JOURNAL OF FACULTY OF ENGINEERING (UNIVERSITY OF TABRIZ)
Year:2005 | Volume:30 | Issue:1 (37) ELECTRICAL ENG.|Papers Count:7

A new methodology based on a functional link net (FLN) artificial neural network (ANN) is introduced for excitation control of a synchronous generator based on fuzzy power system stabilizer (FPSS). This method combines the advantages of the fuzzy controller along with the independence from model identification and fast processing ANN and proposes a new form of excitation controller for a synchronous generator. Delta-rule is used for training of ANN. ANN is trained for different load patterns which are produced by FPSS, and is continued until the total error was smaller than certain value. The analysis of a three-phase short circuit fault condition under various loading conditions of a single machine infinite bus system is presented to illustrate the application of the developed methodology. The obtained results show that the proposed FLN artificial neural network, based on fuzzy controller for power system stabilizer (PSS), can provide very good damping characteristic through wide rang of operating conditions for power system and it has smoother control for system variables; hence it improves dynamic operating of the system considerably.

This paper presents a new approach for detection of high impedance faults in the Electric power distribution systems using wavelet transform. The proposed method has been implemented on a digital protective relay and its behavior is investigated using appropriate simulation software. Details of the design procedure, implementation and the results of performance studies of the proposed relay are given in this paper. The experiment shows that the proposed algorithm performs very well in detecting a high impedance fault with nonlinear arcing resistance. It is clearly shown that the proposed relay is able to accurately distinguish between high impedance faults and other cases such as load and/or capacitance switching.

The speed control of induction motor using a fuzzy sliding mode based variable structure technique is proposed. This technique provides a robust performance of the motor against the variation of the motor parameters as well as uncertainties and disturbances. Modified fuzzy rules are used to eliminate the oscillation around the sliding line without significant reduction of the robustness of the system. Rotor flux is estimated by using an observer in rotor reference frame and then the feedback linearization theory is used in order to decouple the rotor speed and flux. The simulation results showed satisfactory results for the proposed control technique where the load and reference speed are changed.

Idea of notching the armature trailing, in order to improve the railgun performance, is minded by researchers from erstwhile. These improvements are observed by some experimentation. In this paper, railgun is simulated by a FEM code. Then the effects of slit dimensions on current density, field and armature accelerating force are investigated. Finally, Normalized accelerating force is plotted versus slit depth and width. Then optimized slit depth and width are obtained. Using the optimum dimensions, we can increase accelerating force and system efficiency. The variation of maximum current density versus slit depth and width, the armature accelerating force versus slit dimensions will be presented in this paper.

This paper demonstrates a method for obtaining accurate voltage from discrete field sensors; named quadrature method. In using this method, a few number of field sensors is sufficient and it is shown that it can be a tool for measuring electric field and designing voltage transducers. This method would not require a special insulation or electric shielding that is required in most conventional voltage transducer technologies of today. Simulation results show the good accuracy of the method.

In this paper we propose a new approach for solving the edge detection problem using combined isotropic and directional wavelet transform. A primary knowledge about the direction of the gradient of the image, the direction of the probable edges, is extracted by the ordinary separable isotropic wavelet-based edge detector; then based on the computed direction of the probable edges, the adaptive nonseparable directional wavelet-based edge detector is applied on these pixels and the new magnitude of the wavelet transform coefficients is computed and compared to a certain threshold value; then the pixel is classified as edge points or not edge points.  This method detects the edges of the image with superior quality and performance comparing to the ordinary wavelet-based edge detection in the presence of noise; and has less computational cost compared to the directional wavelet-based edge detection methods presented before.

Radar identification algorithms according to time of arrival (TOA), of the pulse train have very good accuracy. Since this accuracy is achieved by hard consequent computation, they are not suitable for real time applications. In this paper a systolic processor has been designed for radar identification using difference time of arrival matrix, ΔTOA. High performance radar identification can be achieved by increasing the size of ΔTOA, which increase the size of systolic array. Increasing the size of systolic array causes the clock propagation delay and asynchronous operation of cells. To overcome the above problems, the ΔTOA matrix has been partitioned. Then the algorithm is applied to the ΔTOA sub matrixes. The simulation results show, the effectiveness of the algorithm when accurate and high-speed identification is needed.