April 2004, Volume 4, Number 1
SZABAT K. — ORLOWSKAKOWALSKA, T. Sensitivity Analysis of State Variable Estimators for TwoMass Drive System 
ŽILKOVÁ, J. — TIMKO, J. Modelling of a Linear Induction Motor in Dynamical Performance 
KOLLÁR, J. — NOVITZKÁ, V. Static Weaving at Dynamic Join Points 
MARTÍNEK, Z. — TŮMA, I. Fault State Selection in the Power System 
VOJTKO, J. — MOJŽIŠ, M. Torque Sensor Modeled by Neural Network 
BOCZAR, T. — FRĄCZ, P. Acoustic Emission of SingleSource Partial Discharges of Various Types Generated Simultaneously 
KOWALSKI, C. — PAWLAK, M. Application of Artificial Intelligence Methods for Rotor Faults Detection of the Induction Motor 
MEŠTER, M. Alternative Transient Program  The Powerful Tool for Power Engineering Simulations 
MIHALÍK, J. — ŠTEFANIŠIN, R. Quantization Algorithms of Standard Videocodecs 
ŠIMOŇÁK, S. — HUDÁK, Š. Petri Net Semantics for ACP Terms 
BOKR, J. Canonical Decomposition 
GROFČÍK, M. Parallel Dataflow Computing Environment 
Summary: 
Krzysztof SZABAT  Teresa ORLOWSKAKOWALSKA Sensitivity Analysis of State Variable Estimators for TwoMass Drive System [full paper] The paper deals with the comparison of two estimation methods for torsional
torque, load speed and disturbance torque reconstruction based on the Luenberger observer and Kalman filter for twomass drive
system. Estimation errors caused by the drive parameter changes were checked and
compared. Due to a lack of analytical methods for gain matrices design for such
estimators, parameters of both estimators were optimised using the same geneticgradient algorithm and the same optimisation index. Such procedure has enabled the detailed comparison of results obtained for these two state
estimators, based on simulation and experimental tests.

Jaroslava ŽILKOVÁ  Jaroslav TIMKO Modelling of a Linear Induction Motor in Dynamical Performance [full paper] This paper deals with modelling of the linear induction motor (LIM). Linear induction motors fall into the category of special electrical machines in which the electric energy is converted directly into mechanical energy of translatory motion. LIMs are the most popular ones from the group of linear electric motors, which consists also of DC motors, synchronous motors, reluctance motors, stepping motors, oscillating and hybrid motors.

Ján KOLLÁR  Valerie NOVITZKÁ Static Weaving at Dynamic Join Points [full paper] Aspect oriented programming is a programming methodology based on separating concerns (aspects) of computation, describing them separately in the form of advices that are applied in clearly selected set of points of a program using pointcut designators. In this paper we will present an approach to the implementation of crosscutting concept of aspect oriented programming using environmental basis of PFL  a process functional programming language. We will concentrate on the advices, defined by pointcut designators with temporal logic operations. We also introduce possible directions in further conceptual solutions based on process functional
paradigm.

Zbyněk MARTÍNEK  Ivan TŮMA Fault State Selection in the Power System [full paper] This paper deals with the selection of fault states that are further assessed in order to work out operability of a power system. The selection of fault states is necessary for large power systems, where assessment of all possible fault states in terms of system operability would be extremely timeconsuming because the number of all fault states can be in the case of complex networks very high. Therefore it is needed to reasonably reduce the number of fault states that are to be assessed. Three different approaches are described in this article. Their goal is to find a set of those faults states that are probable enough in real operation conditions. Only such fault states are then assessed in terms of power reliability and power system operability.

Jozef VOJTKO  Miroslav MOJŽIŠ Torque Sensor Modeled by Neural Network [full paper] This paper describes principle and design of elastomagnetic torque sensor and with it related neural network model. Multilayer feedforward neural network was used. The parameters of this neural network were changed in wide range. The most suitable solution is presented. Other neural network solutions have not been presented in this paper. The basic measuring apparatus was set and output characteristics at various input supply current were analyzed. The substitute circuit model of the sensor was determinated from measured data set. The substitute functions for output characteristics were stated by linear regression. The statistical model was compared with neural network model and errors were computed.

