction of the nominal point of the parameters. The analyses aimed at building neural optimizers in order to acquire such values of constructional parameters of the circuit examined which minimize the variance of selected output values, which means minimizing the dependences of the project parameters on the scatter of the values of technological parameters [10-18]. The most effective solutions of neural optimizers for a dedicated project task were presented. The authors also indicated advantages resulting from an application of optimizing networks in comparison with other methods of optimization.

Słowa kluczowe

EN

neural network; MATLAB; electronic circuit; artificial neural network

PL

sieć neuronowa; obwód elektroniczny; sztuczna sieć neuronowa

• Wydawca: Wydawnictwo Politechniki Poznańskiej
• Czasopismo: Poznan University of Technology Academic Journals. Electrical Engineering
• Rocznik: 2010
• Tom: No. 61
• Strony: 75--87
• Opis fizyczny: Bibliogr. 23 poz.

Twórcy

autor

Józefowicz K.

autor

Rybarczyk A.

• State School of Higher Vocational Education in Leszno

Bibliografia

• [1] Brzózka J., Dorobczyński L.: Programming in Matlab, Wydawnictwo MIKOM, Warszawa 1998.
• [2] Duch W., Korbicz J., Rutkowski L., Tadeusiewicz R.: Monograph: Biocybernetics and biomedical engineering 2000, Volume 6, Neural Networks. AOW Exit, Warszawa 2000.
• [3] Demuth H., Beale M.: Neural Network Toolbox User’s Guide, Version 4, The MathWorks, 2004.
• [4] Hertz J., Krogh A., Palmer R. G. I.: Introduction to the theory of neural computing, Wyd. II, WNT, Warszawa, 1995.
• [5] Mrozek B., Mrozek Z.: Matlab and Simulink. User's guide, second edition, Wyd. Helion, Gliwice 2004.
• [6] Osowski St.: Neural networks from an algorithmical perspective. Ofic. Wyd. 11, WNT, Warszawa, 1996.
• [7] Osowski St.: Neural networks for data processing. Ofic. Wyd. PW, Warszawa 2000.
• [8] Rybarczyk A., Szulc M.: Applying computing package MATLAB for minimization of the variances of output parameters of a self - contained circuit. CMOS, VII Konf. ZkwE’02, Poznań-Kiekrz, 22-24 kwietnia 2002, s. 357-360.
• [9] Rybarczyk A.: Yield Optimization of Analog Filters Using a Statistical Design Approach and Multiparameter Sensitivity Measure, 12th European Conf. on Circ. Theory and Design (ECCTD’95), Istanbul (Turkey), 27-31.08 1995, tom I s. 287-290.
• [10] Rybarczyk A, Józefowicz K., Rybarczyk A.: Designing optimal active filters of higher ranks using artificial neural networks XXVI Conf. IC-SPETO’03, Gliwice-Niedzica, 28-31 maj 2003r., s. 481-484.
• [11] Józefowicz K., Rybarczyk A.: An intelligent system of optimal designing based on recurrent neural networks of the Hopfield type. IX Conf. ZkwE’04, Poznań/Kiekrz, 19-21.04.2004, s. 397-402.
• [12] Józefowicz K., Rybarczyk A.: Procedures Applied in the Intelligent System of Optimal Design Based on Feed-Forward Neural Networks, Mat. Konf. IC-SPETO’2004 (XXVII), Gliwice-Niedzica, 26-29.05.2004, s. 469-474.
• [13] Józefowicz K., Rybarczyk A.: An intelligent system of designing selected electronic circuits based on unidirectional neural networks with radial, basic functions. RBF, X Konf. ZkwE’05, Poznań, 18-19.04.2005, Nr R13, s. 25-26.
• [14] Józefowicz K., Rybarczyk A.: Systems of designing selected electronic circuits using recurrent SSN based on a perceptron (Elman network), XXVIII Konf. IC-SPETO, Gliwice/Ustroń, 11-14.05.2005, vol. 2, s. 333-339.
• [15] Józefowicz K., Rybarczyk A.: A symulator of artificial neural networks in interactible MATLAB environment in a choise of tolerances of selected electronic circuits. XI Scienticic-Technical Conference: ZKwE2006, Poznań, 10-12.04.2006, Nr R25, str. 49-52.
• [16] Józefowicz K., Rybarczyk A.: An automatic generation of artificial neural networks in a choise of tolerances of selected electronic circuits. XXIX International Conference IC-SPETO, Gliwice/Ustroń, 24-27.05.2006, vol. 2, str. 337-341.
• [17]Józefowicz K., Rybarczyk A.: A comparative analysis of selected unidirectional and recurrent networks during determining nominal parameters of electronic circuits and their tolerances. XII Scientific-Technical Conference: ZKwE2007, Poznań, 16-18.04.2007, Nr R 12, s. 27-32.
• [18] Józefowicz K., Rybarczyk A.: The evaluation of design method utilizing artificial neural networks in the task of determining values and tolerances of electronic circuits' nominal parameters. XXX Międzynarodowa Konferencja IC-SPETO, Gliwice/Ustroń, 24-27.05.2007, s. 221.
• [19] Tadeusiewicz R., Lula P.: Neural networks, Training mterials StatSoft, Kraków 2004.
• [20] Opalski L.: Methods and algorithms of optimization of the quality of electronic circuits. OWPW, Warszawa 2002.
• [21] Stybliński M.: Methods of analyzing and optimizing tolerance of electronic circuits' parameters. WNT, Warszawa 1981.
• [22] Kjellstrom G., Taxen K.: Optimization methods for statistical network design. IEEE Proc. ISCAS-76, Boston 1975.
• [23] Tahim K. S., Spencer R.: An integrated approach to manufacturing yield estimation and design centering. Proc. 1978 European on CT and Design, Lausanne, Switzerland. September 1978.