فصلنامه روشهای هوشمند در صنعت برق
پیاپی 18 (تابستان 1393)

Recently tape wound cores، due to their excellent properties، are widely used in transformers for pulsed or high frequency applications. The spiral structure of these cores affects the flux distribution inside the core and causes complication of the magnetic analysis and consequently the circuit analysis. In this paper، a model based on reluctance networks method is used to analyze the magnetic flux in toroidal wound cores and losses calculation. A Preisach based hysteresis model is included in the model to consider the nonlinear characteristic of the core. Magnetic losses are calculated by having the flux density in different points of the core and using the hysteresis model. A transformer for using in a series resonant converter is modeled and implemented. The modeling results are compared with experimental measurements and FEM results to evaluate the validity of the model. Comparisons show the accuracy of the model besides its simplicity and fast convergence.

Application of a substrate integrated waveguide with wiggly ridge shape is presented for excitation longitudinal shunt slot antenna. Two main design equations for design substrate integrated waveguide structure and get parameters of structures، for longitudinal shunt slot excitation by shape wiggly ridge in substrate integrated waveguide are modified. Proposed method is used by applied the crinkle shape to ridge for ridge substrate integrated waveguide structure. This shape wiggly ridge just under longitudinal slot. The slot is place at centreline of substrate integrated waveguide (siw) in center of waveguide and on dielectric copper surface، top of substrate integrated waveguide. Amount of crinkle depth of waveguide centreline is proportional with needful radiation of slot and normalized conductance could be much to increase crinkle depth. In this paper the shunt element distribution assumption for prposed structure is spoted. Results of simulation show، proposed method is suitable candidate for replacing with usual longitudinal shunt slot. Structure’s useful is، low fabrication price، small profile and adaptation with microstrip circuit. Also slot place along waveguide centerline and wiggle depth substitute slot offset، therefore this procedure can suppress second order bim in array containing suggestion structure.

In this paper، a decentralized adaptive controller with using wavelet neural network is used for a class of large-scale nonlinear systems with time- delay unknown nonlinear non- affine subsystems. The entered interruptions in subsystems are considered nonlinear with time delay، this is closer the reality، compared with the case in which the delay is not considered for interruptions. In this paper، the output weights of wavelet neural network and the other parameters of wavelet are adjusted online. The stability of close loop system is guaranteed with using the Lyapanov- Krasovskii method. Moreover the stability of close loop systems، guaranteed tracking error is converging to neighborhood zero and also all of the signals in the close loop system are bounded. Finally، the proposed method، simulated and applied for the control of two inverted pendulums that connected by a spring and the computer results، show that the efficiency of suggested method in this paper.

Estimation of daily load in distribution companies which is performed to present the results to the DMS، is necessary. Daily load forecasting of power systems has traditionally been considered. Because load patterns are influenced by several factors such as climate، economy and society، it is difficult to predict the load exactly. That''s why in recent years the use of intelligent algorithms to predict it، is growing. In this project، the short-term load forecasting is performed in a hybrid approach. Due to the different behavior in different days، various methods have been used to predict the load. With studying different methods of load prediction، finally، finally exponential smoothing algorithm was used to predict the exact load in the weekdays.

Venous Thromboembolism (VTE) is a complication of surgery that may result in sudden death. Unfortunately there is not a reliable model for it up to now where with accurate modeling can be prevented serious these complications. In this paper (VTE) has been model in fuzzy method. At first we have used expert knowledge for modeling. Therefore for beginning risk factors of (VTE) have been identified and then we have helped from papers and scientific references and consultation with surgeons to extract importance of these factors in risk of VTE. Then we have obtained a questionnaire for fuzzy modeling. In this questionnaire all factors are divided into five groups. Each group is designed as a separate fuzzy system. The final output is calculated by community of all groups. Finally we validate this system by data of patients that have been surgery in Shariati hospital in Tehran and you can see that it works appropriately.

Power distribution network configuration must be restructured fast after occurring a fault in order to restoring healthy parts. A logical algorithm is needed to do such restoration. In recent years، by developing GIS systems in distribution networks، data entering and structuring in computers has been changed and dissimilar to conventional softwares، computer cannot view the network configuration. In this paper a proper definition of network agents are suggested in matrix format. By such definition fast detection of fault location and deenergized sections located at its downstream parts is done. Also curtailed load is calculated. Finally with digital data available in the GIS databases، optimal restoration strategy is selected and applied based on available routs. The proposed algorithm is applied on a complicated three feeder distribution network to test its effectiveness and speed. Simulations in MATLAB shows that the proposed algorithm is very fast and robust to any complicated network configuration as well as any number and location on sequential faults.

In this paper، a new 12bit، 200MS/s fully differential sample and hold circuit is presented. In order to increase the linearity and input voltage dynamic range; bootstrapped-switches are used for sampling the input signal. Furthermore، a tunable gain buffer is used as the output stage of the circuit to prevent the loading effects of the succeeding stages on the proposed circuit. The circuit is simulated in HSPICE using 0. 35µm CMOS technology parameters. As it is discussed in the paper، simulation results justify the good performance of the proposed circuit for using in 12bit، 200MS/s applications.

Induction motors (IMs) are widely used in industry including it be an electrical or not. However during starting period، their starting currents are so large that can damage equipment. Therefore، this current should be estimated accurately to prevent hazards caused by it. In this paper، the artificial neural network (ANN) as an intelligent tool is used to evaluate starting current peak of IMs. Both Multilayer Perceptron (MLP) and Radial Basis Function (RBF) structures have been analyzed. Six learning algorithms، backpropagation (BP)، delta-bar-delta (DBD)، extended delta-bar-delta (EDBD)، directed random search (DRS)، quick propagation (QP)، and levenberg marquardt (LM) were used to train the MLP. The simulation results using MATLAB show that most developed ANNs can estimate the starting current peak of IMs with good accuracy. However، it is proven that LM and EDBD algorithms present better performance for starting current evaluation based on average of relative and absolute errors.