Tomasz BOCZAR  Paweł FRĄCZ Acoustic Emission of SingleSource Partial Discharges of Various Types Generated Simultaneously [full paper] The subject matter of this paper is connected with determining identification possibilities of partial discharge (PD) forms based on the analysis of frequency runs of the acoustic emission (AE) pulses generated. In the research carried out so far, the descriptors characterizing the AE pulses measured in the frequency domain were determined, and then their usefulness for the identification of basic PD forms was defined. The results presented here refer to the case when two various singlesource PDs occur at the same time in oil insulation. This paper will present the results of frequency analysis of AE pulses generated separately in the setups modeling surface and singlesource discharges of the multipoint  grounded plane type. Then, for comparison, the results obtained from simultaneous generation of the discharges of both types will be presented. In the Conclusion the results of the comparative analysis of the results obtained will be presented from the angle of determining identification possibilities in the case of generation of singlesource PDs of various types.

Czesław T. KOWALSKI  Marcin PAWLAK Application of Artificial Intelligence Methods for Rotor Faults Detection of the Induction Motor [full paper] The paper deals with rotor fault diagnosis problems of the induction motor using different Artificial Intelligence methods like: neural networks, fuzzylogic and combined neurofuzzy methods. Rotor fault detectors based on these methods were developed. All detectors were trained and tested using measurement data of stator current spectra. The digital signal processor was used for practical realisation of AIbased fault detectors. All detectors were tested in offline as well as online operation. The efficiency of developed neural detectors was
evaluated.

Marián MEŠTER Alternative Transient Program  The Powerful Tool for Power Engineering Simulations [full paper] In this article is presented the capabilities and practical experience with Alternative Transient Program (ATP). ATP is an universal program system for simulation of transient phenomena of electromagnetic as well as electromechanical nature.

Ján MIHALÍK  Radoslav ŠTEFANIŠIN Quantization Algorithms of Standard Videocodecs [full paper] The paper gives a review of algorithms of quantization of the standard videocodecs. On the basis of the results of statistical analysis of videosignals in DCT domain different modifications of linear and nonlinear algorithms of quantization are presented in intraframe and interframe modes of the videocodecs. Finally an evaluation and comparison of their performances by archieved experimental results have been made.

Slavomír ŠIMOŇÁK  Štefan HUDÁK Petri Net Semantics for ACP Terms> [full paper] In computer science community exists an assertion which states, that there will not exist a single formal method for all aspects of the systems to be expressible in satisfying way. We are looking for the solution to this problem in combination of existing methods which offer complementary properties for description of discrete system. In our case the chosen methods are Petri nets and Process algebra. In case of Petri nets, very useful properties are those of automatic generation of system invariants, algorithm for solving the reachability problem and graphic representation of system. Process algebra on the other hand offers easy way for system decomposition. There is a number of reasons why a formal specification is useful, and it grows more important, as the designated system becomes larger and more complex. The advantage of formal methods usage is also the ability of validation of requirements understanding and detection of errors at early stages of system development. For large and complex systems, testing cannot prove the right functionality of the system, because it is incomplete and thus does not check the behavior of the system in all reachable states.

Josef BOKR Canonical Decomposition [full paper] The paper deals with all conceptions, i.e. conceptions given as transition or excitation functions vs. as the factual decomposition with predefined structure, of canonical decomposition, memory modules and their choise and also the part of state at logical object.

Martin GROFČÍK Parallel Dataflow Computing Environment [full paper] The article deals with design and implementation of the parallel computational environment, which is driven by flow of the data. Design is based on the abstract model of the dataflow computer. This computer is developed at the Dapartment of computers and informatics, Faculty of Electrical engeneering and Informatics, Technical University of Košice (DF KPI)[1],[2]. Parallel computational environment presents design of the computational environment, which devlopment is based on the results of the simulation [3]. It modifies abstract design DF KPI to achieve unified workload on the components of the computer and to minimize count of the parallel access to components of the DF KPI. At the end of the document is presented implementation of the proposed design at the computer net environment, which is most achievable environment for parallel computation.
