فصلنامه روشهای هوشمند در صنعت برق
پیاپی 11 (پاییز 1391)

The Degenerate Unmixing Estimation Technique (DUET) is a practical algorithm for source separation in anechoic environment. DUET performs source separation by frequency domain processing and is independent of the number of mixed sources. However performance of this algorithm for long speech signals is very slow. In this research we first down sampled real mixtures and then implemented the modified DUET algorithm. Using the proposed method reduces the implementation time almost one third of the basic algorithm.

In recent years، more attention has been given to the active AMBs thanks to their low maintenance cost، suitability for clean environments، and high speed. One of the popular types of AMBs is the three-pole type with integrated and separate cores. Some deficiencies of this type of bearing are its high oscillation، low stability and low efficiency. This paper seeks to analyse the three-pole AMBs and describe their deficiencies and introduce and analyse a new five-pole AMB. The model thus proposed has lower oscillation as well as more stability and efficiency. The five-pole AMB has been simulated by a controller in the presence of disturbance. The results of simulation demonstrate lower vibration and oscillation in the five-pole AMB in comparison with the three-pole model.

Active noise control is based on the destructive interference between the primary noise and generated noise from the secondary source. An antinoise of equal amplitude and opposite phase is generated and combined with the primary noise. In this paper، performance of the neural networks is evaluated in active cancellation of sound noise. For this reason، feedforward and recurrent neural networks are designed and trained. After training، performance of the feedforwrad and recurrent networks in noise attenuation are compared. We use Elman network as a recurrent neural network. For simulations، noise signals from a SPIB database are used. In order to compare the networks appropriately، equal number of layers and neurons are considered for the networks. Moreover، training and test samples are similar. Simulation results show that feedforward and recurrent neural networks present good performance in noise cancellation. As it is seen، the ability of recurrent neural network in noise attenuation is better than feedforward network.

Physical systems always include constraints and limits. Usually، the limits and constraints، in the control systems، are appeared as temperature and pressure limits or pumps capacity. One of the existing limits in the systems with PID controller is associated with the actuator’s saturation limits. With the saturating of the actuator، the controller’s output and plant’s input will be different and the output signal of controller do not lead the system and their states could not update correctly where this issue makes the system response undesirable. In this paper، by adding a fuzzy compensator that it’s parameters are tuned using imperialist competitive algorithm، the actuator saturation is prevented and the important parameters of the system response، such as setting time and overshoot، are improved.

In order to damping of harmonic propagation phenomenon، active power filters are installed on distribution networks. Active power filter installed on buses may cause harmonic oscillation phenomenon and whack-a-mole. In this paper، a parallel active power filter that works based on voltage detection and applies with the goal of effective reducing and compensation of harmonic propagation and oscillation has been studied. In this paper two case studies consisting of harmonic current and voltage source connections to radial network has been studied. In this paper using simulation results for a ten-buses radial power distribution feeder in different states of nonlinear load connection - harmonic voltage or current source - by selecting fixed gain that is proportional to the inverse of Characteristic impedance these two phenomena effectively reduced. Due to the variable nature of the characteristic impedance of the power distribution feeder automatic adjusting of active power filter gain will be required regardless of the system conditions. Harmonic destructive currenthas been isolated with voltage detection method using of automatically adjusting of active power filter gain and compensation current is produced in order to control of this gain with the aim of total harmonic distortion decreasing and standardizing. By injection of this compensation current can prevent harmonic propagation and whack-a-mole. Because this method of extracting harmonic current is based on voltage and total harmonic distortion control is known voltage detection method.

Applying the direct thrust force control (DFC) method in permanent magnet linear synchronous motor (PMLSM) leads to some important problems. The most important disadvantages of applying this method are electromagnetic force and linkage flux big ripple and variable switching frequency. In this paper space vector modulation (SVM) technique is applied for removing the disadvantages of classic DFC method. SVM technique makes the switching frequency constant and provides continues Voltage space compared with discrete space in classic method. Simulation results confirmed the theory. They show that combining the DFC method with SVM technique removes lots of the disadvantages of classic DFC method like big ripples and variable switching and remains the benefits of this method.

Electromagnetic suspension system with a nonlinear and unstable behavior، is used in maglev trains. In this paper a linear mathematical model of system is achieved and the state feedback method is used to improve the system stability. The control coefficients are tuned by two different methods، Riccati and a new method based on Genetic algorithm. In this new proposed method، we use Genetic algorithm to achieve the optimum values of control coefficients. The results of the system simulation by Matlab indicate the effectiveness of new proposed system. When a new reference of air gap is needed or a new external force is added، the proposed system could omit the vibration and shake of the train coupe and so، passengers feel more comfortable.

Recently، much attention has been paid to Vehicular Ad hoc Network (VANET). VANETs address direct communication between vehicle-to-vehicle and vehicles to roadside units (RSUs). They are similar to the Mobile and Ad hoc Networks (MANET) in their rapid and dynamic network topology changes due to the fast motion of nodes. High mobility of nodes and network resources limitations have made the routing، one of the most important challenges in VANET researches. Therefore، guaranteeing a stable and reliable routing algorithm over VANET is one of the main steps to realize an effective vehicular communications. In this paper، a two-step AODV-based routing protocol is proposed for VANET networks. At first، node-grouping is done using their mobility information such as speed and movement direction. If the first step cannot respond efficiently، the algorithm enters the second step which uses link expiration time (LET) information in the formation of the groups. The goal of the proposed protocol is increasing the stability of routing algorithm by selecting long-lived routes and decreasing link breakages. The comparison of proposed algorithm with AODV and DSR protocols is performed via the Network Simulator NS-2. It is shown that the proposed algorithm increases the delivery ratio and also decreases the routing control overhead.