CONTENTS & ABSTRACTS
In English. Summaries in Estonia
Proceedings of the Estonian Academy of Sciences.
Volume 49 No. 2 June 2000
Interpretability versus adaptability in fuzzy systems; 76–95
Andri RIID and Ennu RÜSTERN
Abstract. This paper considers the most common adaptation techniques for standard and Sugeno fuzzy systems with special regard to post-adaptation linguistic interpretability. The level of adaptation is characterized by the approximation error, but there is no similarly accepted measure for the validity of linguistic interpretation (transparency) and the detection of the latter relies much on empirical observation. Such observation is hardly possible beyond three-dimensional space. The general purpose of this paper is to find out how adaptation algorithms for antecedent and/or consequent parameters, such as gradient descent method, least squares estimation, and clustering techniques, act on the interpretability of the system besides their approximation properties. The comparison of the algorithms is based on the modelling of a simple single-input single-output system. The conclusion that the transparency of the observed system depends on the degree of overlap of neighbouring input fuzzy sets can, however, be generalized for multivariate systems.
Key words: modelling, fuzzy systems, Sugeno systems, fuzzy clustering, gradient descent, least squares estimation, accuracy, interpretability, adaptability.
Supervisory fed-batch fermentation control on the basis of linguistically interpretable fuzzy models; 96–112
Andri RIID and Ennu RÜSTERN
Abstract. Design of a controller for fed-batch fermentation, based on the linguistically interpretable fuzzy models of the process, is investigated. The paper is based on the data provided by the Modelling and Control Competition organized by Industrial Control Centre of the University of Westminster. Fuzzy model of the process is identified from measured data using two different methods: the Yager–Filev fuzzy template modelling and least square estimation with Gustafson–Kessel clustering. The core of the control system is a fuzzy supervisor that is extracted from the identified model. A number of test experiments confirm the applicability of the approach.
Key words: fuzzy modelling, fuzzy control, interpretable fuzzy systems, supervisory control, fed-batch fermentation.
Piecewise linearly approximated sine wave for dynamic quality tests of A/D converters; 113–119
Abstract. An overview of using the piecewise linearly approximated sine wave as a test signal for dynamic quality tests of A/D converters is presented. The basic principles and generation methods of a specific test signal for A/D converters are explained. A comparison of the classical sine wave, proposed test signal, and triangular wave is presented. Suitability of the piecewise linearly approximated sine wave for dynamic quality tests of A/D converters in time and frequency domain and for statistical methods is considered.
Deviation extension problems and nonlinear effects in the phase locked loop frequency demodulators; 120–134
Mart MIN, Vello MÄNNAMA, and Toivo PAAVLE
Abstract. In this paper, possibilities for deviation extension of the phase locked loop (PLL) frequency demodulators are considered and some specific features of the PLL systems are analysed. First, optimization of the 3rd order PLL is used in order to reduce substantially the dynamic phase error; that enables to extend the deviation range. Both theoretical analysis and computer simulation are used to determine the reliable operation limits of the frequency deviation. Second, implementation of the first order PLL as a wide-band frequency demodulator is analysed, assuming an effective out-loop post-processing (filtering) of the output signal. It is shown that the deviation range may be extended almost to the maximum theoretical value, determined by the sampling theorem. Finally, two special periodic signal waveforms, generated in the first order PLL due to the in-loop noise (ripple), are demonstrated and analysed.
Key words: phase locked loop, frequency demodulator, frequency deviation, dynamic phase error, filter optimization, computer simulation, signal distortion.
Biasing error correction in averaging current-mode rectifiers; 135–143
Vello MÄNNAMA and Brett WILSON
Abstract. A current-mode approach to the design of precision rectifiers and absolute value circuits has been shown to extend high-speed signal capability and reduce distortion around the zero-crossing region. However, addition of bias to the bridge diodes to improve high-speed performance restricts linearity in the transition region. This paper develops an error analysis for current-mode rectifiers over various input signal waveforms and presents a bias error cancellation technique whereby the offset effects of bridge bias can be eliminated.
Key words: current-mode design, pre-biased rectifier, error analysis, error cancellation.
Modelling of charge carrier non-isothermal transport in silicon and silicon carbide; 144–154
Enn VELMRE and Andres UDAL
Abstract. Extending our previously published model of charge carrier transport and mobility, a procedure for calculation of the semiconductor thermoelectric coefficients is developed. It is shown that the phonon drag strongly influences the Seebeck and Peltier coefficients even at high temperatures, up to 400 K for Si and 600 K for SiC. As a new feature, the presented model gives a possibility to investigate the influence of the electron-hole scattering on thermoelectric coefficients. A strong underestimation of non-isothermal ÑT-proportional current density terms in most of the semiconductor device simulators is pointed out.
Key words: Seebeck coefficient, Peltier coefficient, Soret coefficient, phonon drag, electron-hole scattering.
Large area 6H–SiC Schottky diode; 155–159
Toomas RANG, Oleg KOROLKOV, and Mihhail PIKKOV
Abstract. This paper describes experimental investigation of a large area 6H–SiC Schottky diode fabricated with diffusion welding technology. A new model for the dependence of the barrier height on the temperature is presented.
Key words: SiC, Schottky structures, barrier height, temperature influence.
Transconductance feedback amplifiers; 160–166
Abstract. Gain-bandwidth independence over a wide range of voltage gain levels is available by employing feedback around a transconductance amplifying element in place of a voltage amplifying element. It is shown that only minor changes are required to the architecture of standard 3-stage voltage operational amplifier circuits to obtain all the advantages of this new configuration, which also include improved voltage slew rates, whilst retaining two high-impedance input terminals.
Key words: transconductance amplifier, feedback, gain-bandwidth product, slew rate, operational amplifier architecture.
Boris Tamm 70; 167